Upload
others
View
12
Download
0
Embed Size (px)
Citation preview
1
WATER ANALYSIS AND CLIMATIC HISTORY OF GILGIT AND
HUNZA VALLEYS (A DENDROCLIMATIC APPROACH)
Muhammad Usama Zafar
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
2013
Department of Environmental Science
Federal Urdu University of Arts, Science and Technology
Gulshan-e- Iqbal Campus, Karachi, Pakistan
ii
CERTIFICATE
Certified that the candidate has completed the thesis under
my supervision
Prof. Dr. Moinuddin Ahmed
(Foreign Professor)
Laboratory of Dendrochronolgy and Plant Ecology of Pakistan,
Department of Environmental Science Federal Urdu University of Arts, Science
and Technology, Gulshan-e-Iqbal Karachi
iii
Research title: Water analysis and climatic history of Gilgit and
Hunza Valleys (A Dendroclimatic Approach)
Submitted by Muhammad Usama Zafar
M. Phil/PhD Scholar
Research supervisor Prof. Dr. Moinuddin Ahmed
Foreign Professor
Laboratory of Dendrochronology
and plant ecology of Pakistan
Department of Botany FUUAST,
Karachi
Co-supervisor Dr. Muhammad Afzal Farooq
Chairman
Department of Environmental
Science FUUAST, Karachi
Graduate Research Management Council
Department of Environmental Science Federal Urdu University of Arts, Science
and Technology Gulshan-e-Iqbal Campus Karachi
iv
i-Note
In Pakistan, family of dendrochronology is very small and still it is emerging as a new science. It
has been introduced in three or four universities throughout the country but not at professional
level. Therefore, I also included water analysis of the same area from where I worked over
dendroclimatology. Although there is no correlation between water analysis and
dendroclimatology but I analyzed both. It was very difficult for me to study the two different
fields and to make comprehensive results. My basic concern was towards dendroclimatology and
for seeking jobs in Pakistan; I included water analysis as a separate part of my thesis.
v
ii- Abstract
Twenty-nine samples from different locations of Upper Indus Basin from Gilgit and Hunza
valleys were selected for the investigation of physico-chemical characteristics. Sampling was
performed during July in the year 2012. Eleven parameters were chosen for water analysis to
assess water quality and to observe the variations among different sites. Physical factors were
analyzed at site while chemical analysis was carried out in the laboratory using standard
techniques of water analysis developed by (American Public Health Association, APHA) and
spectrophotometeric techniques. Multivariate statistical techniques including principal
component analysis (PCA) and cluster analysis (CA) were employed to interpret the data and to
unravel the causes of water pollution. Results of physico-chemical properties showed that values
of all parameters were in accordance with the permissible limits proposed by World Health
Organization (WHO) but the high values of total alkalinity shows that water is of bicarbonate
type.
Knowledge of past climate variability is necessary for understanding present and future climate
tendencies. This study used three species (Picea smithiana, Juniperu sexcelsa and Pinus
gerardiana) ring-width chronologies to investigate palaeo-temperature history in Gilgit and
Hunza valleys Northern Pakistan. The resultant reconstruction is among the first palaeo-series
from Picea smithina produced for Pakistan to date. It is in good agreement with other tree-ring
based records, and with instrumental (both local and grid) data. Ten pine chronologies including
three species were developed. Ring-width measurements were detrended using the
standardization method to preserve as much climatic signals as possible. Crossdating exposed the
presence of a strong common signal among trees. Inter-site comparison showed that a common
control mechanism affected tree growth not only within sites, but also across sites. To determine
whether climate was the main factor that controlled the growth of three species from Gilgit and
Hunza, correlation and response functions were analyzed. Temperature and precipitation were
tested for their relationship with tree growth. Mean monthly temperature and total monthly
precipitation were observed as the primary growth-limiting factor. Chronologies were negatively
correlated with temperature and precipitation of spring season, and climate correlation modeling
showed that temperature and precipitation explained 39-63% variance in the tree-ring data. Tree-
vi
ring data from Picea smithiana Jutial contained the strong temperature signal, was picked for
reconstruction. The Jutial chronology was then used to reconstruct March-June temperatures
back to A.D. 1523. The calibration model explained 38.16% of the variance in temperature, and
all calibration and verification tests were passed at good levels of significance. The reconstructed
temperature was tested over decadal and century time-scale. The coolest decadal time scale
period revealed that 17th
century experienced lowest degree of temperature and ensuing the
period of “Little Ice Age” (LIA). The temperatures reached their maximum in 19th
century over
century time-scale. As Pinus gerardiana Chaprot chronology exhibited strongest temperature
signal among all chronologies therefore, separate exercise was performed where Jutial
chronology reconstruction was compared with Chaprot reconstruction. Two species
demonstrated the common pattern in spring temperatures. However, the temperature
reconstruction from Chaprot was insufficient to produce a long term proxy temperature. This
research has strengthened the Pakistan network of chronology sites, and confirmed that Picea
smthiana, Juniperus excelsa and Pinus gerardiana have great dendro-climatic value. The last
more than 450 years of temperature fluctuations were reconstructed with a high degree of
fidelity. The current reconstruction added similar trend of temperature in comparison with the
other studies throughout central Asia.
vii
iii-DEDICATION
I would like to dedicate my thesis to my beloved mother
viii
iv-Table of Contents
Note............................................................................................................................................ i
Abstract .................................................................................................................................... ii
Dedication ............................................................................................................................... iii
Table of Contents ................................................................................................................. viii
List of Tables .......................................................................................................................... xi
List of Figures ....................................................................................................................... xiii
List of Symbols, Abbreviations or Other (Optional) ........................................................ xiv
Acknowledgements .............................................................................................................. ixx
Part One: Water Analysis of Gilgit and Hunza Valleys
Chapter 1 Water analysis ....................................................................................................... 4
1.1 Introduction .......................................................................................................................... 4
1.2 Research objectives .............................................................................................................. 5
1.3 Review of literature .............................................................................................................. 8
1.4 Materials and methods ....................................................................................................... 10
1.4.1 Sampling and on-site evaluation .................................................................................... 10
1.4.2 Methods for the detection of chemical parameters ....................................................... 12
1.4.2.1 Chloride ................................................................................................................. 12
1.4.2.2 Carbonate alkalinity ............................................................................................... 12
1.4.2.3 Bicarbonate alkalinity ............................................................................................ 12
1.4.2.4 Total hardness ........................................................................................................ 13
1.4.3 Statistical analysis ......................................................................................................... 13
Chapter 2 Results .................................................................................................................. 14
2.1 Temperature ........................................................................................................................ 14
2.2 pH ....................................................................................................................................... 15
2.3 Dissolved oxygen ............................................................................................................... 16
2.4 Total dissolved solids ......................................................................................................... 17
2.5 Electrical conductivity ........................................................................................................ 18
2.6 Salinity ................................................................................................................................ 19
ix
2.7 Chloride .............................................................................................................................. 20
2.8 Total alkalinity .................................................................................................................... 21
2.9 Total hardness ..................................................................................................................... 22
2.10 Sulphate .............................................................................................................................. 23
2.11 Nitrate ................................................................................................................................. 24
2.12 Pearson correlation matrix of all parameters ...................................................................... 29
2.13 Discussion ........................................................................................................................... 34
Part two: Climatic history of Gilgit and Hunza Valleys (A Dendroclimatic Approach)
Chapter 3 General introduction .......................................................................................... 37
3.1 Introduction to dendrochronology ...................................................................................... 37
3.1.1 Brief history of dendrochronology ................................................................................. 37
3.2 Climate of Pakistan ............................................................................................................. 38
3.3 About the study sites .......................................................................................................... 39
3.3.1 Gilgit .............................................................................................................................. 39
3.3.2 Hunza ............................................................................................................................. 40
3.4 Purpose of study ................................................................................................................. 43
3.5 Review of Literature ........................................................................................................... 45
3.5.1 Dendrochronology in Pakistan ....................................................................................... 45
3.5.2 Dendrochronology in China ........................................................................................... 46
3.5.3 Dendrochronology in Nepal ........................................................................................... 47
3.5.4 Dendrochronology in India ............................................................................................ 47
Chapter 4 Chronology development ................................................................................... 49
4.1 Introduction ........................................................................................................................ 49
4.2 Materials and methods ........................................................................................................ 49
4.3 Field methods ..................................................................................................................... 49
4.4 Laboratory preparation ....................................................................................................... 53
4.1.1 Surfacing and crossdating .............................................................................................. 53
4.1.2 Measurement using Velmex ........................................................................................... 53
4.5 Software's used in the analysis ........................................................................................... 53
4.5.1 COFECHA ..................................................................................................................... 54
4.5.2 Chronology development ............................................................................................... 55
4.5.3 ARSTAN ........................................................................................................................ 55
4.5.4 Chronology Statistics ..................................................................................................... 56
x
4.6 Results ................................................................................................................................ 58
4.6.1 Crossdating of all sites ................................................................................................... 58
4.7 Chronology development ................................................................................................... 64
4.7.1 EPS and Rbar ................................................................................................................. 76
4.7.2 Autocorrelation and partial autocorrelation ................................................................... 82
4.8 Chronology comparison ..................................................................................................... 91
4.9 Multivariate analysis........................................................................................................... 94
4.10 Discussion ........................................................................................................................... 96
4.11 Conclusion .......................................................................................................................... 99
Chapter 5 Growth-climate response ................................................................................. 100
5.1 Materials and methods ...................................................................................................... 101
5.2 Climate data ...................................................................................................................... 101
5.2.1 Temperature ................................................................................................................. 102
5.2.2 Precipitation ................................................................................................................. 103
5.3 Results .............................................................................................................................. 105
5.4 Correlation among tree ring chronologies and temperature ......................................... 105
5.5 Correlation among tree ring chronolgies and precipitation .......................................... 106
5.6 Discussion ......................................................................................................................... 119
5.7 Conclusion ........................................................................................................................ 122
Chapter 6 Temperature reconstruction ............................................................................ 123
6.1 Introduction ...................................................................................................................... 123
6.2 Material and methods ....................................................................................................... 123
6.3 Results .............................................................................................................................. 125
6.4 Comparison with Pinus gerardiana reconstruction .......................................................... 133
6.5 Discussion ......................................................................................................................... 135
6.6 Conclusion ........................................................................................................................ 138
Bibliography or References ................................................................................................ 140
xi
vi- List of Tables
Table 1.1 Nearest town, elevation and map location of Gilgit and Hunza valleys. ................ 7
Table 1.2 Analysis parameters and their analytical procedures ............................................ 11
Table 2.1 Values of all sampling sites with eleven parameters from Gilgit and Hunza ...... 26
Table 2.1 Correlation matrix among all parameters ............................................................. 28
Table 2.2 Characteristics of three groups derived from Ward's clustering of the water quality
variables of the samples collected from 29 locations ............................................................. 30
Table 3.1 Ecological characteristics of forest from sampling sites ...................................... 42
Table 4.1 Summary statistics of species from eleven sites collected from COFECHA ....... 59
Table 4.2 Negative (narrow) pointer years from ten sites of Gilgit and Hunza valleys ....... 62
Table 4.3 Positive (wide) pointer years from ten sites of Gilgit and Hunza valleys ............ 63
Table 4.4 Summary of COFECHA statistics ........................................................................ 88
Table 4.5 Summary of Arstan statistics ................................................................................ 89
Table 4.6 Correlation matrix of all chronologies values from ten sites ................................ 91
Table 5.1 Summary of correlation function between tree ring chronologies and monthly
temperature and precipitation data from Gilgit station ......................................................... 114
Table 5.2 Summary of correlation function between tree ring chronologies and monthly
temperature and precipitation data from the relevant 0.5o grid climate database (Mitchell and
Jones, 2005) .......................................................................................................................... 115
Table 5.3 Summary of response function between tree ring chronologies and monthly
temperature and precipitation data from Gilgit station ......................................................... 116
Table 5.4 Summary of response function between tree ring chronologies and monthly
temperature and precipitation data from the relevant 0.5o grid climate database (Mitchell and
Jones, 2005) .......................................................................................................................... 117
Table 5.5 Summary of four tables (5.1-5.4) including only significant signs of postive and
negative corrrlation and response analysis ........................................................................... 118
Table 6.1 Regression analysis of ten chronologies with Gilgit temperature data from different
sites of Gilgit and Hunza valleys .......................................................................................... 126
Table 6.2a Early calibration ................................................................................................. 127
Table 6.2b Late calibration .................................................................................................. 127
xii
Table 6.3 Statistics for March-June actual and reconstructed (1955-2008) temperature data
........................................................................................................................................... …129
Table 6.4a Warm periods ..................................................................................................... 130
Table 6.4b Cold periods ....................................................................................................... 130
xiii
vii- List of Figures
Figure 1.1 Map representing twenty nine sampling sites from Gilgit and Hunza rivers ....... 6
Figure 2.1 Graph shows the box and whisker plot of temperature from all sites ............... 14
Figure 2.2 Graph shows the box and whisker plot of pH from all sites ............................... 15
Figure 2.3 Graph shows the box and whisker plot of dissolved oxygen from all sites ........ 16
Figure 2.4 Graph shows the box and whisker plot of total dissiolved solids from all sites .. 17
Figure 2.5 Graph shows the box and whisker plot of electrical conductivity from all sites . 18
Figure 2.6 Graph shows the box and whisker plot of salanity from all sites ........................ 19
Figure 2.7 Graph shows the box and whisker plot of chloride from all sites ....................... 20
Figure 2.8 Graph shows the box and whisker plot of chloride from all sites ....................... 21
Figure 2.9 Graph shows the box and whisker plot of total hardness from all sites .............. 22
Figure 2.10 Graph shows the box and whisker plot of chloride from all sites ....................... 23
Figure 2.11 Graph shows the box and whisker plot of nitrate from all sites .......................... 24
Figure 2.12 Dendrogram resulting from Ward's cluster analysis 29 samples collected from
Gilgit and Hunza Rivers.......................................................................................................... 30
Figure 2.13 Principal Component analysis (PCA) based on eleven parameters of water samples
collected from Gilgit and Hunza valleys ................................................................................ 32
Figure 2.14 Scree plot of 29 water samples with eleven parameters ..................................... 33
Figure 3.1 Average monthly temperature in Co and rainfall in millimeter of Gilgit station based
on the data period from 1955-2009 ........................................................................................ 40
Figure 3.1 Average monthly temperature in Co and rainfall in millimeter of Gilgit station based
on the data period from 1955-2009 ........................................................................................ 40
Figure 3.2 Map 1 showing the study sites from Gilgit and Hunza valleys. Yellow boxes are the
sites from where samples are collected. The arrow from the second figure (Map 2) highlights the
selected area from Northern Pakistan. ................................................................................... 41
Figure 4.1 Dependence of series intercorrelation with site slope.The red circle shows the slope
from 25o to 35
o and the blue circle represents the slope ranged 40
o to 55
o ............................ 61
Figure 4.1a Picea smithiana Kargah chronology plots. five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively. ................................................ 66
Figure 4.2a Picea smithiana Jutial chronology plots. five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively. ................................................ 67
xiv
Figure 4.3a Picea smithiana Haramosh chronology plots. five figures representing raw,
standard, residual, arstan chronologies and sample depth respectively. ................................. 68
Figure 4.4a Picea smithiana Bagrot chronology plots. five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively. ................................................ 69
Figure 4.5a Picea smithiana Nalter chronology plots. five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively. ................................................ 70
Figure 4.6a Picea smithiana Chera chronology plots. five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively. ................................................ 71
Figure 4.7a Picea smithiana Chaprot chronology plots. five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively. ................................................ 72
Figure 4.8a Juniperus excelsa Chaprot chronology plotes. five figures representing raw,
standard, residual , arstan and sample depth respectively …………………………………….73
Figure 4.9a Juniperus excelsa Nalter Kargah chronology plots. five figures representing raw,
standard, residual, arstan chronologies and sample depth respectively. ................................. 74
Figure 4.10a Pinus gerardiana Chaprot chronology plots. five figures representing raw,
standard, residual, arstan chronologies and sample depth respectively. ................................. 75
Figure 4.1b Running rbar and EPS graph of Picea smithiana from Kargah……………..77
Figure 4.2b Running rbar and EPS graph of Picea smithiana from Jutial. ………………77
Figure 4.3b Running rbar and EPS graph of Picea smithiana from Haramosh………….. 78
Figure 4.4b Running rbar and EPS graph of Picea smithiana from Bagrot……………...78
Figure 4.5b Running rbar and EPS graph of Picea smithiana from Nalter………………79
Figure 4.6b Running rbar and EPS graph of Picea smithiana from Chera.………………79
Figure 4.7b Running rbar and EPS graph of Picea smithiana from Chaprot…………….80
Figure 4.8b Running rbar and EPS graph of Juniperus excelsa from Chaprot…………..80
Figure 4.9b Running rbar and EPS graph of Juniperus excelsa from Nalter……………..81
Figure 4.10b Running rbar and EPS graph of Pinus gerardiana from Chaprot………….81
Figure 4.10b Running rbar and EPS graph of Pinus gerardiana from Chaprot………….81
xv
Figure 4.11 The autocorrelation coefficients (AC) and partial autocorrelation coefficients
(PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval….83-86
Figure 4.12 Two graphs show 200 years chronologies among ten sites. Arrow indicate pointer
years all sites……………………………………………………………………………..........93
Figure 4.13 Dendrogram resulting from Ward's cluster analysis of 200 years (1800-2000)
among ten sites…………………………………………………………………………….......95
Figure 4.13 Principal component analysis of ten sites using the common period of 200 years
(1800-2000)……………………………………………………………………………………95
Figure 5.1 Box-plot of mean monthly temperature of Gilgit station based on the period (1955-
2009)……………………………………………………………………………………………102
Figure 5.2 Box-plot of mean monthly precipitation of Gilgit station based on the period (1955-
2009)…………………………………………………………………………………………...103
Figure 5.3 Hierarchy of method which is followed for correlation and response analysis…105
Figure 5.4 Graph representing correlation coefficients between residual chronologies and
temperature of Gilgit meteorological data from ten sites respectively for 13 months span.107-109
Figure 5.5 Graph representing correlation coefficients between residual chronologies and
precipitation of Gilgit meteorological data from ten sites respectively for 13 months span110-112
Figure 6.1 Actual (red) and reconstructed (dashed) March-June temperature during common
period 1955-2008…………………………………………………………………………….128
Figure 6.2 Scatter plot of the observed and reconstructed temperature of the data that were used
for early calibration period (1955-1985)…………………………………………………….128
Figure 6.3 The Gilgit March-June reconstruction over the entire period of 1523-2008….132
Figure 6.4 Ten year running mean window describes the trend of warm and cold years. Upper
line of graph represents warm years and lower line signifies cold years during 25 year's
intervals………………………………………………………………………………………132
Figure 6.5 Hundred year running mean window describes the trend of warm and cold years.
Red line of the graph represents 100 mean running window during 1620-2000……………132
Figure 6.6 Comparison between the two temperature reconstructions based on 10 years moving
average. Blue line indicates Pinus gerardiana Chaprot while red line shows Picea smithiana
Jutial reconstruction………………………………………………………………………….133
xvi
Figure 6.7 Scatter plot of the Jutial and Chaprot temperature reconstruction based on 168 years
of data (1840-2008). Jutial reconstruction is on Y-xis and Chaprot reconstruction is represented
on X-xis………………………………………………………………………………………134
xvii
viii- List of Symbols, Abbreviations
WWF…………………………………………………………………World Wildlife Fund
BOD………………………………………………………... Biochemical Oxygen Demand
COD………………………………………………………... Chemical Oxygen Demand
TOC………………………………………………………... Total Organic Carbon
PCA………………………………………………………… Principal Component analysis
CA………………………………………………………………….. Cluster analysis
GPS………………………………………………………………… Global Positioning System
GPE………………………………………………………………… Gold Panning Extraction
TDS………………………………………………………………… Total Dissolved Solids
DO………………………………………………………………….. Dissolved Oxygen
WHO……………………………………………………………….. World Health Organization
EC………………………………………………………………….. Electrical conductivity
mg/L…………………………………………………………………milli gram per Litre
T. Alkalinity…………………………………………………………Total Alkalinity
T. Hardness………………………………………………………….Total Hardness
PC……………………………………………………………………Principal Component
PMD…………………………………………………………Pakistan Meteorological Department
Ft……………………………………………………………………..Feet
mm…………………………………………………………………...millimeter
oC…………………………………………………………………….Degree Centigrade
NW…………………………………………………………………. Northwest
N……………………………………………………………………. North
ES……………………………………………………………………Eastsouth
E……………………………………………………………………..East
xviii
W…………………………………………………………………….West
ITRDB………………………………………………………International Tree Ring Data Bank
SNR………………………………………………………….Signal-to-Noise ratio
EPS…………………………………………………………..Expressed population signal
DBH…………………………………………………………Diameter at breast height
ID……………………………………………………………Identity
DPL………………………………………………………….Dendrochronology Program Library
PSKAR………………………………………………………Picea smithiana Kargah
PSJUT……………………………………………………….Picea smithiana Jutial
PSHAR………………………………………………………Picea smithiana Haramosh
PSBAG………………………………………………………Picea smithiana Bagrot
PSNAL………………………………………………………Picea smithiana Nalter
PSCHR………………………………………………………Picea smithiana Chera
PSCHP………………………………………………………Picea smithiana Chaprot
JECHP ……………………………………………………….Juniperus excelsa Chaprot
JENAL………………………………………………………Juniperus excelsa Nalter
JEMOR………………………………………………………Juniperus excelsa Morkhun
PGCHP………………………………………………………Pinus gerardiana Chaprot
ACF………………………………………………………….Autocorrelation Function
PACF………………………………………………………..Partial Autocorrelation Function
m…………………………………………………………….meters
PCReg……………………………………………………….Principal Component Regression
CE……………………………………………………………Coefficient of efficiency
RE……………………………………………………………Reduction of error
RP……………………………………………………………Pearson Product moment correlation
xix
RR…………………………………………………………...Robust Correlation coefficient
RS……………………………………………………Spearman coefficient of rank correlation
RSQ………………………………………………………….Variance explained
xx
xi- Acknowledgements
The author wishes to express his profound gratitude and sincere appreciation to Professor Dr.
Moinuddin Ahmed for his assiduous guidance, critical review, and kind help; which never failed
to inspire thoughtful approaches to the subject.
My sincere thanks are due to Dr. Jonathan Palmer, Director of Gondwana Tree ring Laboratory,
New Zealand who taught me to compute dendrochronolgial data and how to interpret these. I
also thanks to members of Lamont Doherty Earth Observatory Colombia University USA for
providing me the relevant softwares especially to Prof. Brendon M. Buckley for his useful
discussions on many aspects.
I offer my enduring gratitude to my seniors and pioneers of dendrochronology in Pakistan Dr.
Muhammad Wahab and Dr. Nasrullah Khan, and my fellow students Muhammad Akbar and
Alamdar Hussain at Federal Urdu University of Arts Science and Technology, who have helped
me in the collection of my samples. I also offer my appreciation to Azhar Kazmi for his support
in arrangement of the thesis. I owe particular thanks to Dr. Syed Shahid Shaukat, who trained me
in statistical analysis and in particular provided me insights into multivariate analysis.
I thank to my co-supervisor Dr. Muhammad Afzal Farooq for enlarging my vision of science and
providing coherent answers to my endless questions. I cannot forget Professor Dr. Arif Zubair
Dean faculty of science whose penetrating pertinent questions forced me to think more deeply.
Last but not the least, my special thanks are due to my parents, whose have supported me
throughout my years of education, both morally and financially, and I owe a special debt of
gratitude to my brothers, sister and my wife for her understanding, support and most of all
patience during the completion of the thesis.
1
General introduction
This thesis comprises of two portions. Water analysis of Gilgit and Hunza valleys; which
includes investigation of physico-chemical properties of surface water. It is being considered that
drinking water of these areas is like mineral water and it should be supplied to all over Pakistan.
However, during last few decades increased pollution even in watershed areas create a great
concern. To check whether this statement is true or false, water analysis of this area was
performed. As far as water analysis is concerned, little work has carried out in these areas.
Mercury in Pan Amalgamation was found in river in high concentration in 37 samples collected
from 24 different sites of Gilgit and Hunza (Biber et al. 2011). Physico-chemical properties
along with some heavy metals (arsenic, chromium, copper, mercury and lead) of Handrap Lake
and nullah in Ghizer district were tested and no trace metals were found (WWF). A few samples
have been analyzed for only physical and chemical analysis but still not published. In present
study, water samples were not restricted just only to rivers but also few water samples were
collected from nallahs and tap waters (used for the drinking purpose) to check out the differences
among these three types. Water samples were selected from 29 locations followed by standard
sample collection techniques. Physical parameters such as pH, Electrical conductivity,
temperature, total dissolved solids and salinity were identified at site. Further analysis was
performed in laboratories for chemical properties including chloride, total hardness, total
alkalinity, sulphate and nitrate.
The second portion of this thesis describes the climatic history of Gilgit and Hunza valleys using
tree rings. These areas are temperature and moisture dependent which are the limiting factor for
tree growth. Global climate is continuously and rapidly changing and not autonomous. Recent
floods are examples of this global environmental change and glacier melting (Muhammad,
2
2010). The fluctuation of climate cannot be understood from few years of data. Available 40 to
50 years meteorological department data is scarce therefore we need a tool to provide a long term
variations of climate. Dendrochronological work was carried out in northern areas of Pakistan
since 1987. Response function analysis of temperature and precipitation in some of northern
areas including Afghanistan (Khan et al. 2008) Hunza (Esper et al. 2000), Astore and Ayubia
(Ahmed et al. 2010a) were carried out. Here, we concentrated on Picea smithiana, Juniperus
excelsa and Pinus gerardiana of eight locations from Gilgit and Hunza valleys to find out the
history of climate. The samples collected from different locations were analyzed to develop a
network of tree rings in comparison with meteorological and grid data to find out growth-climate
correlation and response. For this purpose, different softwares i.e. COFECHA, DPL
(Dendrochronology Program Library), ARSTAN, Minitab, Correlation and Response Function
analysis (Fritts, 1976), and principal component regression analysis have been used. The sites
which showed highest climatic signals were forwarded for further analysis while those sites
showing no climatic signals were rejected.
