Spatial Data Infrastructures

in Support to Land Cover Monitoring Activities

Brice Mora, PhD

GOFC-GOLD Land Cover Project Office

Spatial Data infrastructures Course (MGI), January 2015

Outline

• What is GOFC-GOLD?

• A Web-portal for Global Land Cover Reference Datasets

- data quality standards

- good accuracy assessment practices

- data distribution licensing

- introduction to relational database concepts

• Data Stream Management in the Context of Forest Monitoring

- forest change detection with remote sensing

- smartphone applications

- community-based monitoring

What is GOFC-GOLD?

• Developed in 1997, originally under the Committee on Earth Observation Satellites (CEOS):

- To test the concept of an Integrated Global Observing System (IGOS)

- To improve use of Earth Observation data to address major problems of global concern

- To improve coordination of national programs

- To improve co-operation between providers and users of Earth Observation data for regional and global applications

• Has become one of the Panels of the Global Terrestrial Observing System GTOS (FAO GTOS Secretariat)

• Sponsors: FAO, WMO, UNEP, UNESCO, ICSU, EC-JRC, ESA, NASA, USGS

Background to GOFC-GOLD

1. RedLatiF – Latin America

2. SAFNET – Southern. Africa

3. Miombo – Southern Africa

4. OSFAC – Central Africa

5. WARN – West Africa

6. SEARRIN – S.E. Asia

7. NERIN – Northern Eurasia

8. CARIN – Central Asia

9. SCERIN – S.Central Europe.

10. SARIN – South Asia

11. BARIN – Baltic-Arctic

GTOS

User

Outreach

Science and

Technical Board

GOFC-GOLD

Executive Committee Project Office

Implementation

Teams

- Land Cover

- Fire

Partnerships e.g. UNISDR WFAG;

CEOS WGCV

Global Strategies &

Frameworks e.g. GEOSS, GCOS

IP

Regional

Networks

Working Groups

- REDD

- Biomass

A Web-portal for Global Land Cover

Reference Datasets

Data portal: Background

• Observation of land cover at global scale useful for many scientific and managerial applications

• Several global land cover (GLC) maps produced using remote sensing data

• Several independently validated GLC datasets which generation required significant efforts to analyse a large number of satellite images and interpret land cover type

• Reference datasets not always easily accessible to the land cover community and not used to full potential despite scarcity of such datasets

Data portal: objectives

The GOFC-GOLD Land Cover Office web portal aims to: • Provide most appropriate databases based on internal quality

criteria and a consolidation • Allow open and easy access to available datasets to the land

cover community (while keeping some data for independent assessments)

• Foster use of recommended practices for land cover validation in the GLC mapping community

• Each dataset provided along with detailed information and recommendations for an appropriate use

• Direct users to most appropriate dataset