Present investigation includes following objectives.
Objective 1
To identify physico-chemical properties and extent of pollution and its concentration in water of
rivers, nullahs, springs and municipal pipe lines at Gilgit and Hunza valleys.
Objective 2
To check out concentration of Chloride, total hardness, total alkalinity, nitrate and sulphate
whether they are in permissible limits in drinking water, nullahs and river water.
3
Objective 3
To develop a network of tree ring chronologies using different species at different sites of Gilgit
and Hunza valleys.
Objective 4
To explore growth-climate (temperature and precipitation) response of various tree species
growing in Gilgit and Hunza valleys.
Objective 5
To reconstruct temperature more than past 400 years from Gilgit and Hunza valleys
1
PART ONE
WATER ANALYSIS OF GILGIT AND HUNZA VALLEYS
1
Chapter No. 1
Water analysis
This chapter describes the importance of water in terms of quality parameters. The problem of
statement is well discussed followed by site description. An over view of water analysis carried
out throughout Pakistan, is given. Sample collection techniques, field and laboratory methods are
discussed. Eleven parameters are detected in which first five are physical (temperature, pH,
electrical conductivity, total dissolved solids, and salinity) and remaining are the chemical
parameters consisting of dissolved oxygen, chloride, total hardness, total alkalinity, sulphate and
nitrate.
1.1-Introduction
Water is a natural source and basic human need but sometimes its quality is deteriorated by
anthropogenic activities. Fresh water shortage is ever-increasing in the water-starved regions due
to increasing population (Seckler et al. 1998). Over extractions of underground water not only
depletes water table but also makes good quality aquifer vulnerable to be contaminated by
unfavorable substances (Shah et al. 2002). Over population and heavy industrialization have
produced a critical situation for water resources (Chaudhary et al. 2001). Most of developed
countries adopted alternative supplies for their domestic use, while in other parts of the world
particularly in developing countries like in Pakistan; alternative supplies are not available to
handle the whole urban population (Farooq, 2008). For the evaluation and water resources and to
reduce the threats of pollution, quality plays an important role rather than quantity. There are
many parameters which represents the water quality and composition in specific localities and
time (Praus, 2005). Certain indicators of surface water quality have been familiarized to measure
the fitness of the water and assumed to be the gauge of quality of water whether water is for the
use of drinking or other industrial or agricultural purposes.
The water quality is measured by its physical, chemical and biological parameters while the
other parameters are heavy metals, pesticides, organic matter including BOD, COD and TOC.
Physico-chemical changes such as pH, alkalinity, hardness, nutrients and other heavy metals
cause sensitiveness to aquatic organisms (Khan et al. 1999). The problem in the assessment of
5
water quality is the complexity of analysis and large number of data sets which contain much
information regarding the behavior of the water. As it is difficult to treat all the parameters in
combination, many researchers interpret the water quality parameters individually by describing
the seasonal variability and their causes.
Indus River is the biggest source of water in Pakistan covering the area of 1,140,000 sq. kms and
has social and economic value. The main source of Indus is in Tibet, it begins in the convergence
of Sengge River and Gar River that drains the Ngangoing Kangri and Gangelise Shah ranges.
The Indus then winds itself from north to south through Gilgit Baltistan just south to the
Karakorum Range then bends to the south, coming out of the hills between Peshawar and
Rawalpindi, plains of Punjab and Sindh and then routes to lower Sindh where it finally falls into
Arabian Sea.
The Indus is nourished by glaciers and snows of the Himalayas, the Karakorum and the Hindu-
Kush that originate in the Indian State of Jammu and Kashmir and The Northern Areas of
Pakistan. The Indus consists of two basins i.e. Upper Basin and Lower Basin. The parts of the
HinduKush and the Karakorum ranges in the northern territory of Pakistan are drained by Gilgit
River (that is my study area) which is bordered with Afghanistan and China in the north. The
Gilgit River combines with the network of different types of rivers including Ghizer, Yasin,
Ishkuman and Hunza River which then finally joins the Indus River near Juglot. The upper parts
of the basin are generally glaciated and covered with permanent snow.
The Hunza River basin is also a part of my study area, actually the sub basin of the Gilgit River
but owing to its substantial size and significance, it is considered as a separate basin. Like Gilgit
River, Hunza River also drains the Karakorum Mountains consisting of large glaciated area
situated in the north. The Karakorum Highway that links Pakistan to China passes across this
basin. Karimabad is the capital of Hunza valley, extended over miles and miles of terraced field
and orchards. To check out quality of water from study areas following objective will be
covered.
1.2-Research objectives
1. To obtain water quality data
2. To examine physical and chemical contamination
6
3. To learn trends of dissolved solid concentrations and its load at different sites
4. To identify whether the concentration of dissolved solids are within the permissible limits
of World Health Organization (WHO, 1993).
5. To recognize the water quality and ecological status through the use of multivariate
statistical techniques
6. To classify possible factors which are responsible for the variation in water quality of
Gilgit and Hunza Rivers
7. To relate multivariate statistical techniques to study homogeneity and heterogeneity
among sampling stations and to differentiate water quality variable for temporal variation
of Gilgit and Hunza Rivers
Figure 1.1: Map representing twenty nine sampling sites from Gilgit and Hunza Rivers
7
Table 1.1: Nearest town, elevation and map location of water collection of Gilgit and Hunza
valleys.
S. No. Locations Elevation in meters Co-ordinates
1 Baseenpur 1683 35o50N, 74
o15E
2 Baseenpur (spring) 1700 35o50N, 74
o15E
3 Kargah 1674 35o50N, 74
o15E
4 Gilgit city 1574 35o54N, 74
o21E
5 Gilgit tap water 1574 35o54N, 74
o21E
6 Jutial 1748 35o53N, 74
o20E
7 Nomal 2507 36o08N, 74
o12E
8 Nalter (spring) 2968 36o07N, 74
o10E
9 Nalter (Lake) 2968 36o07N, 74
o10E
10 Danyore 1580 36o08N, 74
o51E
11 Juglot Gah Nala 1610 36o09N, 74
o51E
12 Haramosh Nala 1600 35o07N, 74
o08E
13 Aliabad Nala 1700 36o09N, 74
o52E
14 Aliabad tapwater 1700 36o09N, 74
o52E
15 Atabad 2400 36o20N, 74
o52E
16 Gulmit 2412 36o20N, 74
o52E
17 Hussaini 2433 36o20N, 74
o52E
18 Ghalapur Nala 2500 36o36N, 74
o51E
19 Khyber Nala 2678 36o34N, 74
o48E
20 Passu 2700 36o46N, 74
o90E
21 Gulkin Nala 2403 36o24N, 74
o52E
22 Batura Glacier 2540 36o30N, 74
o52E
23 Batura Lake 2540 36o30N, 74
o52E
24 Shimshal River 2850 36o20N, 75
o01E
25 Shimshal 2850 36o20N, 75
o01E
26 Morkhun 2780 36o40N, 74
o52E
27 Boiber tributary 3075 36o40N, 74
o52E
28 Boiber Nala 3075 36o40N, 74
o52E
29 Sost River 3075 36o41N, 74
o52E
8
1.3-Review of Literature
The works regarding the analysis of water quality of Gilgit-Baltistan area are scant. Water of
Gilgit and Hunza rivers was analyzed to check concentration of mercury from water samples
collected from 37 sites. The main source of mercury was Pan Amalgamation in the small scale
gold panning and Extraction (GPE). Samples were tested in terms of dissolved and suspended
mercury in water and the main purpose of research work was to create a hydrological modeling
to recognize the source, fate and transport of mercury and to build up scenarios to reduce
mercury concentration to permissible limits (Biber, 2011).
Physical and microbial analysis was carried out from Nomal valley (Ahmed et al. 2007) by
collecting water samples throughout the year except January and February. They found the
highest fecal contamination of water at source in the months of May-August.
Physico-chemical quality of Jhelum River water for irrigation and drinking purposes at District
Muzaffarabad, Azad Kashmir were pointed out by Sarwar et al. (2007). Various physico-
chemicals were checked like pH, Electrical Conductivity, total dissolved solids and suspended
solids. They suggested that Jhelum River water is suitable for drinking and irrigation purpose.
Kabul River and its tributaries were assessed for its organic and faecal coliform strength starting
from Warsak Reservoir to the confluence point of Kabul and Indus Rivers (Khan et al. 1999).
Thirty eight samples offered high concentration of fecal contamination rendering the water unfit
for irrigation and human consumption. Khan et al. (1999) claimed that organic and fecal
contaminations were caused due to the effluent discharged from Khazana Sugar Mills, Colony
Sarhad Textile Mills, Amerjee Papers and Paper Board Mills and from different tannery
industries. Faecal contamination, using most probable number technique, was reported in one of
the residential sector of Islamabad city (Azhar, 1996) and from Risalpur, Pubi and Tarnab
(Ihsan-Ullah et al. 1999).
The whole Lahore city was investigated for its bacteriological quality of drinking water and 530
water samples were collected from different localities during the months of April and May
(Anwar et al. 2010). Among 530 samples, 197 samples were positive for bacteriological
contamination. Anwar et al. (2010) concluded that bacterial contamination is a significant
problem in Lahore.
9
Besides northern areas in Pakistan, surface water of Lower Indus Basin from Kashmor to Keti-
Bander was assessed for its physical, chemical, trace metals and microbiological analysis
(Farooq. 2012). Multivariate analysis using cluster analysis and factor analysis explained that
surface water is of acceptable quality in terms of its physico-chemical properties and the level of
coliform bacteria; however the levels of some heavy metals like lead, mercury and cadmium
exceeded the WHO (1993) permissible limits.
Limnological studies of Keenjhar Lake were conducted by Lashari et al. (2009) which dealt with
the physico-chemical properties of water including temperature, pH, alkalinity, chloride,
conductivity, TDS, turbidity, DO, calcium and magnesium. The outcome of the study
demonstrated that all the parameters of Keenjhar Lake are in accordance with the permissible
limits for aquatic quality characteristics. Physico-chemical analysis for the potable water of
Khairpur city was conducted to investigate the quality of water for the city population (Pirzada et
al. 2011). The concentration of pH, electrical conductivity, total dissolved solid, hardness,
alkalinity, chloride and sulphate were measured. They showed that estimated limits of cations
and anions were safe according to the limits proposed by WHO (1993).
Drinking water for bacteriological contamination was tested from ground and surface water
samples in Rohri city (Shar et al. 2010). They agreed that water was contaminated with total
coliform, including Escherichia coli and other Heterotrophic plate count bacteria in pre- and
post-storage. Fecal contamination was also identified by Shar et al. (2009) in main reservoir;
distribution line and consumer tap in Sukkur city. The reason was in fact the occurrence of
animal excreta in nearby water sources which may increase the number of coliform and
Escherichia coli in the drinking water of Sukkur.
Zubair et al. (2009) used factor analysis for the determination of trace metals from both open and
bore wells and found elevated concentrations of Pb and Zn in well water in both pre- and post-
monsoon periods.
Little work has been carried out from Gilgit and Hunza Rivers in terms of its physico-chemical
characteristics whether surface water is suitable for human consumption. So bearing this in mind,
we focus to estimate homogeneity and heterogeneity among sampling stations and to distinguish
water quality variables for temporal variations in Gilgit and Hunza Rivers.
10
1.4-Materials and methods
1.4.1-Sampling and on site evaluation
Twenty nine locations were chosen from the valleys of Gilgit and Hunza for the estimation of
physico-chemical parameters. Samples were collected in the month of July 2012. The collection
was carried out in such a way that samples did not get contaminated with other substances. At
each site, surface water was collected and kept in polythene plastic bottles formerly washed in
10% nitric acid for 24 hours and rinsed with distilled water. These bottles finally swamped with
sample water also for two to three times. 500 ml water was collected in each bottle and six
parameters were noted at the spot with the assistance of Sension 156 HACH potable multi-
parameter, USA. The parameters were temperature, pH, electrical conductivity, total dissolved
solids, salinity and dissolved oxygen. Two-three drops of nitric acid were introduced in the
bottles so that chemical characteristics of the water could not be deteriorated.
11
Table 1.2: Analysis parameters and their analytical procedures
Serial
No.
Variables Abbreviation Analytical method Units
Physical Parameters
1 Temperature Temp Sension 156
HACH,
oC
2 pH pH Sension 156
HACH,
No
3 Electrical Conductivity EC Sension 156
HACH,
µS cm-1
4 Total Dissolved Solids TDS Sension 156
HACH,
mg L-1
5 Salinity Sal Sension 156
HACH,
%
Chemical Parameters
6 Dissolved Oxygen DO Sension 156
HACH,
mg L-1
7 Chloride Cl-1
Titration (Silver
Nitrate)
mg L-1
8 Total Hardness Ca+Mg Titration (EDTA) mg L-1
9 Total Alkalinity CO3+HCO3 Titration (H2SO4) mg L-1
10 Sulphate SO4 Spectrophotometer mg L-1
11 Nitrate NO3 Spectrophotometer mg L-1
The procedures followed were those described by APHA (2003).
12
1.4.2-Methods for the detection of chemical parameters
1.4.2.1-Chloride
Titration by silver nitrate
25 ml of sample was taken in titration flask and few drops of potassium chromate were added to
it as an indicator, and then titrated with 0.0141N solution of AgNO3 until slight reddish color
attained. The final reading of Chloride was obtained by putting the value AgNO3 consumed in
the burette.
Cl in mg/L = (ml of AgNO3 used in titration * 0.0141(N) * 35,450) / Vol. of sample
1.4.2.2-Carbonate alkalinity
Titration by sulphuric acid
25 ml of sample was taken in titration flask with a few drops of phenolphthalein indicator were
added to it. The solution became pink. The solution was titrated with 0.1N H2SO4 until the pink
color was disappeared.
Alkalinity in mg/L = (ml of sulphuric acid used in titration * 0.1N *50,000) / Vol. of sample
1.4.2.3-Bicarbonate alkalinity
Titration by sulphuric acid
A few drops of methyl orange as an indicator were added to the sample of 25 ml of water. The
sample turned yellow by the addition of indicator. The solution was titrated with 0.1N H2SO4
standard acid which changed the sample from yellow to orange.
Alkalinity in mg/L = (ml of sulphuric acid used in titration * 0.1N *50,000) / Vol. of sample
13
1.4.2.4-Total Hardness
Titration by EDTA
Measured volume of 25 ml of sample was taken in titration flask and few drops of Eriochrome
Black T were added as indicator and titrated with standard solution of 0.01M EDTA until sample
turned blue.
Total Hardness in mg/L = (ml of EDTA used in titration * 1000) / Vol. of sample
1.4.3-Statistical analysis
For statistical analysis, I used statistical software “Minitab” version 11.12. Box and whisker plots
are produced for every site which describe the minimum, maximum and mean values, quartile 1,
quartile 3, and outliers of the data. Correlation matrix among all parameters are created to check
whether the concentration of one parameter affect the concentration of other. Multivariate
techniques (cluster analysis and principal component analysis) are applied to the datasets to yield
comprehensive results.
14
Chapter No. 2
Results
This chapter details the Analysis of water collected from 29 locations of Gilgit and Hunza
valleys. Results of Physico-chemical properties of water are discussed. Finally, these physico-
chemical properties are compared by means of multivariate analysis including cluster and
Principal component analysis.
2.1-Temperature
Figure 2.1: Graph shows the box and whisker plot of temperature from all sites. Central line in
the box is the mean, upper line represents 3rd
quartile and lower line expresses 1st
quartile. Asterisks in the graph show outliers.
Box and whisker plot shows that most of the temperatures of the sites fell from 11oC to 13
oC.
The lowest temperature was 7.4oC which was observed from Nalter (spring). Both samples from
Gilgit showed highest temperatures among all samples (more than 25oC). Although there are no
permissible limits of temperature set by WHO (1993), yet temperature may be helpful in the
growth of some microorganisms in water.
30
20
10
Tem
pera
ture
15
2.2-pH
Figure 2.2: Graph shows the box and whisker plot of pH from all sites. Central line in the box is
the mean, upper line represents 3rd
quartile and lower line expresses 1st quartile. Asterisks
in the graph show outliers.
Most of the samples were found within the permissible limits described by WHO (1993) whereas
seven samples did not expose the same results. The Morkhun valley showed the highest value
(pH=8.75) and the lowest values were observed from three sites i.e. Gilgit tap water, Nalter Lake
and Shimshal River (pH=7.0). Apparently, most of the samples collected from Hunza rivers are
towards the basic side having pH more than 8 (Table 2.1). Lower value of standard deviation
(0.59) defines that pH of all sites is similar.
9
8
7
pH
16
2.3-Dissolved Oxygen
Figure 2.3: Graph shows the box and whisker plot of dissolved oxygen from all sites. Central
line in the box is the mean, upper line represents 3rd
quartile and lower line expresses 1st
quartile. Asterisks in the graph show outliers.
Only one outlier was observed showing the lowest D.O 0.49 mg/L from Nalter Lake. Most
samples expressed D.O ranged 1.4-1.9 mg/L. The mean value was obtained 6.26 mg/L whereas
the highest value was seen from Nalter spring (2.2 mg/L). There is no guideline of D.O described
by WHO (1993). The value of standard deviation of dissolved oxygen (24.9) explains that a little
bit difference in the D.O of all sites occurred.
2.0
1.5
1.0
0.5
Oxygen
Dis
solv
ed
17
2.4-Total dissolved solids
Figure 2.4: Graph shows the box and whisker plot of total dissolved solids from all sites. Central
line in the box is the mean, upper line represents 3rd
quartile and lower line expresses 1st
quartile. Asterisks in the graph show outliers.
No outlier was found in the box and whisker analysis. The highest value of total dissolved solids
touched nearly to 280 mg/L which was far from the WHO (1993) allowable limits. Maximum
value was seen from Boiber tributary and the minimum value was reported from Gulkin Nala
(TDS = 19.3 mg/L). All values of TDS were beneath the permissible limits designed by WHO
(1993). Total dissolved solids have the standard deviation 67.4 which is high showing a
considerable difference among the samples.
300
200
100
0
Tota
l dis
solv
ed s
olids
18
2.5-Electrical conductivity
Figure 2.5: Graph shows the box and whisker plot of electrical conductivity from all sites.
Central line in the box is the mean, upper line represents 3rd
quartile and lower line
expresses 1st quartile. Asterisks in the graph show outliers.
Two sites (Juglot Gah Nala and Boiber tributary) exceeded the allowable limits of drinking water
quality. These two sites represented higher values of conductivity with more than 500 mg/L. The
lowest value was observed from Gulmit site (26.3 mg/L).The other samples have the
conductivity values within the tolerable limits. A high value of standard deviation was obtained
(144.6) indicating a vast difference in the values of conductivity exists among the samples.
500
400
300
200
100
0
Conductivity
19
2.6-Salinity
Figure 2.6: Graph shows the box and whisker plot of salinity from all sites. Central line in the
box is the mean, upper line represents 3rd
quartile and lower line expresses 1st quartile.
Asterisks in the graph show outliers.
Only one outlier was seen representing higher value of 0.3 from Boiber tributery. This value was
not observed from any other sample. All other samples attained values from 0.0 to 0.1. Out of 29
water samples, 14 showed 0.0 while the rest 14 showed 0.1 value of salinity. This parameter also
has no guideline of WHO (1993) acceptable limit of drinking water quality. Lowest value of
standard deviation was obtained among samples i.e. 0.068.
0.3
0.2
0.1
0.0
Salinity
20
2.7-Chloride
Figure 2.7: Graph shows the box and whisker plot of chloride from all sites. Central line in the
box is the mean, upper line represents 3rd
quartile and lower line expresses 1st quartile.
Asterisks in the graph show outliers.
The lowest value was 6.0 mg/L obtained from water sample of Hussaini also being represented
in outlier of box and whisker plot. The highest value was less than 15 mg/L reported from
Shimshal site. Chloride was experimented much lesser than the WHO (1993) tolerable limits.
Occurrence of most samples in terms of chloride was 10-12 mg/L.
15
14
13
12
11
10
9
8
7
6
Chlo
ride
21
2.8-Total alkalinity
Figure 2.8: Graph shows the box and whisker plot of total alkalinity from all sites. Central line
in the box is the mean, upper line represents 3rd
quartile and lower line expresses 1st
quartile. Asterisks in the graph show outliers.
Range of the samples was found to be 600-900 mg/L with no outlier. Total alkalinity was seen
higher in all samples beyond the WHO (1993) drinking water acceptable limits. The range of
values from lower to higher side was 360 to 1200 mg/L. Highest value was inspected from Gilgit
tap water and lowest value from Nalter spring. Value of 207 of standard deviation was reported
from all samples which is highest among all parameters explaining a large difference among the
values of alkalinity.
1200
1100
1000
900
800
700
600
500
400
Tota
l A
lkalinity
22
2.9-Total hardness
Figure 2.9: Graph shows the box and whisker plot of total alkalinity from all sites. Central line
in the box is the mean, upper line represents 3rd
quartile and lower line expresses 1st
quartile. Asterisks in the graph show outliers.
Six samples were scrutinized to be higher with the elevated value observed from Shimshal (240
mg/L). All the six higher values were from Hunza Rivers. The lowest value was experienced
from Jutial (28 mg/L). Most of the samples existed within the range of 50 to 150 mg/L with no
outlier. The mean value was 100 mg/L as apparent from the graph.
250
200
150
100
50
0
Tota
l H
ard
ness
23
2.10-Sulphate
Figure 2.10: Graph shows the box and whisker plot of sulphate from all sites. Central line in the
box is the mean, upper line represents 3rd
quartile and lower line expresses 1st quartile.
Asterisks in the graph show outliers.
Only one outlier was seen touching the highest value of sulphate (119 mg/L) from Aliabad tap
water whereas the water sample collected from Nala of the same site did not have the same value
(only 18 mg/L). The difference might be the clarity of the two samples. Aliabad tap water sample
was more turbid than Nala that might increase the sulphate concentration. The lowest value was
reported from Gulkin Nala and most existence of sample value was 18-44 mg/L. All samples
expressed permissible WHO (1993) limits.
120
100
80
60
40
20
0
Sulp
hate
24
2.11-Nitrate
Figure 2.11: Graph shows the box and whisker plot of nitrate from all sites. Central line in the
box is the mean, upper line represents 3rd
quartile and lower line expresses 1st quartile.
Asterisks in the graph show outliers.
Mean value of nitrate was just exceeding the value of 20 mg/L. Box in the graph covered the
area of 13 to 33 mg/L representing the incidence of most samples. The highest value was above
40 mg/L from Boiber Nala and lowest value from Nomal i.e. 5 mg/L. Importantly; all samples
are within the permissible limits of WHO (1993). No outlier was observed in the analysis.
First 12 locations were from Gilgit River and the last 17 locations were selected from Hunza
River (Table 2.1). We took averages of all parameters from two rivers and then compared these
averages. Temperature, pH, DO, salinity, chloride and nitrate got the same values. The other
parameters TDS, conductivity, total alkalinity, total hardness and sulphate showed dissimilarities
in results. The higher values were obtained from Hunza River explaining Hunza River was more
disturbed than Gilgit River.
The values of twenty nine water samples with eleven parameters are presented in Table 2.1.
World Health Organization limits for safe drinking water are shown on the bottom of the table.
40
30
20
10
0
Nitra
te
25
Above discussion shows that Boiber tributary crossed the WHO (1993) approved limits of pH,
conductivity, total alkalinity and total hardness and same was the case with Boiber Nala. The
values of all parameters were satisfying the corresponding limits (except in few cases) whereas
total alkalinity was found high even in minimum value column. It means that all samples have
high values of alkalinity. Total dissolved solids (TDS), Dissolved Oxygen (D.O), salinity,
chloride, sulphate and nitrate met the acceptable limits designed by WHO (1993).
26
Table 2.1: Values of all sampling sites with eleven parameters from Gilgit and Hunza Rivers
Locations Temperature PH D.O TDS Conductivity Salinity Chloride
Total
Alkalinity
Total
Hardness Sulphate Nitrate
Baseenpur 13.4 8.54 0.98 23 40.2 0 14.00 420 40 10 11.1
Baseenpur
(Spring) 12.31 7.57 0.91 22.3 35.7 0.1 10.00 460 32 10 27.4
Kargah 13 7.6 1.8 24.3 38.2 0.1 10.00 700 32 11 10.9
Gilgit City 27.6 8.1 1.7 31 69.5 0 10.00 480 48 18 37.4
Gilgit tap water 25.2 7 1.65 35 83.6 0.1 9.93 1200 40 11 32.9
Jutial 12.1 7.48 1.89 27.6 43.6 0.1 12.00 560 28 19 12.4
Nomal 12.9 8.65 1.29 110 68.6 0.1 8.00 1020 120 22 5
Nalter (Spring) 7.4 8.2 2.2 24.8 38.5 0.1 10.00 360 132 11 36.7
Nalter lake 9.4 7 0.49 61.4 145.7 0 7.94 960 80 21 9.5
Danyore 16.5 7.5 1.73 86.7 204 0 9.93 1020 100 37 21
Juglot Gah
Nala 12.4 7.77 1.5 211 505 0.1
12.00 860 80 23 9.7
Haramosh Nala 14.2 7.4 1.74 102.2 240 0.1 11.91 900 140 43 26.3
Aliabad Nala 12.3 7.1 1.63 36.5 86.1 0 9.93 840 40 18 22.5
Aliabad Tap
water 13.2 7.42 1.48 47.7 115.4 0
12.00 920 220 119 18.5
Atabad 10.2 7.4 2.01 69.7 164 0 7.94 700 100 21 29.4
Gulmit 12.1 7.86 1.47 85.7 20.3 0 12.00 980 72 32 18.5
Hussaini 10.2 7.75 1.41 71.21 168.9 0 6.00 820 100 24 16.4
Ghalapur Nala 11.8 8.49 1.36 166.9 70.4 0 12.00 700 116 9 13.7
Khyber Nala 16.7 8.54 1.85 171.8 316 0.1 12.00 740 172 78 10.8
Passu 11.3 7.2 1.7 52.6 124.4 0 11.90 880 80 34 33.2
Gulkin Nala 10.8 8.37 2.07 19.3 36.3 0 12.00 540 36 9 17.6
Batura Glacier 11.1 8.55 2.02 52.3 96.2 0 12.00 600 48 19 33.8
Batura Lake 11.1 8.3 1.98 119.9 219 0.1 10.00 1000 168 46 31.5
Shimshal River 10.9 7 2.01 78.3 184.8 0 14.89 800 120 48 36.5
Shimshal 11 7.2 1.92 181 423 0.1 9.93 860 240 91 39.9
Morkhun 13.3 8.75 1.26 182 426 0.1 8.00 840 136 79 36
27
Boiber
Tributery 14.5 8.55 1.46 277 513 0.3
10.00 820 176 86 12.1
Boiber Nala 13.2 8.73 1.7 155 290 0.1 8.00 560 196 60 42.7
Sost River 13.5 8.26 1.7 109.5 187.5 0.1 14.00 900 140 50 12.4
WHO
permissible
limits (1993)
No guideline 6.5-8.5 No
guideline
500
mg/L 400 mg/L
No
guideline 250
mg/L 250 mg/L
150
mg/L
500
mg/L
50
mg/L
28
Table 2.2: Correlation matrix among all parameters
Temperature pH D.O TDS Conductivity Salinity Chloride
Total
Alkalinity
Total
Hardness Sulphate
pH -0.006
D.O -0.067 0.2
TDS -0.039 0.379 0.216
Conductivity -0.01 0.149 -0.134 0.877
Salinity 0.088 0.288 -0.165 0.592 0.561
Chloride 0.014 -0.023 0.131 -0.093 -0.137 -0.125
Total
Alkalinity 0.155 -0.358 -0.071 0.289 0.273 0.067 -0.133
Total
Hardness -0.166 0.189 0.039 0.638 0.603 0.354 -0.099 0.253
Sulphate 0.001 0.083 -0.18 0.565 0.631 0.349 0.032 0.301 0.843
Nitrate 0.126 -0.122 -0.152 -0.089 0.084 -0.09 -0.131 -0.156 0.205 0.142
29
2.12- The Pearson correlation matrix of all parameters
Correlations of all parameters from 29 samples of Gilgit and Hunza Rivers are shown in Table
2.2. Eleven parameters from 29 locations from two rivers were analyzed for correlation. The
significance level at degrees of freedom 27 is checked using r Table in the following order.
1) Greater than 0.349 but less than 0.449 (p<0.05)
2) Greater than 0.449 but less than 0.554 (p<0.01)
3) Greater than 0.554 (p<0.001)
Temperature did not illustrate any significant correlation with all parameters. It was negatively
correlated with pH, DO, TDS, conductivity and total hardness and positively correlated with
salinity, chloride, total alkalinity, sulphate and nitrate. pH was found significantly correlated with
total dissolved solids and total alkalinity at (p<0.05) and positively correlated with all parameters
except chloride and nitrate.
No significant correlation was found among the dissolved oxygen and all parameters. It was
positively correlated with three parameters and negatively correlated with other parameters. TDS
showed significant correlations with four parameters i.e. conductivity, salinity, total hardness and
sulphate at (p<0.001). Conductivity also expressed strong significant positive correlations with
salinity, hardness and sulphate and weak positive correlation with total alkalinity.
Positive significant correlations were observed among salinity, total hardness and sulphate at
(p<0.05) level and chloride did not show any significant correlation with any other parameter.
Total alkalinity and total hardness indicated significant positive correlation with sulphate at
(p<0.05) and (p<0.001) respectively. There was no significant correlation between sulphate and
nitrate.
The above discussion explains that total hardness and sulphate are the two parameters which are
highly correlated with most of the parameters. It indicates that if the concentration of these
variables is disturbed, it is expected to alter the concentration of most variables thereby creating
problem with the quality of water.
30
Figure 2.12: Dendrogram resulting from Ward‟s clustering of 29 samples collected from Gilgit
and Hunza Rivers
Table 2.3: Characteristics of three groups derived from Ward‟s clustering of the water quality
variables of the samples collected from 29 locations
Water
quality
variables
Cluster I
(1,2,4,8,3,18,6,21,22)
Cluster II
(5,7,16,9,10,23,12,29,14,13,20,15,17,24)
Cluster III
(11,27,25,2619,28)
Temperature 13.28 13.07 13.51
pH 8.10 7.56 8.25
D.O 1.66 1.59 1.61
T.D.S 43.50 76.17 196.3
Conductivity 52.07 143.73 412.16
Salinity 0.04 0.035 0.13
Chloride 11.33 10.45 9.98
T. Alkalinity 535.56 924.2 780
T. Hardness 56.89 108.5 166.66
Sulphate 12.89 37.5 69.5
Nitrate 22.33 22.4 25.2
31
The characteristics of the three groups derived from agglomerative cluster analysis are presented
in the sequel (Table 2.3, Fig. 2.12). Temperature was quite similar in all clusters. The pH of the
water was found slightly more alkaline for cluster I and III. Dissolved oxygen was almost similar
in all groups whereas total dissolved solids were remarkably higher in group III. Conductivity of
the samples showed greater values for groups II and III. Salinity was found to be higher in
cluster III while chloride was little bit higher in group I. Total alkalinity was found to be low in
cluster I as compared to other two clusters. Total hardness of groups III was seems to be higher
than the other two groups. Sulphate and nitrate showed greater values for group III.
32
Figure 2.13: Principal Component analysis (PCA) based on eleven parameters of water samples collected from Gilgit and Hunza
valleys
33
Principal Component analysis (PCA) was applied on normalized data sets (11 variables)
separately for 29 locations (Fig. 2.13) to find similarities or dissimilarities among variables.
PCA of the data sets produced first five PCs with eigen values > 1 explaining 80.8 percent of
total variance with respect to quality water data sets. Eigen value measures the significance of
the factor and values greater than one are considered as significant (Shrestha and Kazama,
2007). Six parameters formed a close cluster in combination with sulphate. Hardness and
TDS occurred as a group in the nearby area. The conductivity and alkalinity were found to be
located quite apart from the rest indicating they have least correlation with the other
variables.
The scree plot (2.14) was used to explain the number of PCs to be retained in order to
understand the fundamental data structure (Vega et al. 1998). The scree plot of the present
study showed that first five PCs have eigen values greater than one (Fig. 2.14).together the
first four components explains 70.4% of the total variance inherent in the data set.
Figure 2.14: Scree plot of 29 water samples with eleven parameters.
Among five PCs, PC1 explaining 33% of total variance has weak positive loadings on TDS,
conductivity, salinity, total hardness and sulphate. PC2 (14.5% of total variance) has
moderate positive loadings on DO and nitrate. PC3 with 13.4% of total variance caused
moderate positive loading over pH and alkalinity. PC4 having 10.2% of total variance has
moderate loadings on temperature and chloride and weak loadings over nitrate. PC5 has also
moderate positive loadings on temperature and chloride
34
2.13-Discussion
Northern areas of Pakistan are mountainous rural region with a population of 900,000 living
in villages typically encompass 50-200 households (Nanan et al. 2003). In this region, the
main source of water supply is from melting snows. It comes through channels (river lets)
and small streams to the mouth of village and considered as the only supply of water for the
villages as there are no wells and hand pumps which can substitute this water. The water
from melting snows runs down with the collection of various materials on its way and
converted to turbulent mountainous stream (McCarrison, 1906).
The utilization of such water for domestic purpose may cause harmful diseases. Access to
safe drinking water is the basic human right of every citizen. Safe water is the water
complying with National Drinking Water Quality Standards and meeting the quality in
accordance with WHO (1993) and UNICEF joint report; Access means the availability of
water at least 20 liters per person per day from an improved source within one kilometer.
Present study is a first comprehensive study in which we investigated water resources (river,
stream, lakes and nullahs) of Gilgit and Hunza valleys.
Mean conductivity was found satisfactory among all samples but water from Juglot Gah Nala
exhibited high amount of conductivity. This sample also has high amount of TDS, perhaps
this might be due to the presence and amount of minerals in water added from rocks and
glaciers. Besides all parameters, total alkalinity crossed the tolerable limit of drinking water.
As the rocks of these valleys contain high amount of carbonate and bicarbonate which mixes
with the water passing through it, causing the high alkalinity in water.
Correlation analysis results explained that there are specific relationship pattern among pH,
TDS, conductivity, total hardness and sulphate and these results are also confirmed by
principal component analysis. Correlation analysis results showed the direct significant
positive relationship (P<0.001) between electrical conductivity with total hardness and
sulphate. It means that total hardness (Calcium+Magnessium) and sulphate control the
conductance of surface water of the Rivers as reported in ground water of urban areas of
Karachi (Farooq, 2008).
The cluster analysis of overall data set showed three major groups while the discriminating
variable for the groups were six variables (temperature, pH, DO, salinity, chloride and
nitrate) which showed homogeneity within groups but heterogeneity between the three
35
groups. TDS and hardness also exhibited considerable differences between the groups.
Likewise, vast differences in the mean values were observed between the clusters for
conductivity and total alkalinity. Cluster analysis highlights that at present, most of the
samples collected from Hunza River have high values of TDS, conductivity, total alkalinity
and hardness which may indicate the high concentration of salts in Hunza River as compared
to Gilgit River.
In present study, we found the lowest temperature (7.4oC and 9.4
oC) from Nalter site (lake
and spring respectively). The lowest temperature of water is due to the fact that the water
which we collected from Nalter was coming from glaciers therefore found to be the lowest
among all sites. Similar temperature (8 o
C) was also observed by Islamuddin (2011) who
worked over Nalter Lake. Water temperature from Nomal valley, located at lower height
(2507 m) than Nalter valley (2968 m) was 13 o
C in current study whereas (Ahmed and Shah,
2007) described the temperature of the same valley about 25 oC. The difference in results may
be the difference in collection season.
All the physico-chemical properties from present study and Islamuddin (2011) study are
concurrent with each other and are in accordance with the limits of WHO (1993) except total
alkalinity (Carbonate and bicarbonate) were relatively higher in both studies. Our physico-
chemical characteristics also match with the findings of Jhelum River, District Muzzafarabad
Azad Kashmir study (Sarwar et al. 2007) and with that of some studies of lower Indus Basin
(Farooq, 2012; Pirzada et al. 2011; Lashari et al. 2003) but total alkalinity was lower in their
studies as compared to current studies.
There are no hard and fast rules for the permissible limits of physical properties (temperature,
pH, turbidity, dissolved oxygen, TDS, conductivity and salinity) developed by WHO (1993).
The concentrations of all chemical parameters were found within permissible limits in
samples with the exception of total alkalinity in the samples.
Total alkalinity in terms of drinking water contaminant is not a primary or secondary source.
However, alkaline water has a bitter taste and slippery feel. High alkaline water like in
current study can cause drying of skin. Alkalinity is important for fish and aquatic life
because it acts as a buffer against rapid pH changes (Benjamin, 2002; Hemond, 2000). High
alkalinity is also important in agricultural activity. Use of high alkaline water affects the plant
growth by excessive salts raising osmotic pressure in soil solution and it causes the reduction
of water availability. This high alkalinity results in lower leaf-area index (Tyagi, 2003).
36
Finally it is concluded that physico-chemical properties of surface water of study areas are
within drinking permissible limits of WHO (1993) but present water is considered as
bicarbonate type. However Nano-filteration techniques should be installed at the mouth of
water supply to reduce the total alkalinity of Gilgit and Hunza Rivers.
37
PART TWO
DENDROCLIMATIC HISTORY OF GILGIT AND HUNZA
VALLEYS
37
Chapter No. 3
General introduction
The science of dendrochronology, its brief history and importance in the study of past climate
variations are presented in this chapter. Climate of Pakistan including five provinces is also
described. Brief introduction of Gilgit and Hunza valleys and its climate are discussed.
Review of literature from Pakistan, China, Nepal and India are also presented.
3.1-Introduction to dendrochronology
The systematic study of tree rings pattern designated to a particular event with the passage of
years is known as dendrochronology (Cook and Kariukstis, 1992).
3.1.1-Brief history of dendrochronology
According to Heizer (1956), the early Greeks were considered first to note the annual tree
layers. They also knew that the widths of these layers were dependent on environmental
conditions. Duhemel and Buffon in 1737, two French naturalists examined the frost damaged
layers of 20 rings occurred in the bark of several felled trees. Other investigators confirmed
their observations. Twining (1827) and Charles Babbage (1838) in England recognized
crossdating based on relative ring widths. In 1892, a Russian worker, F. N. Shevedov was the
first person to crossdate the annual rings and determined that structure of these episodes was
because of past climate changes.
Andrew Ellicot Douglas is recognized as the father of dendrochronology. Douglass founded
laboratory of tree ring research at the University of Arizona, Tucson (United States) in 1937
(the first institution specialized only for tree ring studies). Originally, the dendro-
chronological techniques were created to date archaeological structures; but later on tree ring
analysis was used in various disciplines. These include plant ecology, geomorphology,
hydrology, glaciology, seismology, entomology and importantly climatology. An important
facility in the area of Dendrochronolgy is the establishment of the international tree-ring data
bank (ITRDB) in 1974, which contribute to the global scientific community and free access
to the tree-ring data (Grissino-Mayer and Fritts, 1997).
In Pakistan, dendrochronological work began in the late 80s when Ahmed (1988) presented a
paper on problematic issues related to tree age estimate. Later a complete dendro-
38
chronological laboratory was developed by Prof. Dr. Moinuddin Ahmed in 2005 namely
"laboratory for Dendrochronology and Plant Ecology of Pakistan in the Department of
Botany at Federal Urdu University of Arts, Science and Technology Karachi. This is also my
institution where I conducted my research.
3.2-Climate of Pakistan
Pakistan lies in the temperate zone bordering India to the east, Afghanistan to the west, China
to the north and Iran to the south west. Northern Pakistan spans an area of 72,500 square
kilometers between latitude 34o-37
o and longitude 72
o-78
o. The silk route also known as the
Karakoram Highway makes a link to China through Khunjerab Pass.
Pakistan has four seasons; cool to cold winters from December to February, dry and hot
spring from March to May, southwest monsoon rainy summer season from June to August
and the last autumn season that starts in September and ends in November. The coastal area
along Arabian Sea is usually warm while temperature reaches even in negative in some
regions of Gilgit Baltistan and some part of Northern Areas of Pakistan.
The temperature of the capital city Islamabad ranges from 2oC in January to 40
oC in June.
Average precipitation for July and August is about 255 millimeters and comprises the
majority of the annual total 1140 mm approximately.
Baluchistan occupies 44% of Pakistan‟s land area but less population density due to scarcity
of water. Again it has the hot summers usually as high as 50oC and the record-breaking
temperature was 53oC in Sibi on 26
th May 2010. Some cities have temperature below 0
oC on
average like in Ziarat and Quetta in winter.
Khyber Pakhtun Khwa (KPK) is another province of Pakistan formerly known as North West
Frontier Province (NWFP) located in North West of the country and borders with
Afghanistan. Naran (Kaghan) valleys, Swat valley, Kalam and Upper Dir are the areas
famous for its tourism. The climate of KPK varies immensely as it mainly mountainous
region. Most of the northern areas are extremely cold in winter with temperature regularly
below zero. The summer is pleasant with heavy rainfall in some areas like in Swat (1200 mm
approximately) and with low humidity. One of the hottest places of Asia is situated here i.e.
Jacobabad while on the other hand, the northern mountains have temperate weather in the
summer and intensely cold in winter.
39
Azad Jammu and Kashmir meet the lower area of Himalayas including Hari Parbat peak and
Jamgarh peak etc. Azad Kashmir is one of the most beautiful regions of the subcontinent
which receives rainfall in both summer and winter and average rainfall exceeds to 1400mm
and Muzzafarabad and Pattan are considered as the wettest areas in Pakistan.
3.3-About the study sites
Ecological characteristics of sampling sites are presented in Table 3.1.
3.3.1-Gilgit
Gilgit Baltistan is one of the five provinces of Pakistan and is important for its tourism and
water resources. Gilgit valley is situated at the elevation of 1,454 meters (4770 ft) with
latitude and longitude 35o 55‟N and 74
o 20‟E respectively Pakistan Meteorological
Department (PMD), (1961-1990). The region is also famous because three Asian mountains
i.e. The Himalayas, The Karakorum and The Hindu Kush ranges meet here. To the
Northwest, place of interest of Gilgit valley is Kargah which lies 10km from Gilgit town. The
summer season of Gilgit is brief and hot and the summer temperature may rise up to 40oC in
July. The temperature in winter falls below zero. Rainfall is scanty in Gilgit averaging from
120 to 240 millimeters.
Figure 3.1shows average monthly temperature and total rainfall of nearby Gilgit station. The
climatic data is short approximately extending over 50 years. Another problem with the
meteorological station data is that it is away in terms of elevation from tree ring sample
collection sites. Our station data suggests that highest rainfall occurs in late spring (April-
May) which is also known as pre-monsoon period. Minimum rainfall occurs in November,
high temperature in summer (June-August, monsoon period) and lowest temperature occurs
in January.
40
Fig. 3.1: Average monthly temperature in Co and rainfall in millimeter of Gilgit station based
on the data period from 1955 to 2009.
3.3.2-Hunza
The territory of Hunza spans about 7900 square kilometers and borders the Gilgit river basin
in the west, Afghanistan and China in north and the Shigar and Indus River Basin in the
south. The major valleys in Hunza are Nilt, Nagar, Shimshal, Morkhun, Chapursan and
Hanging Glacier in high Karakurum Range. Karimabad is the main town which is also a
popular place for tourism with surrounding mountains of Ultar Sar, Rakaposhi, Hunza peak,
Passu peak, Diran peak and Lady finger peak, all 6000 meters or higher. Maximum
temperature is 27oC in May and minimum sometimes reaches up to -10
OC in January.
Hunza is one of the primary destinations in Pakistan and is the centre piece of tourism in the
northern region. Hunza can be divided into two regions, Lower Hunza and Upper Hunza. The
former of which is also called as Central Hunza and the latter as Gojal. The Central Hunza
starts from Sikanderabad leading up to Karimabad whereas Upper Hunza leads from
Karimabad to Khunjerab extending all the way up to international border with China.
The Hunza valley is also mountainous in the Gilgit Baltistan and is situated to the north of
Hunza River at an elevation of 2438 meters (7998 ft) with latitude and longitude of 36o
16‟N
and 74o 44‟E respectively. The Hunza River joins with tributaries like Chapursan, Khunjerab,
Ghujerab, Shimshal and Hisper Rivers. Nagar is the large Kingdom across the Hunza River.
Passu is the important village for farmers and 15 km away from Gulmit.
0.0
5.0
10.0
15.0
20.0
25.0
30.0
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Temperature (C) Precipitation (mm)
41
Fig. 3.2: Map 1 showing the study sites from Gilgit and Hunza valleys. Yellow boxes are the
sites from where samples are collected. The arrow from the second figure (Map 2) highlights
the selected area from Northern Pakistan.
42
Table 3.1: Ecological characteristics of forest from sampling sites
S.
No.
Species Site Latitude Longitude Elevation
in meters
Slope Aspect
1 Picea
smithiana
Kargah 35o53 74
o11 2989 34
o NW
2 Picea
smithiana
Jutial 35o50 74
o20 3250 40
o N
3 Picea
smithiana
Haramosh 35o53 74
o53 3296 53
o E/S
4 Picea
smithiana
Bagrot 36o01 74
o36 3130 45
o E
5 Picea
smithiana
Nalter 36o02 74
o35 3100 25
o SW
6 Picea
smithiana
Chera 36o9 74
o11 2900 36
o N
7 Picea
smithiana
Chaprot 36o14 74
o16 3000 35
o N
8 Juniperus
excelsa
Chaprot 36o14 74
o16 3130 45
o N
9 Juniperus
excelsa
Nalter 36o90 74
o11 2900 36
o N
10 Juniperus
excelsa
Morkhum 36o37 74
o56 3475 40
o W
11 Pinus
gerardiana
Chaprot 36o14 74
o16 2850 26
o N
43
3.4-Purpose of the study
The formation of annual layers of wood by trees in response to the conditions within their
growing season provides an alternative measure of climate on an accurately dated timescale
(i.e. dendroclimatology). Its use has been crucial to the development of the now famous
Northern Hemisphere “hockey stick” temperature reconstruction for the past millennium
(Mann et al. 1998). With careful site and species selection, targeted areas can be recognized
where tree growth is highly sensitive to available soil moisture and hence can be used to
reconstruct quantified measures of climate beyond the range of instrumental and historical
records. This methodology is highly robust and has been used to develop tree-ring-based
drought reconstructions across the USA for the past millennium (Cook et al. 2004), and
similarly for monsoon Asia (Cook et al. 2010).
In Pakistan, variation in climate significantly impact people and the economy, largely through
restrictions in hydroelectric power generation and accessibility of water for human
population, industry and agriculture. Understanding how past climatic variability developed
and persisted is a timely scientific problem. The key northern region of Gilgit-Baltistan is
particularly fragile as the Indus River passes through the region and many of its tributaries
are sourced here and contribute to our national water resources. The harsh winters and a short
growing season mean local agricultural productivity is vulnerable to the fluctuation in
climate. The minor value of export trade from this region (Rs13.3 million, 2009) does not
capture the importance of subsistence agricultural production for the 2 million people living
there, or feeding the significant numbers of tourist.
The sustainable development of the region depends on knowing the full range of natural
climate variability. Monitored climate records are simply too short (often <60 years) to
capture the range of past conditions – a situation that is even more critical when climate
predictions are attempted. This is a widespread problem often faced throughout the world
and the solution adopted in several other countries, including the USA, has been to use a
substitute or a proxy-climate indicator to provide the long record. Some would argue that
speleotherms provide a similar high resolution proxy but clearly without the spatial resolution
and high sample depth. The only suitable proxy that has been proven to be sensitive to
changes in moisture supply, able to provide broad spatial coverage, has clearly-resolved
annual radial growth, can be exactly dated and provide long enough records, are tree-rings.
This understanding is not new (e.g. Fritts, 1976), but only really over the past decade has the
44
power of tree-ring analysis (dendroclimatology) and its well-developed statistical methods
been brought to bear on the reconstruction of the joint space-time properties of past climate.
Terrible floods have recently devastated Pakistan, and for many other years the variability in
climate is of grave concern and has a significant, widespread, negative impact on the
economy and livelihoods. It is hard to predict the future on the basis of past 50 years climatic
variations however it would be easy to predict or construct a reliable model of future climatic
variation if the past climatic data is 500 years. Therefore, we can be better prepared ourselves
if we know how likely climate fluctuates. This field of science (dendroclimatology) has been
widely and successfully applied overseas to reconstruct past climatic history for hundreds of
years longer than any instrumental records. This enabled them a much clearer understanding
of the climatic pattern and helps with planning anticipated events in the future. We are using
the same technique in Gilgit and Hunza valley of Northern Areas of Pakistan.
We know the selected areas have old conifer trees though being rapidly cut illegally, so this
research opportunity is disappearing and some other limited studies have already proven the
conifers to be sensitive to climate conditions. Therefore, here we propose to develop a
network of tree-ring chronologies to investigate out the climatic history for the Gilgit and
Hunza Valley areas which is the key agricultural regions in the Gilgit-Baltistan Region.
45
3.5-Review of Literature
3.5.1-Dendrochronology in Pakistan
The Himalayan conifer might be the oldest in the region and one Juniper marcopoda tree was
reported by Bilham et al. (1983) with 1200 rings in the Hunza valley Karakurum but
crossdating was not possible.
Dendrochronological work started in 1987 in Pakistan for the first time when Ahmed
explained dendrochronology and its scope in Pakistan. Using tree rings, he (1988a) described
population structure of some planted tree species in Quetta. Within the same year (1988b),
problems which encountered in estimation of age in different tree species were also identified
by him. Tree ring chronologies of Abies pindrow were presented from moist temperate
Himalayan Region of Pakistan by Ahmed et al. (1989). Ahmed with Sarangzai (1991, 1992)
used Juniperus excelsa and Chilghoza (Pinus gerardiana) to estimate the age and growth
rates using standard dendrochronological techniques.
Juniperus excelsa from six different sites of Hunza Karakurum (Chaprot, Morkhun1,
Morkhun2, Morkhun 3, Morkhun4 and Hunza) was used reconstructing modes of regional
climate over the past 500 years (Esper, 2000). He observed more than thousand years (1450)
year‟s old juniper trees. Inter-regional pointer years reflecting common years within
Karakorum and Tien Shan (China) were also observed by Esper et al. (2001). Twenty sites
were analyzed for this purpose out of which 15 from Karakurum and 5 from Tien Shan China
and most concentrated species from these sites was juniper. 1300 years climatic history was
analyzed for Western Central Asia in which 20 individual sites were used in the Northwest
Karakorum of Pakistan by Esper et al. (2002).
Ahmed and Naqvi (2005) constructed tree ring chronologies of Picea smithiana from
Himalayan Range of Pakistan. Khan et al. (2008) identified the dendroclimatic investigations
of Picea smithiana from Afghanistan. In (2009) Ahmed showed some preliminary results for
dendroclimatic investigation using Picea smithiana of Chera and Nalter and presented 600
years chronology. Abies pindrow from Astore and Ayubia was analysed for growth-climatic
response function by Ahmed et al. (2010a). Tree ring chronologies from seven sites of
Karakorum Range were constructed by Ahmed et al. (2010b). Zafar et al. (2010) described
for its chronological work comparison with other sites. The species was same between two
sites i.e. Picea smithiana. A collection of 28 tree ring sites from six different species i.e.
46
Picea smithiana, Juniperus excelsa, Abies pindrow, Pinus gerardiana, Cedrus deodara and
Pinus wallichiana were shown to having dendroclimatic potential and he also explained that
these species are suitable for long term climatic reconstruction (Ahmed et al. 2011) but I will
describe only a few. Climate/growth correlation of the Karakorum Range was described by
Ahmed et al. (2012). Dendroclimatic and dendrohydrological response of the tree species
from Gilgit valleys were also presented by Ahmed et al. (in press). Recently Cook et al.
(2013) reconstructed five hundred years river flow of Indus River by using tree rings.
3.5.2-Dendrochronology in China
Lot of work has been carried out in China in terms of dendroclimatology and
dendrohydrology. Xiang et al. (2000) worked on ten species from three Gorges reservoirs in
which five species do not show distinct ring boundaries. The other species included Cathaya
argyrophylla, Cinnamomum camphora, Gordonia acuminate, Pinus massoniana and
Schefflera delavayi were only 38 to 138 years long showed double and missing rings. These
species were used for preliminary climate modeling and river flow. The climate modeling
expressed significant correlation with current summer rainfall and summer river flow.
2326-years ring width chronology was prepared by Zhang (2003) to check out the climatic
variability on the north eastern Qinghai-Tibetan Plateau using Sabina przewalskii. The
average length of samples was found to be 574 years while only six samples were exceeding
1000 years. Using 13 months window from previous September to current September with
one year lag effect, ring width indices indicated strong positive correlation with temperature
in October of previous growth year and during May and June, it showed positive correlation
with precipitation and negative with temperature.
Yu et al. (2004) reconstructed May-July precipitation in the north Helan Mountain since AD
1726. He used standard chronologies of five tree ring sites, early wood ring width, latewood
ring width, total ring width, minimum early wood density, maximum latewood density and
their climatic response relationship. The rainfall from May to July was reconstructed using
transfer function. His precipitation showed six reconstruction periods with precipitation lower
than mean and eight periods with the precipitation higher than mean and three wet intervals.
Climate response variations between male and female dioecious Fraxinus mandshurica trees
were analyzed by Lushuang et al. (2010). The results obtained were of the evidence that
growth pattern in two genders were similar from 1950 to 1970 but different from 1931 to
47
1940. The climate growth response between male and female was also different as female
trees showed significance relation in November to precipitation while male trees showed
significance relation to temperature in November of the previous year. The final results
suggested that climatic sensitivity was different in male and female and female represented
high climatic signals in comparison with male as female can bear more stress of the
environment.
3.5.3-Dendrochronology in Nepal
Tree ring chronologies from Nepal were discussed by Bhattacharya et al. (1992). Twenty five
sampling sites were used but only ten tree ring chronologies were crossdated. Eight species
were discussed in which two species i.e. Juniper at some sites and Pinus roxburghii created
problem of dating. Various chronology statistics were discussed including subsample signal
strength (sss), signal to noise ratio (SNR) and expressed population signal (EPS).
Cook et al. (2003) developed 32 tree ring chronologies network from the Himalayas of Nepal
and found suitable for reconstruction of temperature over the past few hundred years. He
represented six indigenous tree species which were fir (Pseudotsuga), spruce (Picea
mariana), hemlock (Tsuga), juniper (Juniperus excelsa), pine (Pinus) and elm (Ulmus). The
result showed strongest increase in temperature of October-February season over the past 400
years.
Dendroclimatic (temperature and precipitation) investigation was held to detect climate
perceptions in Langtang Central Park Nepal comprising 250 years tree ring chronological
data using Abies pindrow from two sites. The result illustrated that Abies pindrow can be
used for reconstruction of monthly temperature (Chhetri, 2008).
120 cores from 60 trees of Abies spectabilis from two sites i.e. Chandan bari and Cholangpati
Langtang National Park were crossdated to obtain mean tree ring width, series
intercorrelation and mean sensitivity. Chrononlogies were only 100-300 years old and
negatively correlated with minimum monthly temperature and positively correlated with total
monthly precipitation (Chettri et al. 2010).
3.5.4-Dendrochronolgy in India
Yadav and Singh (2002) demonstrated the dendroclimatic potential of Taxus buccata from
Western Himalayan, India. They developed 345 years ring width chronology and described
48
the indirect correlation of tree growth with pre-monsoon temperature. They also found out a
significant correlation of yew (Taxus buccata) with Abies pindrow chronology.
Cedrus deodara from two sites of Western Himalaya by Pant et al. (2000) were subjected to
densitometric and Response function analysis. Data was obtained from densitometric analysis
for earlywood, latewood, minimum, maximum and mean densities and total ring width.
Response Function analysis was used that indicated significant relationships between pre-
monsoon (March, April, May) and also pre-monsoon climate reconstruction was established
using these two species.
Borgaonkar et al. (2009) presented 458-year tree ring chronology of Himalayan cedar from
three high elevation sites of Western Himalaya (India) in relation to climate and glacier
fluctuations. Dendroclimatic investigations showed significant positive relationship of tree
ring index with winter i.e. December-February and summer precipitation and indirect
relationship with summer temperature while in case of past glacial fluctuation records,
suppressed and released growth pattern in tree ring chronology was noticed, explaining the
rapid retreat of Himalayan glaciers.
A long term rainfall reconstruction of 694-years was established using Pinus gerardiana and
Cedrus deodara from Himachal Pardesh, India by Singh et al. (2009). He developed a
correlation of January-February precipitation of (AD 1310-2005) years concluding that these
months have direct relationship with growth of these species. He also developed
reconstruction of March-July precipitation and explains 46% of variance showing 20th
century was the wettest and 18th
century was the driest period.
In the same way, Cedrus deodara from 11 moisture stressed sites from monsoon shadow
zone of the Western Himalaya, India were used to develop chronology back to AD1353
and showed direct relationship with March, April, May, June (MAMJ) precipitation. The data
obtained from reconstruction explained drought in fifteen and sixteen centuries and MAMJ
precipitation over monsoon shadow zone was directly related to El Nino Southern Oscillation
(ENSO) (Yadav, 2011).
49
Chapter 4
Chronology development
4.1-Introduction
This chapter describes the development of new tree-ring chronologies of Picea smithiana,
Juniperus excelsa and Pinus gerardiana in the study area. I present a brief site description of
standard field methods and laboratory techniques that were applied to collect and prepare the
samples. Tree ring sequences are identified and crossdated by a combination of visual and
computer-aided techniques. Standardization pursues to remove non-climatic signals from raw
tree ring data followed by the selection of suitable standardization method, which plays an
important role in dendroclimatic research. Standardization techniques using negative
exponential curve or linear curve are investigated and their statistics are discussed. Finally,
chronologies are compared by means of correlation analysis and multivariate analysis
including cluster analysis and Principal component analysis.
4.2-Materials and methods
Sampling was carried out from eight sites from which eleven chronologies were produced.
Field sampling was conducted in the month of June and July as in these months, the sites are
easily accessible.
4.3-Field Methods
For the selection of sites, high elevations were targeted because rings of trees were expected
to be most sensitive to (i.e., limited by) climate at such locations. The Swedish increment
borer was used for the collection of all samples from living trees. On average, 15 trees from
each site were sampled from the following species:
1. Picea smithiana from Kargah, Jutial, Haramosh, Bagrot, Nalter, Chera and Chaprot,
2. Juniperus excelsa from Chaprot, Nalter and Morkhun
3. Pinus gerardiana from Chaprot
Those trees were selected having high DBH (diameter at the breast height) with the
assumption of direct relationship between DBH and age (sensu Fritts, 1976, Schweingruber et
al. 1990)., Two cores from each tree were collected from opposite sides of the trees but in the
50
case of Juniperus excelsa, two to three radii were taken. DBH of the trees were measured
using DBH tape. Injuries and branches were avoided following the methods of Stokes and
Smiley (1968).
360 cores from living trees were collected from eleven sites with the lowest elevation at
2850m and highest elevation from 3475m. Pinus gerardiana was existent at lower elevation
whereas Juniperus excelsa species were cored at the highest elevation among all sites (Table
3.1). Most of the sites situated at northern aspect with the minimum and maximum slope of
25o to 53
o respectively (Table 3.1).
Some snaps taken from the forest of the study area
Picea smithiana forest from Kargah
51
Picea smithiana from Bagrot
Picea smithiana from Haramosh
52
Picea smithiana from Nalter
Juniperus excelsa from Nalter
53
4.4-Laboratory preparation
In laboratory; the cores were air dried for two days for further processing. These dried cores
were mounted on wooden groove with the help of water soluble glue and were fixed with
masking tape and again left for drying for 48 hours. Each core at the time of mounting was
given an ID, date of collection, species and site name. Sanding machine with papers of
different grits was used for surfacing the rings following Orvis and Grissino-Mayer (2002).
4.4.1-Surfacing and crossdating
After mounting, the next step was to count the rings from bark to pith and to assign calendar
years under powerful microscope using skeleton plot method followed by Stokes and Smiley
(1968). First those cores were selected whose outside ring was known means that year of
collection were dated from outside to pith. Narrow and wide rings were marked on the
skeleton plot and most narrow rings in the whole stand were circled as pointer years. One dot
was marked after every ten years, two dots after every fifty years and three dots after every
century using lead pencil. Pattern of narrow and wide rings of one core was matched by the
other core of the site. This way, visual crossdating was achieved.
4.4.2-Measurement using Velmex
The ring‟s widths of crossdated cores were measured in millimeter using measure J2X. The
identity was given using the criteria SSSSTTC. The first two SS stands for species, the next
two SS stands for sites, TT stands for tree number and C represents the core number like in
case of Juniperus excelsa from Nalter (JENL101) JE describes species name: Juniper
excelsa, NL shows site Nalter: 10 represents that this is the 10th
tree of the stand and 1
quantifies that this is core number 1 of the two or three. Black mark on the monitor screen
was used to calculate the values by measuring the distance travelled between two successive
rings. The values were stored numerically by the program itself. The program started
counting from bark to pith one by one. Ring widths were measured to 0.001 mm accuracy
selected from the program menu.
4.5-Software’s used in the analysis
Following software were used in overall analysis
Velmex Measure J2X
54
COFECHA Richard Holmes (1934-2003).
DPL (Dendrochronological Program Library) is the package program containing 36
different programs written by Richard Holmes (1934-2003).
ARSTAN (Cook et al. 1986)
Minitab (version 11.12)
4.5.1-COFECHA
The raw ring width measurement taken in millimeter was subjected to COFECHA (Holmes et
al. 1994; Grissino-Mayer 2001) to check the quality of crossdating. Default commands were
followed with 32 year cubic spline 50% wavelength cutoff for filtering; 50 year segment
length with 25 year lagged and 99% confidence interval with 0.3281 critical level of
correlation value to incorporate the results. COFECHA embodies seven parts; part one
describes title page, options selected, summary and absent rings; part two tells graphical
representation in the form of Histogram; part three shows master series with samples depth
and absent rings by year; part four demonstrates Bar plots of master dating series; part five
illustrates correlation of each series by master series; part six represents potential problems
including low correlation, divergent year to year changes absent rings and outliers. The
second part of COFECHA is of much importance and signifies the following results;
Number of dated series which tells how many samples in a stand is crossdated.
Master series which tells the longest crossdated core in the whole series.
Total rings and total dated ring in the whole stand.
Series intercorrelation which shows how much pattern of rings is similar or dissimilar
to one another.
Average mean sensitivity is the measure of relative differences in widths between two
adjacent rings.
Flags which are the source of problems in crossdating.
Here, we adopted an approach where we concentrated first, second and fifth parts to explain
our results.
55
4.5.2-Chronology development
Dendrochronologists don‟t use ring width measurement to find past climatic variations as
climatic signals in tree ring widths are small so these signals must be enhanced by indexing
procedure. Mean chronology of a given site can be obtained by averaging the indices of many
trees. The random non climatic noise caused by any measurement errors cancel one another
and signal to noise ratio is enhanced. If there is greater climatic variations among the sample
ring width we require small number of cores to extract signal-to-noise ratio.
4.5.3-ARSTAN
Software ARSTAN was used to transfer cross-dated raw data to develop standardized
chronology. We developed master chronology through first deterending method include
Standard, Residual and Arstan chronologies. The residual chronology, with mean index value
of 1 removes autocorrelation and strengthens exogenous signals. Sample depth is included in
graphs to note where sample size begins to decrease substantially.
ARSTAN stands for Auto Regressive STANdardization. Trends (systematic changes) in the
trees were removed from the software ARSTAN followed by Cook (1985). Deterending with
a cubic spline of 32 years was adopted for standardization to minimize the loss of low
frequency signal in the series using Arstan code (Holmes, 1992; Cook et al. 1986) and ring
width index of each sample was obtained by dividing raw ring width value with
corresponding smoothed value. The standardization minimizes the unwanted information
known as noise and maximizes the required variation explained by Cook and Holmes (1986)
and three chronology values were obtained i.e. raw chronology, standard chronology and
residual chronology. The raw ring width chronology is just averaged non standardized raw
data. The residual version of chronology is produced by autoregressive modeling of the
detrended measurement series and chronology is averaged to standardized value (mean
index=1). Robust mean value function produces chronology with strong common signal and
without persistence. If there is no autoregressive modeling, standard chronology is produced.
The pooled autoregression is reincorporated into residual version to produce Arstan
chronology (Holmes, 1994).
56
-lag +lag +- lag
Hierarchy shows chronologies are divided into three i.e. residual, standard and arstan chronologies.
Residual chronology with no lag (previous) year effect
Standard chronology with lag year effect
Arstan chronology with some lag year effect
4.5.4-Chronology statistics
ARSTAN describes the following statistics; statistics of raw tree-ring measurement, statistics
of standard tree ring measurements, statistics of residual tree ring measurement and statistics
of arstan tree ring measurements. First year, last year, total years, mean index, standard
deviation, skewness coefficient, kurtosis coefficient, mean sensitivity and series correlation
are common results in these four chronology statistics. We also used the auto and partial
autocorrelation up to back ten years using 95% confidence interval (t-1 to t-10). The
minimum common year period among all samples is mentioned (from starting year to ending
year). Besides all other statistics, four main statistics are much important including Rbar,
SNR, EPS, and SSS (Cook and Kariukstis, 1990) which are also further discussed.
According to Briffa and Jones (1990), Rbar is the average correlation between all possible
series and was calculated for 50 years windows lagged by 25 years. It is an indication of
common variance and is independent of sample size. The EPS partly depends on sample size,
measures how well the finite chronology compares with a theoretical infinite population
(Wigley et al. 1984). Its values range from zero to one with no test for a threshold level of the
statistics however Wigley et al. (1984) recommended the value of 0.85 for a threshold.
Subsample Signal Strength is the measure of a subset of index time series which describes the
chronology of a larger set of index time series and is measured as the quotient of EPS values
of a subset and reference sample.
Chronology development
Residual chronology
Standard chronology
Arstan chronology
57
Whereas Rtotal is the average correlation between corresponding time interval of index time
series, Rwithin is the average correlation between corresponding time interval of index time
series from different cores taken from the same tree, Rbetween is the average correlation
between corresponding time interval of index time series from different cores taken from
different trees and Reffective is the effective correlation coefficient describes between and
within cross section signal.
58
4.6-Results
4.6.1-Crossdating of all sites
On average, 67.7% of all collected samples were crossdated. Picea smithiana from all seven
sites showed good crossdating as compared to Juniperus excelsa from three sites. Crossdating
was found least successful with the samples of Juniperus excelsa from Nalter, where only
30% of cores were forwarded for the chronology construction. On average, high rejection rate
(30%) was caused due to lack of tree ring pattern which mean to say that there were too many
missing or false rings that confound our ability to identify the pattern of narrow and wide
rings that allows for crossdating. However, relatively poor physical quality of the samples
could not be observed. Several cores were too short that they were rejected from the stand or
some of tree samples, rings were too narrow to be measured. Trees from few sites like Picea
smithiana from Jutial and Nalter, nearly all cores were crossdated.
Crossdating success was accomplished with high replication i.e. every site consisted of 23
radii from 15 trees. The minimum 12 radii from 20 trees were observed in Juniperus excelsa
from Nalter. Rest of the (uncrossdated) samples from this site is preserved to crossdate in the
future due to shortage of time. Another factor that gives the additional support to the better
achievement of crossdating was the low occurrence of missing rings i.e. only 0.99% of all
crossdated rings were absent in the radii.
Portions of the time series with two or more series were found to be 445 years on average
with the highest portions presented in Picea smithiana from Nalter, Chera and Haramosh
respectively. Picea smithiana from other three sites (Jutial, Bagrot and Chaprot) were of
more than 400 years and least year portion was observed in case of Picea smithiana from
Kargah. Juniperus excelsa from three sites have the portion more than 300 years. The shortest
series (212 years) was found in Pinus gerardiana Chaprot. Most of the missing rings were
narrow in the years 1971 and 1917, and were nearly absent in all species.
Series intercorrelation and mean sensitivity are the projections of year to year variability in
the chronology (Fritts, 1976). Individual series correlation occurred in Picea smithiana Jutial
and Haramosh which means every core of the site showed good correlation with the other
samples. The lowest individual series correlation was obtained in Juniperus excelsa and
Picea smithiana from Chaprot which indicates that Chaprot site has the minimum correlation.
But Pinus geradiana from the same site reversed the results i.e. good individual series
59
correlation. The results showed that different species from the same site expressed different
correlation among all the cores. Picea smithiana from the other sites stated good correlation
ranging from 0.479-0.875.
Table 4.1: Summary statistics of species from eleven sites collected from COFECHA. 1 =
percentage of core samples that were crossdated; 2 = the portion of the time series with two
or more series; 3= individual series correlation; 4 = individual mean sensitivity; 5 = highest
correlation with 50 years dated segment; 6= lowest correlation with 50 years dated segment;
7 = mean measurement of rings; 8 = percentage of missing rings
Site 1 2 3 4 5 6 7 8
PSKAR 70% 367 0.47-0.79 0.19-0.30 0.75 0.20 0.96 0.053%
PSJUT 90% 479 0.72-0.96 0.24-0.40 0.98 0.89 0.90 0.177%
PSHAR 67% 520 0.74-0.88 0.27-0.42 0.94 0.75 0.60 0.010%
PSBAG 67% 460 0.51-0.81 0.22-0.39 0.95 0.62 0.74 0.037%
PSNLT 90% 601 0.49-0.77 0.15-0.32 0.75 0.36 0.83 0.076%
PSCHR 60% 596 0.38-0.87 0.15-0.43 0.80 0.52 0.96 0.231%
PSCHP 57% 496 0.27-0.58 0.20-0.27 0.67 0.34 0.96 0.021%
JECHP 73% 339 0.11-0.85 0.21-0.35 0.71 0.35 0.93 0.213%
JENLT 30% 378 0.37-0.65 0.19-0.32 0.73 0.50 0.68 0.021%
JEMOR NIL 495 0.01-0.16 0.37-0.55 0.31 -0.9 0.53 0.023%
PGCHP 73% 212 0.61-0.83 0.22-0.44 0.67 0.34 0.97 0.228%
PSKAR= Picea smithiana Kargah, PSJUT=Picea smithiana Jutial, PSHAR=Picea smithiana
Haramosh, PSBAG= Picea smithiana Bagrot, PSNLT= Picea smithiana Nalter, PSCHR=
Picea smithiana Chera, PSCHP= Picea smithiana Chaprot, JECHP= Juniperus excelsa
Chaprot, JENLT= Juniperus excelsa Nalter, JEMOR=Juniperus excelsa Morkhun, PGCHP=
Pinus gerardiana Chaprot,
Results suggest that climate does not significantly limit tree growth at some sites. An
indication of how climate is the limiting factor for the growth of trees is obtained from mean
sensitivity, which is a measure of year to year variability in ring-width (Fritts, 1976). Values
range from zero (no change from one ring to the next) to a value of two indicating highest
sensitivity exists in tree samples that might be climate or other factors cause year to year
changes (Fritts, 1976). Here in this study, the individual mean sensitivity values reveal low
variability. The highest individual mean sensitivity values occurred in Picea smithiana
Haramosh (0.27-0.42) while the least was in Juniperus excelsa Nalter (0.19 to 0.32). It means
60
individual core of Juniperus excelsa from Nalter has the less sensitivity to climate. Picea
smithiana from first four sites exhibited good individual sensitivity.
The mean ring width was 0.82 mm (millimeter) with the fastest growing trees unsurprisingly
from the lowest altitudinal site (Pinus gerardiana from Chaprot) where the mean
measurement for all growth rings was 0.97 mm (Table 4.1).
On average, the highest correlation of 50-years dated segments with 25 years lagged was 0.79
while the lowest was 0.48 calculated from all sites (Table 4.1). The highest correlation among
segments occurred in 17th
and 20th
century whereas the lowest correlation among segments
happened in 19th
century. The results indicated that 17th
and 20th
century among all sites and
all species crossdated well while on the other hand 19th
century proved low occurrence of
crossdating among all sites and all species. The present study stated that except for Juniperus
excelsa from two sites (Nalter and Morkhun), good crossdating occurs between two cores
collected from the same tree, among the same species growing at the different sites, between
the different species growing at the same sites and even different species growing at different
sites. Juniperus excelsa from Morkhun did not show reliable results for further investigation
and was therefore rejected.
61
Fig. 4.1: Dependence of series intercorrelation on site slope. The red circle shows the slope
from 25o to 35
o and the blue circle represents the slope ranged 40
o to 55
o.
Six sites were situated at the lower steep slope whereas four sites were sampled from higher
steep slope (Fig. 4.1). The lower values of series intercorrelation occurred from 25o to 35
o
while the higher values happened from 40o to 55
o (Fig. 4.1). It is also apparent from the
Figure that series intercorrelation values at lower steep slope range 0.55-0.75 and in case of
higher steep slope, it starts from 0.65 and ends at 0.92. It means series intercorrelation is also
affected by steepness of site (Fenwick, 2003). Trees growing on steep slopes produced ring-
width pattern common to more individuals within a site which are designated by higher series
intercorrelation values.
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
20 25 30 35 40 45 50 55
Seri
es
Inte
rco
rre
lati
on
Slope Rank
62
Table 4.2: Negative (narrow) pointer years from ten sites of Gilgit and Hunza valleys. Plus
sign indicates the presence of pointer year in the site.
Year PSKAR PSJUT PSHAR PSBAG PSNAL PSCHR PSCHP JECHP JENLT PGCHP Similar
sites
2001 + + + + + + + + + 9
1985 + + + + + 5
1974 + + + + + 5
1971 + + + + + + 6
1961 + + + + + 5
1947 + + + + + + + 7
1917 + + + + + + + 7
1877 + + + 3
1865 + + + + 4
1810 + + + + + 5
1802 + + + + + 5
1785 + + + + 4
1742 + + + + + 5
1717 + + + + 4
1707 + + + + + 5
1701 + + + + 4
1626 + + + + + + 6
1603 + + + + + 5
1602 + + + + + 5
1574 + + + + 4
1573 + + + + 4
1572 + + + + 4
1492 + + + 3
63
Table 4.3: Positive (wide) pointer years from ten sites of Gilgit and Hunza valleys. Plus sign
indicates the presence of pointer year in the site.
Year PSKAR PSJUT PSHAR PSBAG PSNAL PSCHR PSCHP JECHP JENLT PGCHP Similar
sites
2005 + + + + + 5
1994 + + + + + 5
1981 + + + + + 5
1973 + + + + + 5
1960 + + + + 4
1958 + + + + + + 6
1945 + + + + + 5
1924 + + + + + 5
1906 + + + + 4
1883 + + + + + + + 7
1834 + + + + 4
1826 + + + + 4
1804 + + + + + 5
1766 + + + + + 5
1748 + + + + + 5
1747 + + + + + 5
1661 + + + + 4
1621 + + + + + 5
1596 + + + + 4
Negative pointer years (narrow rings) from all ten sites are described in Table 4.2. Most of
the pointer years are similar in case of Picea smithiana with the higher degrees in Jutial,
Haramosh, Bagrot and Chera. Juniperus excelsa and Pinus gerardiana from Chaprot showed
the least percentage of pointer years. Juniperus excelsa from Nalter explained the lowest
resemblance of pointer years among all sites. The most evident year was experimented 2001
which was seen in almost all chronologies. 1971, 1947, 1917 and 1626 years also revealed
strong occurrence too.
64
In the same way, positive pointer years (wide rings) were also found similar in case of Picea
smithana as compared to other species (Table 4.3). The highest common positive pointer
years were witnessed in Picea smithiana from Chera. Picea smithiana from Jutial, Haramosh
and Chaprot presented close resemblance. The most evident years among all chronologies
were 1883 which was seen in seven sites followed by 1958.
Overall, similar pointer years were experimented in current study, due to similar climatic
conditions that limit the tree growth. Out of three species, Picea smithiana from every site
and from different exposures showed close resemblance. So it appears that ring-width pattern
was affected by variation in temperature because narrow rings were formed during the below
average temperature.
4.7-Chronology development
Chronologies from all ten sites are presented in Figs. 4.1a – 4.10a. Each Figure is made up of
five plots: the raw chronology (which is the biweight robust mean of crossdated ring-width
measurements); standard, residual and arstan chronologies respectively (i.e. based on single
detrended data) and sample depth (which explains how number of series changes with respect
to time). The most important feature of the chronology plots is the close resemblance among
all the raw chronologies.
In most of the chronologies, noticeable period of above average growth occurred in 16th
century (i.e. up to 1600). The period of below average encountered during the last century
(1900-2000) in almost all chronologies. It indicates that many sites show increased growth in
16th
century or it may point to the fact that growth of trees were better during early growing
season where as sites show decreased growth in 20th
century where the growth was declining.
Another factor may be the climate because surprisingly all the chronologies exhibited same
above and below average growth during 16th
and 20th
century respectively so it points to
favorable conditions for growth of trees during 16th
century and unfavorable conditions
during 20th
century.
In the present study it was analyzed that how many cores attained the age more than 300 by
making the mean among all chronologies. On average, it was observed that every site
contained 10 cores attaining the age more than 300 years. So every species at every site, large
number of crossdated cores was found to be more than 300 years. It means that all species
forests were as old as 300 years and can tell the history of past 300 years climate. However,
65
Picea smithiana from some sites got the age more than 500 years. These results were also
confirmed by taking average of sample size that covers the area of 300 years (Fig. 4.1a-4.10a)
where we got thirteen cores from all chronologies more than 300 years on average.
66
Fig. 4.1a: Picea smithiana Kargah chronology plots. Five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively
67
Fig. 4.2a: Picea smithiana Jutial chronology plots. Five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively
68
Fig. 4.3a: Picea smithiana Haramosh chronology plots. Five Figures representing raw,
standard, residual, arstan chronologies and sample depth respectively
69
Fig. 4.4a: Picea smithiana Bagrot chronology plots. Five Figures representing raw, standard,
residual, arstan chronologies and sample depth respectively
70
Fig. 4.5a: Picea smithiana Nalter chronology plots. Five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively
71
Fig. 4.6a: Picea smithiana Chera chronology plots. Five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively
72
Fig. 4.7a: Picea smithiana Chaprot chronology plots. Five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively
73
Fig. 4.8a: Juniperus excelsa Chaprot chronology plots. Five figures representing raw,
standard, residual, arstan chronologies and sample depth respectively
74
Fig. 4.9a: Juniperus excelsa Nalter chronology plots. Five figures representing raw, standard,
residual, arstan chronologies and sample depth respectively
75
Fig. 4.10a: Pinus gerardiana Chaprot chronology plots. Five figures representing raw,
standard, residual, arstan chronologies and sample depth respectively
76
4.7.1-EPS and Rbar
Positive autocorrelation was observed within the chronologies. It was also clearly obtained
between different chronologies while on the other hand; negative autocorrelation was seen
among chronologies but less than positive autocorrelation.
The output file obtained from program ARSTAN (Fig.4.1b-4.10b) describes the running EPS
and Rbar as an overview of the reliability of chronology with respect to time for Picea
smithiana, Juniperus excelsa and Pinus gerardiana. All chronologies were found reliable for
reconstruction up to 300 years on average. However, Picea smithiana from four sites Jutial,
Haramosh, Bagrot and Nalter showed suitability of samples up to 400 years. The lowest
reliability among all sites was observed in Pinus gerardiana Chaprot i.e. only 160 years.
Picea smithiana from Kargah, Chaprot and Juniperus excelsa from Chaprot satisfied the
threshold limit of EPS.
EPS and Rbar looks weak in the case of Picea smithiana from Chera before late 1600s (Fig.
4.6b) clearly showing that crossdating is not correct up to 1600. Juniperus excelsa from
Nalter too looks weak with clear dating issue (Fig. 4.9b). It is suggested that collection of
more samples from these two sites will be helpful to produce better chronology.
77
Fig. 4.1b: Running Rbar and EPS graph of Picea smithiana from Kargah
Fig. 4.2b: Running Rbar and EPS graph of Picea smithiana from Jutial
78
Fig. 4.3b: Running Rbar and EPS graph of Picea smithiana from Haramosh
Fig. 4.4b: Running Rbar and EPS graph of Picea smithiana from Bagrot
79
Fig. 4.5b: Running Rbar and EPS graph of Picea smithiana from Nalter
Fig. 4.6b: Running Rbar and EPS graph of Picea smithiana from Chera
80
Fig. 4.7b: Running Rbar and EPS graph of Picea smithiana from Chaprot
Fig. 4.8b: Running Rbar and EPS graph of Juniperus excelsa from Chaprot
81
Fig. 4.9b: Running Rbar and EPS graph of Juniperus excelsa from Nalter
Fig. 4.10b: Running Rbar and EPS graph of Pinus gerardiana from Chaprot
82
4.7.2-Autocorrelation and partial autocorrelation
Autocorrelation is the serial dependence of the observations in time series and is calculated
by autocorrelation function (ACF) and partial autocorrelation function (PACF).
Autocorrelation function tells measures the correlations between observations at different
times a part (lags) while the partial autocorrelation describes the autocorrelation at different
lags by allowing the effects of autocorrelation at intermediate lags. According to Brown and
Rothery (1993), Both ACF and PACF are used to select Autoregressive models that best
describes the time series.
Autocorrelation properties of all chronologies from ten sites were investigated by calculating
the autocorrelation coefficients (ACs) and partial autocorrelation coefficients (PACs) for the
first ten lags i.e. one lag being equal to one year (Figs. 4.11). The structure of autocorrelation
is quite similar across all sites. In all chronologies there was continuous drop from lag 1 to
10. All autocorrelations were found to be positive except it was pronounced to be negative in
4 lag in case of Pinus gerardiana Chaprot. The PACFs for the chronologies showed the
predominance of AR (1) which is in agreement with the models selected by the ARSTAN
program. The first two lags were found to be high in all chronologies for PACFs.
83
Picea smithiana from Kargah
Picea smithiana from Jutial
Picea smithiana from Haramosh
Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients
(PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval. (To
be continued.,)
84
Picea smithiana from Bagrot
Picea smithiana from Nalter
Picea smithiana from Chera
Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients
(PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval. (To
be continued.,)
85
Picea smithiana from Chaprot
Juniperus excelsa from Chaprot
Juniperus excelsa from Nalter
Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients
(PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval. (To
be continued.,)
86
Pinus gerardiana from Chaprot
Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients
(PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval.
Summary of COFECHA and ARSTAN statistics is shown in the Tables (4.4, 4.5)
respectively. The mean segment length produced by COFECHA program was 296 years
(Table 4.4) with values ranging from 167 years (Pinus gerardiana Chaprot) to 375 years
(Picea smithiana Nalter). Most of the crossdated cores were established from Picea
smithiana of Jutial i.e. 36 cores from 20 trees. The oldest living tree (619 years) belonged to
Picea smithiana from Nalter. The series intercorrelation (the mean correlation with master
chronology produced by COFECHA program) was 0.693 on average.
Mean EPS value from all sites was obtained 0.94 and was much higher from the threshold
value (EPS>0.85). There was a remarkable increase in SNR value in Picea smithiana Jutial
indicating that the agreement between time series (i.e. the strength of crossdating) is better
which may or may not be related to climate. Rbar within the trees was higher for the single
detrended data (mean Rbar within trees= 0.70) whereas mean Rbar values and mean Rbar
between the trees remained nearly equal (mean Rbar= 0.50; mean Rbar between the trees=
0.48) (Table 4.5).
Much strongest common signal between trees was found in Picea smithiana of Jutial. This
chronology also has the highest value of mean sensitivity with the smallest first order
autocorrelation (Table 4.4). The next strongest common signal was observed in Juniperus
excelsa Chaprot and Picea smithiana at Haramosh. Both these two chronologies have also
strong mean sensitivity and low autocorrelation. This may be due to occurrence of all three
sites on a very steep slope. The weakest mean sensitivity and highest first order
autocorrelation was seen in Picea smithiana Chaprot chronology (Table 4.4). It is clear from
87
the Tables (4.4 and 4.5) Picea smithiana from Jutial has highest values of mean sensitivity,
EPS, SNR, mean correlation among all radii (Rbar), mean correlation within trees, mean
correlation between trees and percentage variance in first eigen value. These all values define
the climatic signal in a chronology. The higher the values, the greater will be the climatic
signals.
88
Table 4.4: Summary of COFECHA statistics
S.
No
Species &
site name
Cores∕trees 1st order
autocorrelation
Series
Intercorrelation
Mean
sensitivity
Max period
In years
Mean
length of
series
Common
years
Cross
dated
cores
1 PCSM
Kargah
30∕15 0.55 0.671 0.241 1475-2008
(534)
269 1908-2008
(101)
21
2 PCSM
Jutial
40∕20 0.50 0.913 0.358 1523-2008
(486)
360 1837-2008
(172)
36
3 PCSM
Haramosh
30∕15 0.51 0.849 0.319 1467-2009
(543)
339 1760-2009
(250)
20
4 PCSM
Bagrot
30∕15 0.63 0.735 0.304 1480-2009
(530)
273 1870-2004
(135)
20
5 PCSM
Chera
30∕15 0.39 0.720 0.278 1394-2005
(612)
360 1797-2005
(209)
18
6 PCSM
Nalter
40∕20 0.61 0.636 0.215 1387-1986
(619)
375 1800-1986
(186)
35
7 PCSM
Chaprot
30∕15 0.71 0.562 0.229 1520-2008
(489)
276 1870-2008
(139)
17
8 JUEX
Chaprot
30∕15 0.54 0.549 0.310 1670-2008
(338)
249 1887-2008
(121)
22
9 JUEX 40∕20 0.67 0.563 0.236 1676-2009 287 1828-2009 12
89
Nalter (332) (181)
10 PIGE
Chaprot
30∕15 0.69 0.735 0.289 1737-2008
(271)
167 1895-2009
(114)
22
PCSM= Picea smithiana, JUEX= Juniperus excelsa, PIGE= Pinus gerardiana, SD= standard deviation
Table 4.5: Summary of Arstan statistics
S.
No
Species &
site name
(EPS) (SNR) Rbar Within
trees rbar
Between
trees rbar
Common
period
Eigen
value#1
Percent
variance
Commulative
variance for
first five PCs
1 PCSM
Kargah
0.954 20.681 0.508 0.776 0.497 100 yrs 9.543 47.7% 83.1%
2 PCSM
Jutial
0.993 148.329 0.805 0.893 0.802 100 yrs 28.560 79.3% 91.2%
3 PCSM
Haramosh
0.973 36.134 0.707 0.845 0.698 100 yrs 10.470 69.8% 94.1%
4 PCSM
Bagrot
0.913 10.522 0.429 0.897 0.413 100 yrs 5.684 40.6% 86.2%
5 PCSM
Chera
0.945 17.13 0.488 0.670 0.476 100 yrs 9.959 55.3% 80.9%
6 PCSM
Nalter
0.951 19.37 0.564 0.717 0.554 100 yrs 8.951 59.7% 81.2%
7 PCSM 0.889 8.05 0.335 0.371 0.333 100 yrs 6.42 40.8% 89.0%
90
Chaprot
8 JUEX
Chaprot
0.929 13.033 0.275 0.331 0.270 100 yrs 10.142 40.8% 75.4%
9 JUEX
Nalter
0.914 10.578 0.281 0.715 0.266 100 yrs 8.517 31.5% 58.9%
10 PIGE
Chaprot
0.963 26.132 0.543 0.775 0.535 100 yrs 12.430 56.5% 85.9%
91
4.8-Chronologies comparison
Fig. 4.12 shows the chronologies comparison from ten sites. Only the outer two hundred years
that all chronologies shared were extracted for comparison (PGCHP; EPS >0.85 just only 160
years). From this Figure it is clear that 1802, 1810, 1865, 1917, 1944, 1971, 1985 and 2001 years
were narrow in all chronologies. The results were also confirmed by making inter site
comparison of tree-ring index chronologies, developing a correlation matrix over the same
common two hundred years as shown in Table 4.6. It is checked whether all these chronologies
were inter-correlated with one another or not. All the species were tested to find out the positive
or negative correlation.
Table 4.6: Correlation matrix of all chronologies values from ten sites
PSKAR PSJUT PSHAR PSBAG PSNAL PSCHR PSCHP JECHP JENAL
PSJUT
0.614 ***
PSHAR
0.595 ***
0.641 ***
PSBAG
0.478 ***
0.737 ***
0.56 ***
PSNAL
0.402 ***
0.288 ***
0.232 **
0.294 ***
PSCHR
0.487 ***
0.602 ***
0.606 ***
0.678 ***
0.462 ***
PSCHP
0.344 ***
0.295 ***
0.203 **
0.398 ***
0.247 ***
0.342 ***
JECHP
0.383 ***
0.293 ***
0.459 ***
0.189 *
0.009
0.155
0.067
JENAL
0.206 **
-0.01
0.385 ***
-0.033
-0.107
-0.052
-0.108
0.469 ***
PGCHP
0.383 ***
0.539 ***
0.289 ***
0.456 ***
0.012
0.399 ***
0.248 **
0.222 **
-0.146
The period of analysis is from 1800 to 2000 while the asterisks below each value indicate its
level of significance: (i.e. ***P<0.001; **P<0.01; *P<0.05). All sites are positively correlated
and significant at (p<0.05, 0.01and 0.001) with the exception of Juniperus excelsa from Nalter
which expressed a negative relationship with some Picea and Pinus chronologies. The highest
correlation (0.737) was recorded between PSBAG and PSJUT followed by PSBAG and PSCHR
(0.678). The Juniperus excelsa Nalter site showed no relationship with five sites of Picea
92
smithiana (Jutial, Bagrot, Nalter, Chera and Chaprot) however, it was found significantly
correlated with PSKAR (P<0.01) and PSHAR (P<0.001). A good correlation was also observed
between the two Junipers (P<0.001).
93
Fig. 4.12: Two graphs show 200 years chronology similarities among ten sites. Arrows indicates the pointer years among all sites.
0
200
400
600
800
1000
1200
1400
1600
1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900
PSKR PSJL PSHR PSBG PSNL PSCH PSCR JECH JENL PGCH
0
200
400
600
800
1000
1200
1400
1600
1800
2000
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
94
4.9-Multivariate analysis
Multivariate statistical analyses were employed to better understand the similarities and
ecological status of the studied system (May et al. 2006; Hayal et al. 2009; Pejman et al. 2009).
Cluster analysis (CA) and principal component analysis (PCA) were performed among the ten
chronologies and the characteristics of three groups derived from these analyses are presented in
Fig. 4.13.The highest similarity level was observed among first three sites, and the lowest was
exhibited in Juniperus excelsa Nalter (not shown), which seems to be an outlier. Perhaps this
might be due to Nalter is dry temperate site and Juniperus excelsa and Picea smithiana are the
characteristics species of this area however, Juniperus excelsa in this valley are growing on a
slope where soil moisture is much better due to the permanent snow on the top therefore, this site
is different from others. Picea smithiana from Jutial and Chaprot formed a separate cluster.
Second cluster was formed among four chronologies in which Picea smithiana from Haramosh
and Bagrot, and Juniperus excelsa from both sites showed similarities. A third cluster was
apparent among four chronologies Picea smithiana from Kargah, Nalter and Chera while the
fourth cluster was comprised of Pinus gerardiana from Chaprot.
The PCA confirmed the results of the cluster analysis as shown in Fig. 4.14, where the first three
PCs (principal components) exceed values of 1.0, and therefore show the reliability of PCA
(Shrestha and Kazama, 2007). According to the eigen value criterion, only PC‟s that exceed 1.0
are considered as significant, and the Kaiser criterion explains that only first five factor groups
could be used because successive eigen values are less than one. PC1 resulted positive with the
highest value of Picea smithiana Jutial (0.41) whereas the lowest value was seen in Juniperus
excelsa Nalter (0.061) further indicating its outlier status. Four clusters were observed (Fig.4.14):
the first grouping; Picea smithiana from Jutial, Chaprot, Bagrot and Haramosh; a second
grouping Picea smithiana from Kargah and Nalter. Although being different species, Picea
smithiana Chera and Pinus gerardiana from Chaprot formed a third, while Juniperus excelsa
evidenced similar results in this analysis as observed in cluster analysis by forming separate
clusters. Hence, the final conclusion from multivariate analysis (CA and PCA), is the
construction of four groups among ten chronologies (Figs. 4.13; 4.14) having close resemblances
i.e. 1- Picea smithiana (Jutial and Chaprot), 2- Picea smithiana (Haramosh and Bagrot), 3- Picea
95
smithiana (Nalter and Chera) and 4- Juniperus excelsa (Chaprot and Nalter) with the correlation
values of 0.295, 0.560, 0.462 and 0.469 respectively.
Fig. 4.13: Dendrogram resulting from Ward‟s cluster analysis of 200 years (1800-2000) among
ten sites.
Fig. 4.14: Principal component analysis of ten sites using the common period of 200 years
(1800-2000).
96
4.10-Discussion
All forests studied for research were disturbed due to human cutting, however most sampled
trees were presented in forested areas that were not easily accessible, and therefore suffered
minimal human disturbance. Conifers were more representative than broadleaf species and Picea
smithiana was the most common genus at eleven sites. Picea smithiana from seven sites,
Juniperus excelsa from three sites and Pinus gerardiana from one site were selected. Each site
and species provided good crossdating under the microscope. Every tree species presented
similar narrow rings not only among the species but also among the sites, pointer years or rings
of one species and site overlap with other species and site. Therefore, it is suggested that the area
and species fell under similar climatic conditions.
COFECHA statistics indicated that the highest correlation occurred in Picea smithiana from
Jutial and lowest correlation is shown by Juniperus excelsa from Chaprot. It is suggested due to
highest crossdated cores from Picea smithiana of Jutial, highest correlation occurred. However
Picea smithiana from Haramosh and Bagrot, and Pinus gerardiana from Chaprot showed good
correlation too. The percentage of missing rings was found in all chronologies with high
percentage in Pinus gerardiana from Chaprot and Picea smithiana from Chera (nearly 0.23%).
These two species are at lower elevation among all sites. Hence the tendency to exhibit locally
absent rings was more predominant at lower elevations, as would be expected due to greater
moisture stress (Fritts, 1976). The similar tendency of missing rings was also observed at lower
elevation in Nepal (Bhattacharya et al. 1992).
Longest mean length of series was found in Picea smithiana from Nalter (375 years) and
smallest mean length series was found in Juniperus excelsa from Chaprot (249 years). Picea
smithiana from Haramosh had highest common years i.e. 250 years and Picea smithiana from
Kargah had lowest common years i.e. only 101 years. It means that more cores of Picea
smithiana from Haramosh spanned at least the past 250 years. The forest of Haramosh trees
spanned the age of 250 years while forest from Picea smithiana Kargah had the minimum age of
just only 100 years. The forest of Haramosh is not easily accessible compared with Kargah, and
therefore intensive cutting of trees was minimal.
97
Picea smithiana from Jutial, Haramosh and Bagrot had attained the mean sensitivity values of
0.3 while Picea smithiana from other two sites achieved mean sensitivity values of 0.2.
According to Speer (2010), a series with mean sensitivity values around 0.1 is so complacent that
it is difficult to crossdate, and series with mean sensitivity values more than 0.4 is sensitively
tricky to crossdate. Juniperus excelsa from Morkhun attained mean sensitivity value more than
0.4, meaning that the rings were extremely narrow as to render crossdating too difficult, and
therefore these trees were not included for further analysis. Mean sensitivity around 0.2 is
generally accepted (Fritts, 1976; Speer, 2010) because they are sensitive enough to crossdate and
reconstruct past climate. In the present study, mean sensitivity values for all series were around
0.2. The chronology statistics are consistent with the result of Ahmed et al. (2010, 2011) and
Borgaonkar et al. (2009).
The values of inter-series correlation (Rbar) and signal-to-noise ratio (SNR) from most of the
sites showed that Rbar and SNR increased with increasing growth-rates. In other words, fast
growing sites have better scenarios for climatic studies. However, for Picea smithiana from
Haramosh and Pinus gerardiana from Chaprot, the reverse was the case. Here, slow growth rates
showed the best prospects. Ahmed et al. (2011) explained that Pinus gerardiana and Pinus
wallichiana from Chitral Gol National Park and Astore, respectively, expressed best prospects
with faster growth rates. In the present study we have inverse results as compared to the findings
made by Ahmed et al. (2011). The differences in results may be the differences in the ecological
tolerances of the species situated at different elevations.
We identified 22 negative pointer years that were found among all chronologies. The most
consistent of these are present in maximum number for the years 2001, 1971, 1947, 1917, 1877,
1802, 1742, 1626, 1603, 1572 and 1492. The year 1877 (observed in current analysis) was the
biggest ENSO drought of instrumental times across much of Asia (Aceituno et al. 2009) and it
was the first of the great Victorian droughts that led to millions of deaths around the globe. It
shows up very strongly in Vietnam too. Nineteen positive pointer years were witnessed common
in all chronologies having consistent positive pointer years in 1958, 1883 and 1804. Our results
are quite similar with the findings of Esper et al. (2001) who described 429 tree ring-width
values from twelve Juniperus excelsa sites and three mixed sites (Juniper, Picea and Pinus) from
northwest Karakorum of Pakistan and southern Tien Shan in Kirghizia for extreme years since
98
1427 (Esper et al. 2001). A comparison between Karakorum and Tien Shan showed similar 17
negative inter-regional pointer years (1917, 1877, 1871, 1833, 1806, 1802, 1790, 1742, 1669,
1653, 1611, 1605, 1591, 1572, 1495, 1492, and 1483 AD) and eight positive inter-regional
pointer years (1916, 1804, 1766, 1703, 1577, 1555, 1514, 1431 AD). Esper et al. (2001)
demonstrated that these regional pointer years from Northern Karakorum of Pakistan and Tien
Shan Kirgizia were due to extreme climatic conditions which limited the tree growth on large
scale independent of site ecology, from the lower, arid to the upper humid timberlines and in
different exposures. From our analyses, the authors determined that the main limiting factor for
tree growth was temperature variations.
In the current study we note the incidence of three successive negative growth rings (1572-
1574). The year 1572 was also evidenced in the study of Esper et al. (2001), indicating that
climate was harsh during these three consecutive years across a broad region. Another noticeable
point is the occurrence of similar trend after every two hundred years; 1602 and 1802, 1717 and
1917, 1785 and 1985, 1802 and 2001. It may suggest that cycle of extreme climate year
happened after every two hundred years. The existence of 2005 positive pointer year probably
hints to the occurrence of extensive rainfall that might be the reason of favorable growth of ring
experienced in most of the northern parts of Pakistan including Gilgit and Hunza valleys.
The principal component analysis and correlation analysis exposed a high degree of
resemblances among most of the chronologies. The reason is that all chronologies were
developed from the same time period and sites were located in a region that is affected by similar
climatic and weather pattern.
In general, the chronologies from Gilgit and Hunza valleys have produced some helpful results
that should be extended by more intensive sub-regional sampling. The chronologies from Picea
smithiana proved to be excellent in internal dating and hence strong common signal has resulted.
It suggests that this species should be studied further. One chronology of Juniperus excelsa and
one of Pinus gerardiana from Chaprot indicate moderate potential in common signal evaluation.
Ring-width chronologies of high elevation Juniperus excelsa (Morkhun) provided no utility for
dendroclimatological signals.
99
The strong correlation among the ten chronologies is observed because many pairs of the sites
were situated at small distances from each other. Therefore, variety of species has quite similar
results due to similar ecological conditions. In addition, likely ecological conditions and close
resemblance in site histories, common climatic influences occurred in the whole region. With the
help of systematic sampling program, ring-width of Picea smithiana, Juniperus excelsa and
Pinus gerardiana in Gilgit and Hunza valleys focused on the potential of past climatic records.
4.11-Conclusion
Ten chronologies were developed from Gilgit and Hunza valleys, Pakistan. The chronologies
spanned 271-619 years. The resemblance of pointer years among sites exhibited a common
growth pattern. These facts highlight the status of careful site selection for dendroclimatic
research. Juniperus excelsa from Morkhun exhibited extreme sensitivity, such that crossdating
was not achievable for this study. An inter-site comparison shows that common signal occurs not
only between individuals of a single site, but also in different species on larger (regional) scales.
A similar pattern of pointer years, high correlation, high values of mean sensitivity, EPS and
SNR values demonstrates the high potential for further investigations for dendroclimatic,
dendrohydrological and drought years‟ reconstruction for the past 500 years or more. However
for better understanding of ring pattern before 1700 AD, the sample size should be increased.
100
Chapter No. 5
Growth-climate response
The next step after chronology development is to check the relationship between tree growth and
climate, to investigate the conditions that most affected the rate of cambial cell divisions during
the annual growth period. Trees that grow in dry temperate sites, such as those used for this
study may have a very short growing period of a few weeks. According to Fritts (1976) and
Blasing et al. (1984), growth-climate relationships can be examined by calculating correlations
between tree-ring index chronologies and climatic data. The response function is a form of
regression equation, in which the climate is used as the independent variable and tree ring data as
dependent variable. Response function analysis is a multivariate technique that is used to
determine the tree ring response to climatic factors, so that these results can be used to deduce
the climatic conditions into the past. This multiple regression technique first employs Principle
Component analysis (PCA) on the monthly climatic data in order to develop a new set of
uncorrelated (i.e., orthogonal) variables for regression. These uncorrelated variables are then
compared against tree growth through correlation statistics to produce a set of correlation
coefficient (i.e. response function) that tells us about the relationship between tree growth and
climate (Fritts et al. 1971; 1974).
(Ahmed et al. 2011, 2012) have demonstrated that climate influences the growth of several
Pakistan conifers by examining the strength of temperature and precipitation signal in tree ring
chronologies from Picea smithiana, Juniperus excelsa, Pinus gerardiana, Cedrus deodara, Abies
pindrow, and Pinus wallichiana. Several other tree species have been shown to be useful for
dendroclimatic investigations in the Himalayan and Karakorum regions (e.g., Ahmed, 1987;
Bhattacharya and Yadav, 1999; Yadav and Park, 2000; Singh and Yadav, 2007; Singh et al.
2009; Ahmed et al. 2009; Ahmed et al. 2011, Esper et al. 2002). Many Himalayan conifers and
broad-leaved tree species have been successfully used for dendroclimatic studies, but most of
these studies were limited to the eastern Himalaya (Ahmed et al. 2011). The western Himalaya
are home to widespread forests with a number of different species divided between the moist and
dry temperate areas of the country (Champion et al. 1965; Ahmed et al. 2006).
101
Here we investigated the growth climate correlation and response in order to identify the
climate variables that have significant effect on some species of the genus Pinus. Tree ring data
from ten chronologies were correlated against mean monthly temperature and total monthly
precipitation. The results are presented below.
5.1-Materials and methods
To determine which chronology best suited our study; we developed a preliminary correlation
among climatic data from the Gilgit station and gridded data (Mitchell and Jones, 2005) with
three chronologies using point-by-point regression. The percent variance for correlation was
obtained at 95 percent confidence interval using software known as CORRELATION AND
RESPONSE FUNCTION with packaged software DPL from the LDEO (Lamont Doherty Earth
Observatory) TRL website, as described by Fritts (1976). The Pearson correlation coefficients
were calculated among tree ring chronologies and the monthly series of temperature and
precipitation from both station and grid data for the 13 months period ranging from previous
October to current October (Figs. 5.1; 5.5). The residual version of each chronology was used to
estimate the growth-climate correlation coefficients because the residuals version is pre-whitened
in order to remove the low order persistence due to autocorrelation (Cook, 1985). PCA was
performed on the climate variables by employing a variance maximizing rotation of original
variable space (Richman, 1986). The first principal component explains the common variance of
the chronology set and hence point out to the regional growth signal.
5.2-Climate data
Like other parts of the world, the scarcity of long meteorological records for statistical
calibration to the tree rings in the Himalayas is a problem (Cook et al. 2003; Bhattacharya et al.
1992). One solution of this difficulty has been to introduce seasonal data into a 0.5o latitude-
longitude grid data set that can be useful for its proximity to the tree ring location (Cook et al.
2003). Here we used gridded climate data from the CRU TS 2.1(http:/www.cru.uea.ac.uk/) for
the region over northern Pakistan for comparison with the Gilgit meteorological data
(temperature and precipitation). The Gilgit local station sits at an elevation of 1460 m, whereas
Hunza has no local climatic station. Therefore, Gilgit‟s climatic data is considered to be the most
representative of the local climate for all of the ten sites. The records for the Gilgit local station
102
span from 1955-2009, and are significantly shorter than the gridded product that extends from
1901-2002 (Mitchell and Jones, 2005). For many of the tree ring samples sites, no local records
exist, and the ones that do are from much lower elevation. Therefore, each tree ring chronology
was estimated with Gilgit local station data and also with CRU data.
5.2.1-Temperature
The Gilgit meteorological station is located in Gilgit (35o55N, 74.20E) in close vicinity to most
of the tree ring sites. Although short (just greater than 50 years), its instrumental record is the
longest in Northern Pakistan. Box and whisker plot of Gilgit monthly temperature (oC) is
presented in Fig. 5.1. Maximum temperature peaks in July while the lowest temperature is
recorded in December and January. The mean annual temperature in Gilgit is 15.93oC obtained
by averaging fifty years data. The warmest year on record was 1961 when the average annual
temperature reached 17.53oC, while 1989 was the coldest year on record with annual mean
temperature of 14.53oC.
Figure 5.1: Box plot of mean monthly temperature of Gilgit station based on the period (1955-
2009). Asterisks in the figure show the outliers from the data.
103
5.2.2-Precipitation
Like temperature, the precipitation data used in the current study were recorded at Gilgit. The
observation period was the same 1955-2009. Box and whisker plots of Gilgit monthly
precipitation in millimeters are presented in Fig. 5.2. As expected, there is far more variability
with precipitation than with temperature, as seen by the outliers above the mean. Maximum
precipitation is generally seen for the months of April and May, with November being on
average the driest. The mean total precipitation from the station is approximately 131.4 mm
annually, nearly similar in both halves of the record. 1996 was the wettest year on the record
where total precipitation reached nearly 251.7 mm per annum. The driest year was 1977 where
the annual precipitation accounted for just 40.7 mm. The precipitation data is scattered evenly
throughout the year with no consistent change. The minimum sum of 141.9 mm (November) and
maximum sum of 1384 mm (May) throughout the entire period of (1955-2009).
Figure 5.2: Box plot of mean monthly precipitation of Gilgit station based on the period (1955-
2009). Asterisks in the figure show the outliers from the data.
104
From the above discussion, it is expected that annual radial tree growth is sensitive to
temperature and precipitation, because in this high and arid region tree ring growth can be
restricted by extreme variations of both variables. If precipitation is sufficient, then temperature
is the expected principal factor to restrict tree growth, and vice versa. Gilgit experiences high
temperature in summer and relatively little rainfall throughout the year, hence, temperature and
precipitation both are likely to limit growth. The climate of the region is dry temperate (Ahmed
et al. 2006) and is characterized by maximum temperature during summer months (June-
August). The rainfall, maximum occurs during late spring that is in the months of March and
April. The mean temperature at Gilgit station is between 20-25oC in summer months and mean
precipitation during late spring lies between 20 to 30 mm (see section 3 Fig. 3.1).
The hierarchy (Fig. 5.3) shows the method that is adopted for further analysis. Out of three
chronology versions, the pre-whitened residual chronology (RES) was used for all correlation
and response function analyses because the AR modeling of the detrended series (Cook, 1985)
increases the confidence of the climate growth relationship without the possible spurious
correlation of trend in data. The RES series were compared with climate data to identify
significant months as shown below.
105
Figure 5.3: Hierarchy of method which is followed for correlation and response analysis
5.3-Results
5.4-Correlation among tree-ring chronologies and temperature
The Growth-temperature relationships for all ten chronologies from ten sites against the Gilgit
station data are shown in figure 5.2. Low positive correlation coefficients (54) were found in
comparison with negative coefficients (76), which imply that high temperature has a negative
impact on tree-growth during the growing seasons (March-May). All tree ring indices showed
similar positive trend in temperature correlation with respect to the winter season (previous
December to January) including previous November also. On the other hand, three species
exhibited different response in terms of spring and summer seasons. Picea smithiana from all
CHRONOLOGY
STANDARD ARSTAN RESIDUAL
CORRELATION FUNCTION
LOCAL CLIMATE
GRIDDED CLIMATE
RESPONSE FUNCTION
LOCAL CLIMATE
GRIDDED CLIMATE
106
sites showed negative correlation in spring season (March-May). Juniperus excelsa from two
sites exhibited negative relationship with temperature from May to July, and this might be
considered as summer season whereas Pinus gerardiana presented a long term negative
correlation including five months (March-July).
5.5-Correlation among tree-ring chronologies and precipitation
Like temperature, the relationship between tree growth and precipitation was seen from Gilgit
precipitation data (Fig. 5.3). In contrast with temperature, correlation comparison explains more
positive correlation (92) and less negative correlation coefficients (38) pointing to the fact that
more rainfall is good for tree growth during its growing season.
Synchronization occurred across all chronologies when comparison was made in correlation
analysis. The results showed that tree ring indices were positively correlated with precipitation
for the spring season, in particular the four months of February to May, with the exception of the
two species from Nalter, having slightly different results for these months (Picea smithiana
showed negative correlation in the month of May and Juniperus excelsa showed low positive
correlation in the months of February-May. Most probably it happened due to the occurrence of
these two species as Nalter is wet site, receives extensive amount of permanent snow on the top.
The overall results indicate that different species have different response to climate even when
situated at the same site locations (as in case of temperature, the two species from Nalter showed
different response), or different species may exhibit a similar response to climate even when
situated at different area (like in case of precipitation, all species exhibited similar response).
107
Figure 5.2: Graphs representing correlation coefficients between residual chronologies and
temperature of Gilgit meteorological data from ten sites respectively for a 13 month span.
Shaded areas indicate the significant months over 13 month period. (To be continued.,)
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Kargah
-0.6
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Jutial
-0.6
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Haramosh
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icin
ets
Picea smithiana Bagrot
108
Figure 5.2: Graphs representing correlation coefficients between residual chronologies and
temperature of Gilgit meteorological data from ten sites respectively for a 13 month span.
Shaded areas indicate the significant months over 13 month period. (To be continued.,)
-0.2
0
0.2
0.4
0.6
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Nalter
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nt
Picea smithiana Chera
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Chaprot
-0.6
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
cie
nts
Juniperus excelsa Chaprot
109
Figure 5.2: Graphs representing correlation coefficients between residual chronologies and
temperature of Gilgit meteorological data from ten sites respectively for a 13 month span.
Shaded areas indicate the significant months over 13 month period. Values on the y-axis are
correlation coefficients.
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
iice
nts
Juniperus excelsa Nalter
-0.6
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Pinus gerardiana Chaprot
110
Figure 5.3: Graphs representing correlation coefficients between residual chronologies and
precipitation of Gilgit meteorological data from ten sites for a 13 month span. Shaded areas
indicate the significant months over 13 month period. (To be continued.,)
-0.2
0
0.2
0.4
0.6
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Kargah
-0.2
0
0.2
0.4
0.6
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Jutial
-0.2
0
0.2
0.4
0.6
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Haramosh
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Bagrot
111
Figure 5.3: Graphs representing correlation coefficients between residual chronologies and
precipitation of Gilgit meteorological data from ten sites for a 13 month span. Shaded areas
indicate the significant months over 13 month period. (To be continued.,)
-0.4
-0.2
0
0.2
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Nalter
-0.2
-0.1
0
0.1
0.2
0.3
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Chera
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Picea smithiana Chaprot
-0.2
0
0.2
0.4
0.6
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Juniperus excelsa Chaprot
112
Figure 5.3: Graphs representing correlation coefficients between residual chronologies and
precipitation of Gilgit meteorological data from ten sites for a 13 month span. Shaded areas
indicate the significant months over 13 month period. Values on the y-axis are correlation
coefficients.
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Juniperus excelsa Nalter
-0.2
-0.1
0
0.1
0.2
0.3
0.4
pO pN pD J F M A M J J A S O
Co
rre
lati
on
co
eff
icie
nts
Pinus gerardiana Chaprot
113
Correlation and response function analyses were performed to find out the consistency among
the results. The four Tables (5.1-5.4) explain the comparison among residual chronologies and
Gilgit meteorological and grid data. Only significant correlation signs are inserted in the Tables.
High variance (chronologies versus Gilgit meteorological data) was obtained in correlation
analysis (Table 5.1) as compared to the variance obtained (chronologies versus grid data) in
correlation (Table 5.2). Besides this, the highest variance was seen for the Chaprot site (Pinus
geradiana and Juniperus excelsa) for both local and grid comparison, respectively (Tables 5.1,
5.2). Juniperus excelsa from Nalter exhibited the lowest variance in both local and grid
comparison (Tables 5.1, 5.2).
We then summarized these Tables by counting the positive and negative significant signs (Table
5.5) to pick the best seasons that had the most influence on tree growth. In the case of
temperature, almost all chronologies showed significant positive response to winter season,
particularly so for previous December and current January (Table 5.5). The spring season
(March-June), had a strong negative effect on tree growth. While in the case of precipitation,
significant positive response of spring season (February-May) to tree growth was identified.
Hence in both cases (temperature and precipitation), the spring season response of March-May is
dominant across all the chronologies.
114
Table 5.1: Summary of correlation function between tree-ring chronologies and monthly temperature and precipitation data from
Gilgit station
Temperature Precipitation
Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O Total Variance
PSKAR + + + + 43.35%
PSJUT + - - - - + 62.20%
PSHAR + - - - + + + 55.34%
PSBAG + - - - + + + + 59.18%
PSNAL + + + + + + - 60.38%
PSCHR + + + + + 57.73%
PSCHP + - + + 39.55%
JECHP + - - - + + 59.31%
JENAL - - - 36.84%
PGCHP + - - - - - + + 62.96%
PSKAR= Picea smithiana from Kargah, PSJUT= Picea smithiana from Jutial, PSHAR= Picea smithiana from Haramosh, PSBAG= Picea
smithiana from Bagrot, PSCHP= Picea smithiana from Chaprot, JECHP= Juniperus excelsa from Chaprot, JENAL= Juniperus excelsa from
Nalter, PGCHP= Pinus gerardiana from Chaprot
115
Table 5.2: Summary of correlation functions calculated from tree-ring chronologies and monthly temperature and precipitation data
from the relevant 0.5o
grid climate database (Mitchel and Jones, 2005)
Temperature Precipitation
Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O Total Variance
PSKAR + + - 21.86%
PSJUT - - - - + + + + 33.73%
PSHAR - - - + + + 34.59%
PSBAG - - + + + + 38.94%
PSCHR - + + + 31.95%
PSNAL + + + - 37.68%
PSCHP + + + 33.10%
JECHP - - + + + + + 44.35%
JENAL - - + + + 30.82%
PGCHP - - - - + + + + + 40.12%
116
Table 5.3: Summary of response function between tree-ring chronologies and monthly temperature and precipitation data from Gilgit
station
Temperature Precipitation
Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O
PSKAR + + - - - - + + + - - -
PSJUT + + - - - + + +
PSHAR + + + - - - + +
PSBAG + + - - - - + + + + + + -
PSNAL + + - + + + + + - -
PSCHR + + + - - - + + - + +
PSCHP + + - - + + +
JECHP - + + - -
JENAL - + - - +
PGCHP + + + - - - - - + + +
117
Table 5.4: Summary of Response functioncalculated from tree-ring chronologies and monthly temperature and precipitation data
fromrelevant Grid data (Mitchell and Jones, 2005)
Temperature Precipitation
Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O
PSKAR + - - + - - - -
PSJUT + - - + +
PSHAR + - - + + + -
PSBAG + - - + - + + + + +
PSNAL + + + + + + + -
PSCHR + - - + + + + +
PSCHP + - + + + -
JECHP - + + + + + +
JENAL + + - - - - + + + +
PGCHP + - - + + + - +
118
Table 5.5: Summary of four tables (5.1-5.4) including only significant signs of positive and negative correlation and reponse analysis
Temperature Precipitation
pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O
Positive 0 0 13 2 1 0 0 0 1 1 0 0 0 0 0 1 0 5 8 6 1 0 0 0 0 0 Table
5.1
Negative 0 0 0 0 0 3 4 6 6 3 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0
Positive 0 0 0 0 0 0 0 0 1 1 0 0 0 0 2 2 2 8 6 5 7 1 0 0 0 0 Table
5.2
Negative 2 0 0 0 0 2 2 5 2 3 2 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0
Positive 0 1 9 7 4 1 0 0 1 1 0 0 0 4 4 0 0 3 8 6 3 0 0 0 0 0 Table
5.3
Negative 2 0 0 0 1 5 6 2 3 2 1 2 2 1 0 0 1 0 0 0 0 2 2 2 0 0
Positive 2 1 1 5 1 0 1 0 3 2 0 0 0 2 3 0 1 3 5 7 6 1 5 0 0 1 Table
5.4
Negative 0 0 0 0 1 6 4 4 1 1 1 0 0 0 0 0 0 0 0 0 0 4 0 1 0 1
Cumulative
Positive
2 2 23 14 6 1 1 0 6 5 0 0 0 6 9 3 3 19 27 24 17 2 5 0 0 1
Cumulative
Negative
4 0 0 0 2 16 16 17 12 9 4 2 2 1 0 0 1 0 0 0 0 9 2 3 0 1
119
5.6-Discussion
Correlation analysis was performed to assess the relationship between tree growth and climate,
using nearby station and gridded data, in conjunction with traditional Response function analysis
(Fritts, 1976). A window of 13 months from previous October to current October was used
because tree growth season in this area is considered to initiate around March and terminate by
the end of September, with some carry-over effects from the prior year.
The important consideration for dendroclimatic studies is the potential age of trees (Ahmed et al.
2011). For example, Picea smithiana from some sites exceed 500 years (see section 4) and older
ages are obtained using the same species from India (Singh et al., 2004; Singh and Yadav, 2007).
Juniperus excelsa by Esper et al. (1995) from Morkhun were found to be the oldest with some
trees significantly greater than 1000 years and no other species were older than junipers. We
have collected samples from the same place but were unable to crossdate due to difficulties in
accounting for locally absent and/or false rings.
The correlation between site chronologies declined with increasing separation distance (Ahmed
et al. 2011). This was observed both among sites of the same species and among sites composed
of different species. A much stronger correlation was sometimes found between two different
species growing at the same site than between sites of the same species with a little separation of
0.5 kilometers. These findings sustain the practice of dense multi-species tree-ring network for
better spatial and temporal coverage to account for the effects of local topography. The best
vision for this in the Karakoram Range appears in terms of Cedrus deodara and Pinus
gerardiana and matches with the reported studies made by neighboring India by Borgaonkar et
al. (2009). However, in another study based on seven sites from Karakorum Range of Northern
Pakistan, this trend is not seen (Ahmed et al. 2012). Our results match with the findings of
Ahmed et al. (2012) where we observed Juniperus excelsa and Picea smithiana from the same
site (Nalter) but found no significant correlation. It may therefore not be necessary that
correlation of site chronologies declined with increasing distance.
The heterogeneous nature of the tree-ring network with respect to distance between chronologies
is identical to the findings made by Archer and Blenkinship (2010), who established high
120
heterogeneity among different climate station data in the Karakoram, and claimed that the
heterogeneity would make for a successful reconstruction of hydroclimate over the Karakoram.
A summarized result of significant monthly climate correlations and percentage of variance
explained were chosen. Low correlation occurred in the comparison of tree ring indices and grid
climate as compared to correlation between tree ring indices and local climate, possibly due to
the fact that grid climate has been interpolated to cover a large spatial area, whereas local
meteorological data reflects the climate of a specific site.
The same pattern across all the species can be seen for temperature in this study, owing to
temperature‟s greater spatial homogeneity than for rainfall. March-June temperature revealed the
most significant negative correlation in our analyses, suggesting that these months have a
negative influence on tree growth if the temperature surges. Higher temperature increase
evapotranspiration, and results in the decrement of soil moisture leading to low tree growth.
There is a strong positive correlation to previous December in almost all species, which indicates
that all species require higher than average temperature in this month for better growth.
A strong positive correlation was also observed during correlation and response analysis in
spring (February-April) rainfall suggesting that spring rainfall enhances tree growth. The rainfall
response was positive with the greatest correlations for February-May. No months were found to
be significant during summer and autumn months (June-September), regarded as the monsoon
period, because our sites are located in a dry temperate area where the monsoon winds have very
little influence. A similar broad outline was also seen in response function analysis, which
supported the outcomes of the correlation function.
Treydte et al. (2006) used Juniperus excelsa from Bagrot site (Hunza). They detected the highest
correlated month for precipitation was July. Their study was based on oxygen isotope
concentrations. We have Picea smithiana from the same site but correlated months are February-
May. The different rainfall response between Treydte et al. (2006) and present study might be
the elevation. Treydte et al. (2006) selected low and high elevations Juniperus excelsa. We have
Picea smithiana from the same site at the elevation of 3130 m but it did not show the similar
response. The difference in results may be the difference in elevation in species or amount of
moisture in soil. Juniperus excelsa used in Treydte et al. (2006) reconstruction grow on the drier
121
site of the dry temperate area whereas Picea smithiana grow on better sites where the moisture is
available. It might also be suggested that Treydte et al. (2006) and Yadav et al. (2002) studies
used RCS which created an artificial positive trend bias in their results – purely an artifact of the
procedure.
Ahmed et al. (2011) found a poor correlation to gridded climate data by the high elevation site
Bagrot 5 compared to the low elevation site Bagrot 1. The Morkhun site was also used by
Treydte et al. (2006) for rainfall reconstruction, and was shown to have a third type of response –
cold and dry. These sites were more than 1000 years old from high elevation (3900 m). The
explanation was given by Ahmed et al. (2011) that residual or weakened bands of cloud from
summer monsoon only reach the highest zones of forest, driven by the strong orographic effects
of mountain ranges. As our sampling sites are located either below or away from the zone
influenced by the summer monsoon there is a difference in summer rainfall correlations.
Cedrus deodara and Pinus gerardiana from district Chitral were analyzed for its dendroclimatic
potential which disclosed significant negative correlation with temperature and significant
positive correlation with precipitation in the spring season (Khan, 2011). The results of Khan
(2011) support the current study, which agrees with the findings of Wahab (2011) who worked
over district Dir by using Cedrus deodara and Picea smithiana.
Cook et al. (2003) presented a network of 32 sites from Nepal, including Picea smithiana and
Pinus wallichiana to produce temperature reconstruction from two species that share a positive
correlation with summer (June-July) temperatures. Here we have positive correlation to June-
July temperature only in the case of Picea smithiana from Nalter, while Picea smithiana from
other sites did not respond the same. Perhaps this is because Nalter is a cold place surrounded by
snow covered peaks and covered with clouds throughout the year with plenty of soil moisture.
Therefore hot summers might be expected to enhance tree growth of tree as in the case of Nepal
(Cook et al. 2003).
The current study sites are located at higher elevations and receive extensive winter snowfall.
This winter snowpack results in more available soil moisture than at other dry temperate sites, so
increased moisture loss may not considerably affect the moisture available later in the year. High
temperature in winter can favor rapid net photosynthesis and increased physiological activity that
122
can lead to the early initiation of cambial activity, rapid growth or formation of wide rings
(Tranquillini, 1964; Fritts, 1976). This clarifies the positive response of winter temperature,
especially in December, to the variations between chronologies. Borgaonkar et al. (2009) showed
a similar pattern of relationship for Cedrus deodara of neighboring India from the Western
Himalaya. Singh et al. (2009) also detected a positive relationship of tree-ring index series with
winter (December-February) temperature and summer precipitation. Our data explains that most
of the sites expressed a positive relationship with December.
In the spring (March-May), the situation is different. The temperature rises gradually above the
average annual value while the amount of precipitation is very small, which leads to reduced
ring-width formation due to higher evapotranspiration. Therefore, more rainfall in these months
(March-April) is conductive to better growth. Several other studies also indicated similar
response of pre-monsoon (spring season) on Western Himalayan conifers (Borgaonkar et al.
1994, 1996; Yadav et al. 1999; Yadav and Singh, 2002).
5.7-Conclusion
It was concluded that growth for all three species was directly affected by a combination of
temperature and precipitation. The tree-ring data were positively correlated with previous winter
temperatures with highest correlation for previous December to current January, and negatively
correlated with temperature over the entire period of the spring season with the strongest and
most consistent relationships for March-June. Chronologies were also positively correlated with
spring season precipitation with the highest relationship with February to May.
123
Chapter No. 6
Temperature reconstruction
6.1-Introduction
Several studies highlight the importance of reconstructing past temperature variability before the
instrumental period for comparing the natural and anthropogenic climatic changes (Briffa and
Osborn, 1997; Jones et al. 1996; Man et al. 1998). These studies have been conducted primarily
with conifers, like Abies pindrow, Cedrus deodara, Picea smithiana, Pinus geradiana and Pinus
wallichiana from the Himalayan region (Ahmed et al. 2011; Bhattacharya and Yadav, 1999).
Conifers including Juniperus excelsa were used to reconstruct past climate back to AD 600 from
the Karakorum Range (Esper et al. 2002). Tree-ring chronologies from the Himalayas of Nepal
were used to reconstruct the past 400 years of temperature (Cook et al. 2003). Recently, Cook et
al. (2013) successfully reconstructed past 500 years of Indus river flow by using tree ring
chronology in comparison with Partab flow data. In the present study, we reconstruct past
temperature more than 400 years by using Picea smithiana, Juniperus excelsa and Pinus
gerardiana chronologies from sites at Gilgit as these substitute records, will provide valuable
data for climate change studies with regional and global perspective.
One current dilemma is the conflicting reports that some Himalayan glaciers are rapid retreating
when they are in fact advancing (e.g. Owen, 2009; Cook et al. 2013). Here, we intend to predict
the temperature influences and investigate the similarity and synchronicity of past responses to
known major cooling events such as the “Little Ice Age” (LIA) (Luckman &Villalba, 2001).
This will provide supporting information to WAPDA that will contribute efforts attempting to
model future outcomes.
124
6.2-Materials and methods
For temperature reconstruction we followed the calibration and verification techniques of Fritts,
(1976) and Cook and Kairiukstis, (1990). Precipitation reconstruction model was performed
separately but we got poor results and it is still important to explore the past precipitation in
future. The four months which showed significance in correlation analysis were used to make a
season (March-June) as these months had the most influence on tree growth and relationship was
stable in time. Tree ring indices were used as the predictors and local temperature was used as
the predictand. The point by point regression option was chosen with a predictor and predictand
common period from 1955-2008, and this time span was split into two equal calibration and
verification period.
Simple linear regression was used to transform the tree ring data into the estimates of four month
window of temperature. The data were divided into two periods (1955-1985 and 1986-2005)
prior to regression, using one for calibration and other for verification, respectively, and then
reversing the order for comparison of both periods. The statistics used for the calibration and
verification periods are the Pearson correlation coefficient, Spearman correlation coefficient,
Reduction of error (RE) and coefficient of efficiency (CE). These values are often lower in
verification period as compared to calibration period. If RE and CE are negative then regression
estimates is bad in verification period. Positive RE and CE is the evidence of validity of
regression model (Cook et al. 1994; 1999). However, according to Brendon M. Buckley in
personnel communication, the RE and particularly CE in the verification periods can be thrown
way off by one or two anomalous values. Any positive value for both implies model fidelity, and
negative values implies poor model fidelity, but that lack of fidelity can often be explained by a
single bad value. A simple plot of the act-est data can go a long way to explaining this. “PCReg”
from the LDEO (Lamont Doherty Earth Observatory) TRL website program was used to develop
the reconstruction of temperature for the past 400 years. The gauge data was also seasonalized
using “SEA” program from the packaged software DPL.
Reconstructions were carried out between tree-ring chronologies and Gilgit meteorological data
by picking the significant months from correlation analysis (Table 5.1) to make a season as
correlation analysis exhibited highest percent variance in overall correlation and response
functions. Reconstruction model was performed one by one to the tree ring chronology and
125
Gilgit local data to check out which chronology best suited for reconstruction. Picea smithiana
showed negative correlation with March-June (spring season) temperature, whereas Juniperus
excelsa showed negative correlation May-July of the growing period (see Table 5.1).
6.3-Results
Here we present a reconstruction of March-June temperature variation for the past 400+ years
using ring-width chronologies of Picea smithiana, Juniperus excelsa and Pinus gerardiana from
Gilgit and Hunza valleys. Table 6.1 shows regression analysis results for the ten chronologies.
Five of the sites accounted for the maximum retained variance from the model, while the other
five were rejected from further analyses. The five retained sites include Pinus gerardiana from
Chaprot, Picea smithiana from Jutial, Picea smithiana from Haramosh and Bagrot. Picea
smithiana from Kargah, Nalter, Chera and Chaprot were rejected. A dramatic change happened
in Juniperus excelsa Nalter which has the ability to make season but failed to pass regression
analysis.
The highest explained variance was observed in Pinus gerardiana Chaprot, but the data are too
short i.e. only more than 150 years for reconstruction. (EPS value reliable up to 1840 see section
4) whereas Picea smithiana from Jutial data is reliable up to 1530 (EPS>400) therefore selected
for further reconstruction.
126
Table 6.1: Regression analysis of ten chronologies with Gilgit temperature data from different
sites of Gilgit and Hunza valleys.
Chronology used Months (seasons) Variance explained
Picea smithiana Kargah No season No regression
Picea smithiana Jutial 4 months (March-June) 38.16%
Picea smithiana Haramosh 3 months (April-June) 13.94%
Picea smithiana Bagrot 3 months (April-June) 16.77%
Picea smithiana Nalter No season No regression
Picea smithiana Chera No season No regression
Picea smithiana Chaprot No season No regression
Juniperus excelsa Chaprot 3 months (May-July) 32.71%
Juniperus excelsa Nalter 3 months (May-July) No regression
Pinus geradiana Chaprot 5 months (March-July) 44.50%
The correlation analysis indicated that tree growth was affected by March to June temperatures,
the season of reconstruction hence a linear regression model was developed to reconstruct past
temperature for the Gilgit region. During the common period of tree ring index and station data
(1955-2008), the reconstruction accounted 38.16% of the variance. Split calibration and
verification were employed to assess the statistical reliability of this model. The reduction of
error (RE) and coefficient of efficiency (CE) statistics were both positive in the verification
period, indicating significant skill in the tree ring estimates (Fritts, 1976) (Table 6.2a and 6.2b).
The model calibrated on the early period (1955-1985) explained 38.16% of the variance in
March to June temperatures, while the model based on late period (1985-2008) expressed 21.9%
of the variance. Both models passed all test of verification. RE>0.3, CE>0.1 demonstrated the
excellent performance of both models. The tree-ring data explains 28.48% of the variance in
temperature using the entire period of 1955-2008 for calibration.
127
Because of the better verification statistics and variance (Tables 6.2a and 6.2b), the early
calibration model was used to reconstruct (March-June) temperature back to 1523.
Table 6.2a and 6.2b: Calibration and verification statistics for the early (a) and late (b) periods.
RP= Pearson‟s product moment correlation coefficient; RR= the robust correlation coefficient;
RS= the spearman‟s coefficient of rank correlation; RSQ= variance explained; RE= Reduction of
error; CE= coefficient of efficiency.
Table 6.2a: Early calibration
Calibration (1955-1985) Verification (1986-2008)
Statistics Value Statistics Value
RP 0.618 RP 0.468
RR 0.615 RR 0.438
RS 0.518 RS 0.348
RSQ 0.381 RE 0.382
CE 0.197
Table 6.2b: Late calibration
Calibration (1986-2008) Verification (1955-1985)
Statistics Value Statistics Value
RP 0.468 RP 0.618
RR 0.438 RR 0.615
RS 0.348 RS 0.518
RSQ 0.219 RE 0.219
CE 0.355
From the Fig. 6.1, it is clear that the reconstructed data displayed similar trends and amplitude as
the observed data over most of the common period. The reconstructed temperatures were higher
(by more than 1oC) than the observed data in the following years: 1957 and 1979. The
reconstructed temperatures were found to be lower than the recorded data by more than 1oC in
1956 and 1980. No year represented the most serious disagreement between the two records.
128
Fig. 6.2 shows the linear relationship between instrumental record and reconstructed data for the
same period. The two data sets are fairly correlated (Pearson product moment correlation;
RP=0.53) by correlation equation of y=8.90+5.11x.
Figure 6.1: Actual (red) and reconstructed (dashed) March-June temperature reconstruction during
common period 1955-2008. The estimation explains 38.16% of the actual variance in this common
period.
Figure 6.2: Scatter plot of the observed and reconstructed temperature of the data that were used
for early calibration period (1955-1985). X-xis represented actual data while Y-xis represented
reconstructed data for the common period (1955-2008).
On the basis of RBar and EPS statistics (see Section 4); Picea smithiana from Jutial
reconstruction is reliable back to 1523 so the proxy record is more than 400 years longer than the
15
16
17
18
19
20
21
22
1950 1960 1970 1980 1990 2000 2010
reconstructed actual
129
instrumental record. The mean temperature obtained over the entire period of reconstructed data
was 18.22oC just lower than the mean of March-June temperatures calculated from Gilgit actual
data (actual data=18.42oC). The Table 6.3 displays that values of basic statistics are quite similar
with each other.
Table 6.3: Statistics for the March-June actual and reconstructed (1955-2008) temperature data
Actual data Reconstructed data
N (years) 54 54
Mean (oC) 18.42 18.32
Median (oC) 18.40 18.42
Standard deviation (oC) 0.88 0.84
Standard Error (oC) 0.12 0.11
Minimum (oC) 16.7 16.8
Maximum (oC) 20.8 20.4
The reconstruction March-June temperatures for the entire period are shown in the Fig. 6.3. Ten
years running mean window described the tendency of the warming and cooling trends (Fig. 6.5)
over a decadal time scale (short term period). Spring temperatures have been steadily increasing
over the eleven intervals of the record considered as warm periods; 1564-1573, 1590-1608,
1615-1626, 1630-1650, 1692-1714, 1768-1787, 1794-1817, 1821-1834, 1854-1869, 1909-1922
and 1953-1979 whereas it have been decreasing in ten intervals of the record also known as cold
periods; 1537-1551, 1574-1589, 1651-1671, 1675-1691, 1735-1749, 1755-1768, 1837-1853,
1870-1908, 1923-1938 and 1980-1991. The warmest anomaly occurred in 1602, 1807 and 1862
with the temperature values 18.81, 18.71 and 18.77 respectively. The coldest anomaly
throughout the data was recorded in 1684 and 1901 where the temperature values were 17.47 and
17.41 respectively.
130
Tables 6.4 (a) and (b): The warmest 6.4(a) and coldest 6.4(b) non overlapping 25 year‟s periods
and the warmest 6.4(a) and coldest 6.4(b) non overlapping 10 years periods of Gilgit and Hunza
Valleys temperature reconstruction in oC. Dep. = departures from the mean of 18.22
oC over the
entire reconstruction period of 1523-2000.
Table 6.4(a): Warm periods.
Warm 25 years intervals Warm 10 years intervals
Intervals Mean Departures Intervals Mean Departures
1630-1654 18.42 0.20 1537-1546 18.43 0.21
1768-1792 19.91 1.69 1574-1583 18.93 0.71
1794-1818 18.41 0.19 1615-1624 18.88 0.66
1890-1904 19.25 1.03 1630-1639 18.42 0.20
1953-1977 18.33 0.11 1675-1684 18.64 0.42
1768-1777 19.91 1.69
1794-1803 18.40 0.18
1890-1899 19.20 0.98
1900-1909 18.61 0.93
1964-1973 18.41 0.19
Table 6.4(b): Cold periods.
Cold 25 years intervals Cold 10 years intervals
Intervals Mean Departures Intervals Mean Departures
1651-1675 18.07 -0.15 1564-1573 17.97 -0.25
1692-1716 17.70 -0.52 1600-1609 17.22 -1.00
1870-1894 17.80 -0.42 1661-1670 15.86 -2.36
1692-1701 16.93 -1.29
1703-1712 17.80 -0.42
1735-1744 17.85 -0.37
1779-1788 17.15 -1.07
1821-1830 17.93 -0.29
1854-1863 17.42 -0.80
1870-1879 17.84 -0.36
1910-1919 16.52 -1.70
1923-1932 17.82 -0.40
1980-1989 17.29 -0.93
131
Tables (6.4a and 6.4b) describe the coldest and warmest periods over 10 years and 25 years non
overlapping intervals in the reconstruction. The warmest ten-years periods occurred during 1768-
1777 where the mean temperature was 19.91oC i.e. 1.69
oC above the average 1523-2008
reconstructed mean. This warm period was also evident in the 25 years reconstructed interval.
The coldest 25-year interval was seen during 1651-1675 where the lowest anomaly happened in
(1661-1670) ten years interval with the mean temperature of 15.86oC lower than 2.36
oC than the
average reconstructed temperature.
Overall, the results reveal that the warmest periods occurred in the second half of 18th
century
(roughly concurrent with the large “mega-droughts” noted for Southeast Asia and beyond, by
Buckley et al. (2007; 2010), Cook et al. (2010), D‟ Arrigo et al. (2012), and Sano et al. (2009)
and the very last and early periods of 19th
and 20th
centuries, respectively. The coldest periods
were observed for the 1660s, 1780s and 1910s. It is also calculated that after the long 20 years
warm interval (1890-1909), cold ten years occurred later on i.e. 1910-1919.
To find out temperature anomalies over centennial time scale (long term period); we used 100
years moving average presented in Fig. 6.5.The 17 century was apparently cool in first half of
the 17th
century and warm during 1640-1660 which then changes to average over the entire
period of 1740 in the present reconstruction. Thereafter, 1740 to 1760 experienced cooling
period with the minimum cooling anomaly (1748-1788). The 19th
century recorded the protracted
warmth period including last two decades of 18th
century and first decade of 20th
century. The
highest anomaly was observed during 1860-1880. The 20th
century was on the average scale
representing absence of any warmth period during the century.
132
Figure 6.3: The Gilgit March-June reconstruction over the entire period of 1523 to 2008.
Figure 6.4: 10 years running mean window describes the trend of warm and cold years. Upper
line of graph represents warm years and lower line signifies cold yearsduring 25 years intervals.
Figure 6.5: 100 years running mean window describes the trend of warm and cold years. Red
line of the graph represents 100 mean running window during 1620-2000.
15
16
17
18
19
20
21
1500 1600 1700 1800 1900 2000
Tem
pe
ratu
re o
C
year
15
16
17
18
19
20
21
1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000
Tem
pe
ratu
re (
C)
years
133
6.4-Comparison with Pinus gerardiana reconstruction
Pinus gerardiana Chaprot showed good correlation in correlation function (see chapter 5, Table
5.1) and also represented strong variance in transfer function (Table 6.1), therefore picked for
reconstruction also. The four months similar to Jutial were chosen for reconstruction over the
period of approximately 150 years (1850-2008). The Chaprot reconstruction was then compared
with the Jutial reconstruction within the same period (1850-2000). Figure 6.6- shows the Picea
smithiana Jutial and Pinus gerardiana Chaprot temperature reconstruction plots. While the two
records exhibited a reasonable degree of similarity over most of the common periods, there are
shorter intervals where the two proxies were significantly different. From the beginning up to
1870, Jutial temperatures were lower than Chaprot temperatures with the exception of a short
interval that occurred where Jutial temperatures were tend to higher for the next five years and
then long term decline happened from 1876 to 1920. Higher Jutial temperatures (compared to
Chaprot reconstruction) occurred in the late 1950s to 1990s and then dropped in the last few
years.
Figure 6.6: Comparison between the two temperature reconstructions based on 10 years moving
average. Blue line indicates Pinus gerardiana Chaprot while red line shows Picea smithiana
Jutial reconstruction.
The interesting periods where Jutial reconstruction caused a very sharp drop were 1860-1870 and
1910-1920. Over the entire common period (1840-2008), the mean temperature calculated in
Jutial was 17.83oC, less than the Chaprot reconstruction where the mean was 18.35
oC.
17
17.5
18
18.5
19
1825 1850 1875 1900 1925 1950 1975 2000
Tem
pe
ratu
re o
C
10 per. Mov. Avg. (pgchap) 10 per. Mov. Avg. (psjutl)
134
The correlation between two records was found highly significant (The Pearson‟s product
moment correlation coefficients: RP = 0.60, P<0.001) over the common interval of years (1840-
2008). However, as described earlier, the strength of correlation was not stable over time.
Changing correlations between 25 years sub-periods between two reconstructions are well
explained (Fig. 6.6). Here, we divided the reconstructed data into 50 years interval for
comparison.
Figure 6.7: Scatter plot of the Jutial and Chaprot temperature reconstructions based on 168 years
of data (1840-2008).
The three intervals were found to be highly significant (P<0.001), but there were less or more
values among three periods. The highest correlation occurred during second half of the 20th
century (Pearson‟s product moment correlation coefficients: RP = 0.60, P<0.001). The weakest
agreement was seen in first half of the 20th
century where RP = 0.57 although still significant
(P<0.001). The most disagreement years were observed i.e. 1945 and 2000 where the two
reconstructions have the values 16.45oC, 18.62
oC (Jutial) and 18.68
oC, 20.34
oC (Chaprot)
respectively.
15
16
17
18
19
20
21
15 16 17 18 19 20 21 22
Juti
al r
eco
nst
ruct
ion
in o
C
Chaprot reconstructed in oC
135
6.5-Discussion
Picea smithiana tree-rings performed well as predictors of Gilgit temperatures. This is directed
by a good amount of variance explained and also by stability of the models for both early and
late periods. The calibration and verification statistics showed similar and reliable values. Pinus
gerardiana explained the highest variance in the same months probably because of existence of
this species at lower elevation. The negative influence of spring temperature on tree growth
means that the high temperature leads to internal water deficit in the early growing season due to
increased soil moisture loss by evapotranspiration.
Other temperature sensitive tree-ring records in the surrounding area offer a reference to the
validation of our reconstruction in the Asian region. Published tree ring temperature records for
India (e.g. Hughes, 1992; Borgaonkar et al. 1996; Yadav et al. 1997), Western Central Asia
(Esper et al. 2002), Nepal (Cook et al. 2003) and Tibet (e.g. Wo, 1992; Brauning, 1994; Wo and
Shao, 1995) are difficult to compare with the present study results, possibly due to the
differences in reconstructed seasons or the differences in tree ring species. However, some
similarities have been noted at inter-decadal and century time-scale.
Esper et al. (2002) described the past 1300 year climatic history for Western Central Asia from
tree-rings. The studies were carried out from Tien Shan of Kirghizia and Northwest Karakorum
of Pakistan. Most of the samples were collected from Juniper. Juniperus turkestanica chronology
from Tien Shan site was found to be significantly correlated with Tien Shan summer temperature
(July-August) of nearby station (Narin). The warmest decades occurred between AD 800-1000
and the coldest decades occurred in the first half of the 17th
century. We found cold years in the
first decade of the 17th
century in decadal time scale, and in the first half of 17th
century over the
century time scale. The results are in agreement with the conclusions of Esper et al. (2002).The
cold periods of the early 20th
century were observed in the Yantze River temperature
reconstruction on the Tibetan Plateau which agrees well in present study (Liang et al. 2008),
however, the difference occurred in summer and spring season. Although Esper et al. (2002) and
Liang et al. (2008) studies were based on summer temperature reconstruction (July-August) our
study was based on spring temperature (March-June), and yet still some of the results match.
136
The reason may be the occurrence of tree-ring species. The current reconstructed study sites
(Gilgit and Hunza) and Southern Tien Shan are some 500 km distant, and the Tibetan Plateau is a
similar distance from the current study sites. The current reconstructed sites were present in
Northwest Karakorum and Himalayan regions, and therefore correlate with spring season
variability. Sites in the Northwest Karakorum are influenced by westerly winds and as well as
monsoonal depression whereas sites (used in Esper et al. 2002 reconstruction) in Southern Tien
Shan and Tibetan Plateau are affected by strongly continental climate without transportation of
precipitation from Arabian Sea. Therefore chronologies from Tien Shan and Tibetan Plateau
affected by summer temperature whereas our study sites influenced by spring temperature.
Another possibility is the selection of tree species. Esper et al. (2002) reconstruction based on
Junipers which was affected by summer temperature as in case of present study where Juniperus
excelsa from both sites also showed negative correlation in the months of June and July. Current
reconstruction was established using Picea smithiana out of three species having influence with
spring temperatures.
Several studies have been carried out in Himalayan region that propose that climatic variability
in this area correlate with spring temperature and El-Nino/Southern Oscillation (Dey and
Bhanukumar, 1983; Douville and Royer, 1996; Li and Yanai, 1996; Overpeck et al.
1996).Various tree-ring chronologies in the Western Himalayan Region have been used in
extending seasonal climatic records (e.g., Bhattacharya and Yadav, 1999). In this contest,
Hughes and Davis (1987) made a pioneer effort in the analysis of Abies pindrow and Picea
smithiana in the Kashmir valley which later contributed detailed reconstructions of mean
temperatures for spring (April-May), late summer (August-September) precipitation since 1780
at Srinagar, Jammu and Kashmir based on width and density of annual rings of Abies pindrow
(Hughes, 1992). Borganonkar et al. (1996) reconstructed pre-monsoon (March-May) temperature
back to 19th
century using ring-width data of Cedrus deodara from Simla and Kanasar.
Yadav et al. (1997) extended this data by April-May temperature reconstruction back to AD
1698 from Western Himalayan region of India. Cedrus deodara, Picea smithiana and Pinus
wallichiana were used in overall analysis. This study showed that the first three decades of the
18th
century (1700s-1720s) were warm in which warmest anomaly occurred about 1713-1722.
The 19th
century showed prolonged warmth in 1850s-1870s, while the period of 1810s-1830s
137
was remarkably cool. Any warming trend in recent decades of the 20th
century was not
evidenced in our reconstruction.
Our reconstruction has warm periods during the 19th
century from 1850 to 1870 with the
warmest 10 year mean anomaly for 1854-1869. The similar warmth period from 1850-1870 were
also observed by Yadav et al. (1997) but according to Bradley and Jones (1993), this century has
been described monotonically cool in high-altitude northern hemisphere regions. The trend of
parallel cooling decades was also experimented during 1730-1750 in present reconstruction and
the reconstruction of Yadav et al. (1997). The 1730s were also found to be cool in the spring
temperature reconstruction valley of Kashmir in the northwestern Himalaya (Hughes, 1992,
1994). The presented reconstructed temperature for 1900s, 1920-1930 and 1980s is also in
agreement with the reconstruction of Yadav et al. (1997) and also similar with the trend noted in
the spring temperature reconstruction in recent decades of the 20th
century in the valley of
Kashmir (Hughes, 1992, 1994). Mean March-June temperature reconstruction of present study in
inter-decadal time scale pointed out a weakened elevated temperature towards later parts of 20th
century in inter-decadal time scale which was believed to be the result of anthropogenic activities
(deforestation). The cool decades for the late 1830s were also experienced in current debate with
those in the British Isles, Central Europe, Scandinavia, the polar Urals and central Korea
(Hughes, 1994). So the impact of these cool decades over climate and tree growth through these
larger areas appears to have been severe.
Cook et al. (2003) reconstructed the past more than 400 years of Nepali temperature by using the
long term Khatmandu record for the five-month season of February-June. Our March-June
reconstruction compares well with the Cook et al. (2003) Nepal reconstruction. Interestingly,
there is no evidence for the abrupt extreme cold events in 1815-1822 (from the Tambora eruption
in Indonesia) in present reconstruction and Karakorum tree ring data reconstruction made by
Esper et al. (2002). Nor is there any support for it in Srinagar or Simla reconstructions. In
contrast, these cold events have been observed in Wo and Shao (1995), Brauning (1994), Liang
et al. (2008) and Cook et al. (2003) reconstructions from eastern Tibet and Nepal respectively. It
is plausible that the 1815-1822 cold event had its greatest impact over eastern Nepal and Tibet
and did not spread far into the western Himalayas and Karakorum. The tree ring network used
138
for the present reconstruction is geographically weighted towards the western Himalayas and
Karakorum.
Distinct cold and warm intervals can be seen from the current investigation, including a period
consistent with “Little Ice Age”. The 17th
century appears to be cooling whereas the 19th
century
is markedly higher than that of other centuries. NASA (2011) describes the occurrence of “Little
Ice Age” in three particularly cold intervals: 1650s, 1770s and 1850s which spread throughout
Europe, North America and Asia. The above periods were marked for the expansion of mountain
glaciers. These periods were also witnessed where we have coolest period with the highest
departures from the mean temperature. The warmest period of the last millennium (Medieval
warm period) could not be investigated as our reconstructed data is too short. Some studies show
that this warm period occurred before the occurrence of LIA (NASA, 2011; Mann et al. 2009).
This all talked about variations in temperature at the level of individual countries. A comparison
of our reconstruction temperature series for Gilgit and Hunza valleys agrees well against
temperature reconstruction covering the past millennium in central Asia (Shi et al. 2012).
Typical cooling and warming patterns appear to be better reflected in our results and Asian
millennium reconstruction (Shi et al. 2012). Comparing the two reconstructions (present and Shi
et al. 2012), some common features are evident. For example, the warming of the 19th
and the
cooling of 17th
century are pronounced in both. The results indicate that a cooling trend occurred
throughout central Asia in 17th
century, before temperature began to rise in the 18th
century
before reaching a maximum by the end of 19th
century.
6.6-Conclusion
Ring-width chronologies of Picea smithiana were used to reconstruct mean March-June (spring)
temperatures back to A.D. 1523. The record is based on a simple linear regression technique
where we calibrated our Picea smithiana chronology against March-June temperature. The
calibrated model explained 38.16% of the variance in temperature and passed all calibration and
verification tests at the 95% level of significance. The reconstruction exhibits a strong positive
correlation with the instrumental data and is characterized by annual to multiyear variations of
cool and warm periods. The 19th
century experienced a prolonged warmth period over a
centennial scale with the highest temperatures in the 1850s-1870s. The coldest 20 year period
139
was 1890-1910. The consistency observed on decadal scales between the present reconstruction
of March-June temperatures, the mean temperatures of April-May reconstruction from the valley
of Kashmir (Hughes, 1992, 1994) and the mean temperature reconstruction of April-May in the
Western Himalaya of India (Yadav et al. 1997) indicates the potential for reconstructing
regional-scale climatic changes using tree rings. The “Little Ice Age” can also been observed
from our reconstruction, and matches with the reported cooling intervals of NASA. It is
recommended that collecting more samples from other locations of the area from sensitive trees
would produce handful results to extend our chronology in finding of Medieval warm anomaly.
140
Refernces
Aceituno, P., M. Prieto, M. Solari, A. Martenez, G. Poveda, M. Falvey, 2009. The 1877-1878 El-
Nino episode: associated impacts in South America. Climate Change 92:389-416.
Ahmed, K., and S. Shah, 2007. Physical and microbial assessment of drinking water of Nomal
Valley, Northern Areas, Pakistan. Pak. J. of Zool., 39: 367-373.
Ahmed, M., 1987. Dendrochronology and its scope in Pakistan. In modern trends of Plant
Science Research in Pakistan. Dept. of Botany, Peshawar University.
Ahmed, M., 1988a. Population structure of some planted tree species in Quetta. J. Pure and
Applied Science, 7:25-29.
Ahmed, M., 1988b. Problems encountered in age estimation of forest tree Species. Pak. J. Bot.,
20(1): 143-145.
Ahmed, M., 1989. Tree-ring chronologies of Abies pindrow (Royle) Spach., from Himalayan
region of Pakistan. Pak. J. Bot., 21(2): 347-354.
Ahmed, M., and A. M. Sarangzai, 1991. Dendrochronological approach to estimate age and
growth rate of various species from Himalayan region of Pakistan. Pak. J. Bot., 23(1):
78-89.
Ahmed, M., and A. M. Sarangzai, 1992. Dendrochronological potential of few tree species from
Himalayan region of Pakistan. A preliminary investigation. J. Pure Applied Science,
11(2):65-72.
Ahmed, M., and H. Naqvi, 2005. Tree-ring chronologies of Picea smithiana and its quantitative
description from Himalayan region of Pakistan. Pak. J. Bot., 37(3): 697-707.
Ahmed, M., J. Palmer, N. Khan, M. Wahab, P. Fenwick and E. R. Cook, 2011. The
dendroclimatic potential of Conifers from Northern Pakistan. Dendrochronologia 29: 77-
88.
Ahmed, M., M. Wahab and N. Khan, 2009. Dendroclimatic investigation in Pakistan, using
Picea smithiana (Wall) Boiss. Preliminary result Pak. J. Bot., 51(5): 2427-2435.
Ahmed, M., M. Wahab, N. Khan, M. U. Zafar and J. Palmer, 2010b.Tree ring chronologies from
upper Indus Basin of Karakorum Range, Pakistan. Pak. J. Bot., SI 42:295-307, 2010.
Ahmed, M., N, Khan and M. Wahab, 2010a. Climatic response function analysis of Abies
pindrow (Royle) Spach. Preliminary results. Pak. J. Bot., 42(1): 165-171.
Ahmed, M., N. Khan, M. Wahab, U. Zafar and J. Palmer, 2012. Climate/Growth correlations of
tree ring species of the Karakorum Range, North Pakistan. IAWA Journal 33: 51-61.
Ahmed, M., T. Hussain, A. H. Shaikh, S. S. Hussain, M. F. Siddique, 2006. Phytosociology and
structure of Himalayan forests from different climatic zones of Pakistan. Pak. J. Bot.,
38(2): 361-363.
Anwar, M. S., S. Lateef and G. M. Siddiqui, 2010. Bacteriological quality of drinking water in
Lahore. Biomedica. 26:66-69.
141
Archer, D., and S. Blenkinsop, 2010. Heterogeneity and spatial representativeness of station
climate and tree ring records as climate change indicators in the Upper Indus Basin. Talk
presented at International Conference/Workshop on Dendrochronology held at Federal
Urdu University, Karachi. November 15, 2010.
Azhar, H., 1996. Determination of drinking water quality from source to consumer in Islamabad.
(Dissertation) Health services academy, Faculty of Medicine, Quaid-e-Azam University
Islamabad.
Benjamin, M., 2002. Water Chemistry. New York: McGraw-Hill.
Bhattacharya, A., C. Vadana and R. R. Yadav, 1999. Tree ring studies in the Eastern Himalayan
region: Prospects and Problems. IAWA Journal 20(3): 317-324.
Bhattacharyya, A., and R. R. Yadav, 1999.Climatic reconstruction using tree ring data from
tropical and temperate regions of India-A Review.IAWA Journal 20(3): 311-316.
Bhattacharyya, A., V. C. Lamarche and M. K. Hughes, 1992. Research Report, Tree Ring
Chronologies from Nepal. Tree Ring Bulletin 52: 1992.
Biber, K., 2011. Mercury fate and transport in Hunza and Gilgit rivers, Northern Areas, Pakistan-
A numerical modeling approach. GSA annual meeting in Minneapolis, 241-28.
Bilham, R., G. B. Pant and J. C. Jacoby, 1983. Dendroclimatic potential of Conifer trees from the
Sir Sar range in the Karakoram. Mausam 7: 45-50.
Blasing, T. J., A. M. Solomon and D. N. Duvick, 1984. Response function revisited. Tree Ring
Bulletin. 44: 1-15.
Borgaonkar, H. P., G. B. Pant, R. Kumar, 1994. Dendroclimatic reconstruction of summer
precipitation at Srinagar, Kashmir India since the last 18th
century. The Holocene 4(3):
299-306.
Borgaonkar, H. P., G. P. Pant, K. R. Kumar, 1996. Ring width variations in Cedrus deodara and
its climatic response over the Western Himalaya. International Journal of Climatology
16:1409-1422.
Borgaonkar, H. P., S. Ram and A. B. Sikdar, 2009. Assessment of tree-ring analysis of high
elevation Cedrus deodara D. Don from Western Himalaya (India) in relation to climate
and glacier fluctuations. Dendrochronologia 27: 59-69.
Bradley, R., and P. Jones, 1993. „Little Ice Age‟ summer temperature variations: their nature and
relevance to recent global warming trends. The Holocene 3, 4: 367-376.
Brauning, A., 1994. Dendrochronology for the last 1400 years in eastern Tibet. Geo Journal 34:
75-95.
Briffa, K. R., and T. J. Osborn, 1999. Seeing the wood from trees. Science 284: 926-27.
Briffa, K., and P. D. Jones, 1990. Basic chronology statistics and assessment. In E. R. Cook and
L.A. Kairiukstis (Eds.), Methods of Dendrochronology: application in the environmental
sciences (pp. 137-152). Dordrecht: Kluwer Academic Publishers.
142
Brown, D., and P. Rothery, 1993. Models in Biology: mathematics, statistics and computing.
Chichester: John Wiley and Sons Ltd.
Buckley, B. M., K. J. Anchukaitis, D. Penny, R. Fletcher, E. R. Cook, M. Sano, C. Nam, A.
Wichiekeeo, T. T. Minh and T. M. Hong, 2010. Climate as a contributing factor in the
demise of Angkor, Cambodia. Proceedings of the National Academy of Sciences of the
United States of America.
Buckley, B. M., K. Palakit, K. Duangsathaporn, P. Sanguantham and P. Prasomsin, 2007.
Decadal scale droughts over northwest Thailand over the past 448 years: links to the
tropical Pacific and Indian Ocean sectors. Climate Dynamics 29:63-71.
Champion, G. H., S. K. Seth and G. M. Khattak, 1965. Forest types of Pakistan. Pakistan Forest
Institute, Peshawar., pp-238.
Chaudhary, V., K. P. Singh, G. Jacks and P. Bhattacharya, 2001.Ground water contamination at
Ludhiana, Punjab, India. J. Indian Water Works Associa.,, 33: 251-261.
Chhetri, P. K., 2008. Dendrochronological analysis and climate change perceptions in Langtang
National Park, Central Nepal. Climate change and disaster impact reduction. School of
Applied Sciences, North Umbria University.
Chhetri, P. K., and S. Thapa, 2010. Tree ring and climate change in Langtang National Park,
Central Nepal. Our Nature 8:139-143.
Climate Normals Of Pakistan, (1961-1990). Pakistan Meteorological Department, Karachi.
Cook, E. R., 1985. A time series analysis approach to tree ring standardization. PhD dissertation,
University of Arizona, Tucson, USA.
Cook, E. R., and L. A. Kariukstis, 1990. Methods of Dendrochronology, Springer, New York.
Cook, E. R., and R. L. Holmes, 1986. Guide for computer program ARSTAN. Adapted from
user manual for program ARSTAN in Tree Ring Chronologies of Western North
America: California eastern Oregon and north great Basin.
Cook, E. R., C. Woodhouse, C. Eakin, D. Meko, & D. Stahle, 2004. Long term aridity changes in
the Western United States. Science 306:1015-1018.
Cook, E. R., D. M. Meko, D. W. Stahle and M. K. Cleaveland, 1999. Drought reconstructions for
the continental United States. Journal of Climate 12: 1145-1162.
Cook, E. R., J. Palmer, M. Ahmed, C. A. House, P. Fenwick, M. U. Zafar, M. Wahab and N.
Khan, 2013. Five centuries of Upper Indus Flow from tree rings. Journal of Hydrology
486: 365-375.
Cook, E. R., K. J. Anchukaitis, B. M. Buckley, R. D‟Arrigo, G. C. Jacoby and W. E . Wright,
2010. Asian Monsoon Failure and Mega drought During the Last Millennium. Science
328: 486-489.
Cook, E. R., K. R. Briffa and P. D. Jones, 1994. Spatial regression methods in
dendroclimatology: a review and comparison of two techniques. International Journal of
Climatology 14: 379-402.
143
Cook, E. R., P. J. Krusic and P. D. Jones, 2003. Dendroclimatic signals in long tree ring
chronologies from the Himalayas of Nepal. International Journal of Climatology 23 (7):
707-732.
Cook. E. R., and Kariukstis. L. A, 1992. Methods of Dendrochronology (Application in the
Environmental Science).
D‟Arrigo, R., K. J. Anchukaitis, B. M. Buckley, E. R. Cook, R. Wilson, 2012. Regional climatic
and North Atlantic Oscillation signatures in West Virginia red cedar. Global and
Planetary Change 84-85: 8-13.
De, A. K., 2002.Environ Chem., (4th
Edn.). New Delhi, India: New Age International Publishers
(232).
Dey, B., and O. S. R. U. Bhanukumar, 1983.The Himalayan winter snow cover and summer
monsoon rainfall over India. J. Geophys. Res., 88: 5471-5474.
Douville, H., and J. F. Royer, 1996. Sensitivity of Asian summer monsoon to an anomalous
Eurasian snow cover within the Metro-France GCM. Climate Dynamics 12: 449-466.
Dubey, N., 2003. A comparative status of drinking water of Bhopal city filteration plants and
ground water with special reference to heavy metals and organo chemical. PhD thesis,
Barkatullah University Bhopal.
Esper, J., 1994. Versucheiner Klimarekonstruktionmit Halfe der Dendrochronolgie am Beispiel
des Hunza- Tales in Karakorum. Unpubl. Master thesis, University of Bonn, 128pp.
Esper, J., A. Bosshard, F. H. Schweingruber and M. Wininger, 1995. Tree rings from the upper
timberline in Karakorum as climatic indicators for the last 1000 years.
Dendrochronologia 13: 79-88.
Esper. J., 2000. Long term tree variations in Juniperus at the upper timber line in the Karakorum
(Pakistan).The Holocene 10(2): 253-260.
Esper. J., Fritz. H and M. Winiger, 2002. 1300 years of climatic history for Western Central Asia
inferred from tree rings. The Holocene 12(3): 267-277.
Esper. J., K. Treydte, H. Gartner and B. Neuwirth, 2001. A tree ring reconstruction of climatic
extreme years since 1427AD for Western Central Asia. Palaeobotanist 50 (2001): 141-
152.
FAO., 2003. Review of World Water Resources by country water reports, Rome, Italy, 110.
Farooq. M. A., 2008. The study of groundwater pollution resulting from storm water infilteration
in Urban areas of Karachi. PhD dissertation. Federal Urdu University of Arts, Science
and Technology.
Farooq. M. A., 2012. Surface water quality assessment and apportionment of pollution sources of
lower Indus Basin from Kashmor to Keti-Bander using multivariate statistical technique-
144
A case study. Research project, Report submitted to Higher Education Commission,
Pakistan.
Fenwick. P., 2003. Reconstruction of past climates using Pink Pine tree ring chronologies. PhD
dissertation. Lincoln University, New Zealand.
Fowler, A., 1988. Climate reconstruction from tree rings. Weather and Climate, 8: 33-45.
Fritts, H. C., 1971. Dendroclimatology and dendroecology. Quaternary Research 1(4): 419-49.
Fritts, H. C., 1974. Relationships of Ring widths in arid sites conifers to variations in monthly
temperature and precipitation. Ecol. Mons., 44(4): 411-440.
Fritts, H. C., 1976. Tree Ring and Climate. London: Academic Press.
Grissino-Mayer, H. D., 2001. FHX2 – Software for analyzing temporal and spatial patterns in
fire regimes from tree rings. Tree Ring Research 57: 115-124.
Grissino-Mayer, H. D., and H. C. Fritts, 1997. The International Tree-Ring Data Bank: an
enhanced global database serving the global scientific community. The Holocene. 7: 235-238.
Grissino-Mayer, H. D., and H. C. Fritts, 1997. The International Tree-Ring Data Bank: an
enhanced global database serving the global scientific community. The Holocene. 7: 235-238.
Hayal, B., and B. Hiilya, 2009. Detection of seasonal variations in surface water quality using
discriminate anlaysis. Environ. Monit Assess., (DOI 10.1007/s10061-009-0772-5).
Heizer, R. F., 1956. The first dendrochronologist. Amer. Antiq., 22, 186-188.
Hemond, H., and E. Fechner-Levy, 2000. Chemical Fate and Transport in the Environment, 2nd
Edition. San Diega, California: Elsevier Science.
Holmes, R. L., 1992. Program ARSTAN (Version B-1992). Laboratory of Tree Ring Research,
University of Arizona, Tucson, USA.
Holmes, R. L., 1994. Dendrochronology Program Library- User manual, Laboratory of Tree-
Ring Research University, Tucson, Arizona USA.
Holmes, R. L., R. K. Adams, H. C. Fritts, 1986. Tree ring chornologies of western North
America, California, eastern Oregan and northern Great Basin.Chronology Series IV,
University of Arizona, Tucson, AZ, USA.
Hughes, M. K., 1992. Dendroclimatic evidence from the western Himalaya. In Bradley RS,
Jones PD (eds). Routledge Press: London; Climate Science Since A.D. 1500, 415-431.
Hughes, M. K., 1994. Lessons from past global change. Journal of Korea Forest Energy 14: 23-
34.
Hughes, M. K., and A. C. Davis, 1987. Dendroclimatology in Kashmir using tree ring widths and
densities in sub alpine conifers. In: L. Kariukstis, Z. Bednarz and E. Feliksik (eds.).
Methods of Dendrochronology. East/West approaches: 163-175. Warsaw, International
Institute of Applied Systems Analysis/ Polish Academy of Sciences.
Ihsanullah., K. Misal, N. Tariq, Khattak and A. Sattar, 1999. Determination of different
contaminants in Selective drinking water samples. The Nus., 36: 91-97.
145
Islamuddin., 2010. Hydro-chemical investigations of high altitude alpine lakes of Gilgit and
Ghizer districts, Northern Areas of Pakistan. M. Phil dissertation University of Peshawar,
Pakistan.
Jayaraman, P. R., T. Devi and T. Nayar, 2003. Water quality studies on Karamana River
Thiruvananthapuram District South Karela, India. Pollution Research.22: 89-100.
Jones. P. D., R. S. Bradley and J. Jouzal., editors, 1996. Climate variations and forcing
mechanism of the last 2000 years. Berlin springer.
Khan, A. R., A. Haq, W. A. Khan, M. Akif, M. Khan and M. Riaz, 1999. Quality characteristics
of potable water of Mardan city (Pakistan) and surrounding areas. J. Chem. Soc. Pak., 21:
106.
Khan, N., 2011. Vegetation ecology and dendrochronology of Chitral. PhD dissertation.
Laboratory of Dendrochronology and plant ecology of Pakistan, Department of Botany
Federal Urdu University of Arts, Science and Technology, Karachi, Pakistan.
Khan, N., M. Ahmed and M. Wahab. 2008. Dendroclimatic investigation of Picea smithiana
Boiss., from Afghanistan. Pak. J. Bot., 40(3): 1063-1070.
Khan. A. R., M. Akif, S. Wadud and K. Khan, 1999. Pollution studies of Kabul River and its
tributeries for the assessment of organic strength and Feacal coliform. J. of Chem. Soc.
Pak., 21:111-113.
Lashari. K. H., A. L. Korai, G. A. Sahoto and T. G. Kazi, 2009. Limnological studies of
Keenjhar Lake, District Thatta, Sindh, Pakistan. Pak. J. Ana. Env. Chem., 10: 39-47.
Li, C., and M. Yanai, 1996. The onset and inter-annual variability of the Asian summer monsoon
in relation to land sea constrast. Journal of Climate 9: 358-375.
Liang, E., X. Shao and N. Qin, 2008. Tree ring based summer temperature reconstruction for the
source region of the Yangtze River on the Tibetan Plateau. Science Direct 61: 313-320.
Luckman, B., R. Villalba, 2001. Assessing the synchroneity of glacier fluctuations in the western
cordillera of the Americas during the last millennium. In: V. Markgraf, ed.,
Interhemispheric Climate Linkages. Academic Press, New York: 119-140.
Lushuang, G., C. Zhang, X. Zhao and K. V. Gadow, 2010. Gender-related climatic response of
radial growth in Dioecious Fraxinus mandshurica trees. Tree Ring Research 66(2): 105-
112.
Mann, M. E., Z. Zhang, S. Rutherford, R. Bradley, M. Hughes, D. Shindell, C. Ammann, G.
Faluvegi and F. Ni, 2009. Global Signatures and Dynamical Origins of the Little Ice Age
and Medieval Climate Anomaly. Science 27: 1256-1260.
Mann, M., R. Bradley, & M. Hughes, 1998. Global-scale temperature patterns and climate
forcing over the past six centuries. Nature 392: 779-787.
146
May, A., E. Mutasem, D. Mark, 2006. Factor influencing development of management strategies
for the Abou Ali River in Lebanon. Sci. Total Environ., 362:15-30.
McCarrison, R., M. B. Captain I.M.S, 1906. Observation on endemic goiter in Chitral and Gilgit
Valleys, Agency Surgeon Gilgit Kashmir.
Muhammad, N., 2010. The express Tribune October 18th
2010.
Nanan, D., F. White, I. Azam, H. Afsar and S. Hozhabri, 2003. Evaluation of water, sanitation
and hygiene education intervention on diarrhea in northern Pakistan. Bulletin of World
Health Organization 81:3.
NASA (National Aeronautics and Space Administration) retrieved 2011-02-28.
Orvis, K. H., and H. D. Grissino-Mayer, 2002. Standardizing the reporting of abrasive papers
used to surface tree ring samples. Tree Ring Research 58: 47-50.
Overpeck, J., D. Anderson, S. Trumborte and W. Prell, 1996. The south west Indian monsoon
over the last 18000 years. Climate Dynamics 12: 213-225.
Owen, L.A. 2009. Latest Pleistocene and Holocene glacier fluctuations in the Himalaya and
Tibet. Quaternary Science Reviews 28: 2150–2164.
Pant, G.B, K. R. Kumar, H. P. Borgaonkar, N. Okada, T. Fujiwara and K. Yamashita, 2000.
Climatic Response of Cedrus deodara tree-ring parameters from two sites in the western
Himalaya. Canadian Journal of Forest Research 30(7): 1127-1135.
Parashar, C., N. Verma, S. Dixit and R. Shrivastava, 2008. Multivariate analysis drinking water
quality parameters in Bhopal, India. Envion Monit Assess.,140: 119-122.
Participatory Adoptive Management Plan Handrap Lake, Ghizer district Northern Areas,
Pakistan. WWF for Nature – Pakistan, Northern Areas.
Pejman, A., G. Bindani, A. Karbassi, N. Mehrdadi and B. Esmail, 2009. Evaluation of spatial
and seasonal variation in surface water quality using multivariate statistical techniques. J.
Env. Sci Tech., 6: 467-476.
Pirzada, T., M. M. A. Talpur, A.D. Rajper, M. R. Shah and S. Naseem, 2011. Quality
Assessment of Potable Water of Khairpur city. J.Chem. Soc. Pak., 33: 604-611.
Praus, P., 2005. Water quality assessment using SVD- based principal component analysis of
hydrological data. Water S. A., 31: 417-422.
Richman, MB., 1986. Rotation of Principal Components. Journal of Climate 6: 293-335.
Sano, M., B. M. Buckley and T. Sweda, 2009. Tree-ring based hydroclimate reconstruction over
northern Vietnam from Fokienia hodginsir eighteenth century mega-droughts and
tropical Pacific influence. Climate Dynamics 33: 331-340.
147
Sarwar. S., F. Ahmed and J. Khan, 2007. Assessment of the quality of Jhelum River water for
irrigation and drinking at District Muzaffarabad Azad Kashmir. Sarhad. J. Agric., 23:
1041-1046.
Schweingruber, F. H., L. Kairiukstis and S. Shiyatov, 1990. Sample selection. In E. R. Cook and
L. A. Kairiukstis (Eds.), Methods of dendrochronology: applications in the environmental
sciences (pp. 23-25). Dordrecht: Kluwer Academic Publishers.
Seckler D., U. Amerasinghe, D. Molden, R. Desilva and R. Barker (1998). World water demand
and supply 1990-2025: Scenerios and issues, International Water Management Institute,
Colombo, Sri Lanka, 40.
Shah, T., and Deb Roy, 2002. Intensive use of ground water in India, International Water
Management Institute, IWMI-TATA Water Policy Program, Anand, India, 16.
Shar, A. H., Y. F. Kazi, M. Zardari and I. H. Soomro, 2009. Bacteriological Quality of Drinking
water of Sukkur City. Pak. J. Med. Res., 48:414-420.
Shar, A. H., Y. F. Kazi, N. A. Kanhar, I. H. Soomro, S. M. Zia and P. B. Gumro, 2010. Drinking
water quality in Rohri city, Sindh, Pakistan. African J. Biotechnol., 42: 7102-7107.
Shi, F., F. Yang, T. Nakatsuka, M. Sano and Pages Asia 2k Project member, 2012. Multi-
centennial temperature variability in Asia over the past 1000 years. (In press).
Shrestha, S., and F. Kazama, 2007. Assessment of surface water quality using multivariate
statistical techniques: A case study of Fuji River Basin, Japan. Environmental Modeling &
Software, 22: 464-475.
Singh, J. and R.R. Yadav, 2007.Dendroclimatic potential of millennium-long ring width
chronology of Pinus gerardiana from Himachal Pradesh, India. Current Science 93(6):
833-837.
Singh, J., R. R. Yadav and M. Wilmking, 2009. A 694-year tree ring based rainfall
reconstruction from Himachal Pardesh, India. Climate Dynamics 33(7): 1149-1158.
Singh. J., R.R. Yadav, B. Dubey and R. Chaturvedi, 2004. Millenium long tree ring width
chronology of Himalayan cedar from Garhwal Himalaya and its potential in climate
change studies. Current Science 86(4): 590-594.
Speer, J. H., 2010. Fundamental of Tree Ring Research. The University of Arizona Press
Tucson.
Standard methods for the examination of water and waste water treatment, American Public
Health Association (APHA), American Water Works Assocation (AWWA), Water
Environmental Federation, 2003.
Stokes, M. A. and T. L. Smiley, 1968. An introduction to Tree Ring dating. The University of
Chicago Press, Chicago Illinois (Reprinted by the University of Arizona Press, Tucson.
148
Tabachnick, B. G., and L. S. Fidell, 1983. Using multivariate statistics, Harper and Row, New
York, pp. 509.
Tranquillini, W., 1964. The physiology of plants at high altitude. Annual review of plant
physiology 15: 345-362.
Treydte, K. S., G. H. Schlesr, G. Helle, D. C. Frank, M. Winige, G. H. Hang and J. Esper, 2006.
The twentieth century was the wettest period in northern Pakistan over the past
millennium. Nature 440(27): 1179-1182.
Tyagi, N. K., 2003. 5 Managing saline and alkaline water for high productivity. Water
productivity in agriculture: Limits and opportunities for improvement. Central Soil
Salinity Research Institute, Haryana, India.
Vega, M., R. Pardo, E. Barrado and L. Deban, 1998. Assessment of seasonal and polluting
effects on the quality of river water by exploratory data analysis. Water Research. 32: 3581-
3592.
Wahab, M., 2011. Population dynamics and dendrochronological potential of Pine tree species
from District Dir. PhD dissertation. Department of Botany, Federal Urdu University of
Arts, Science and Technology, Karachi, Pakistan.
Welch, P. S., 1952. Limnological methods. New York: Mc-Graw Hill.
WHO‟s Guidelines for drinking water quality setup in Geneva, Switzerland, 1993.
Wigley, T. M. L., K. R. Briffa and P. D. Jones, 1984. On the average value of correlated time
series, with application in dendroclimatology and hydrometeorology. Journal of Climate
and Applied Meteorology 23: 210-213.
Wu, X., 1992. Dendroclimatic studies in China. In Climate Since A.D. 1500, Bradley RS, Jones
PD (eds). Routledge Press: London.
Wu, X., and X. Shao, 1995. Status and prospects of dendrochronological study in Tibetan
Plateau. Dendrochronologia 13: 89-98.
Xiong, L., N. Okada and T. Fujiwara, 2000. The dendrochronological Potential of ten species in
three Georges Reservoir Region of China. IAWA Journal 21(2): 181-196.
Yadav, R. R., 2011. Long term hydroclimatic variability in monsoon shadow zone of western
Himalaya, India. Climate Dynamics 36: 1453-1462.
Yadav, R. R., and J. Singh, 2002. Tree ring analysis of Taxus buccata from the Western
Himalaya, India, and its Dendroclimatic potential. Tree Ring Research58 (1/2): 23-29.
Yadav, R. R., and J. Singh, 2002. Tree-ring based spring temperature patterns over the past four
centuries in western Himalaya. Quarternary Research 57: 299-305.
Yadav, R. R., and W. K. Park, 2000. Precipitation reconstruction using ring-width chronology of
Himalayan Cedar from Western Himalayan, India: Prelimanary results. Proc. Indian
Acad, Sci., 109(3): 339-345.
149
Yadav, R. R., W. K. Park and A. Bhattacharya, 1999. Spring-temperature variations in Western
Himalaya, India as reconstructed from tree ring: AD 1390-1987. The Holocene 9 (1):85-
90.
Yadav, R. R., W. K. Park and A. Bhattacharyya, 1997. Dendroclimatic reconstruction of April-
May temperature fluctuations in the western Himalaya of India since A.D. 1698.
Quaternary Research 48: 187-191.
Yan, T., and M. J. Sadowsky, 2007. Determining sources of fecal bacteria in waterways.
Environ. Monit. Assess., 129: 97-106.
Yu, L., S. Jiangfeng, V. Shishov, E. Vaganov, Y. Yinke, C. Qiufang, S. Junyan, W. Lei and I.
Djanseitov, 2004. Reconstruction of May-July precipitation in the north Helan mountain,
Inner Mangolia since AD 1726 from tree ring late-wood widths. Chinese Science Bulletin
49(4): 405-409.
Zafar. M. U., M. Ahmed, M. Akbar and A. Hussain, 2010. Standard Tree Ring Chronologies of
Picea smithiana from two new sites of Northern Area Pakistan. World Applied Sciences
Journal 11(12):1531-1536.
Zhang, Q. B., G. Cheng, T. Yao, X. Kang and J. Huang, 2003. A 2326 year tree ring record of
climatic variability on the northeastern Qinghai-Tibetean Plateau. Geophysical Research
Letters 30(14): 1739.
Zubair, A., H. N. Abbasi and A. Hussain, 2009. Hydro chemical interpretation of storm water
impact on Ground water using Factor Analysis. J. Env. P. Soc., 3: 117-123.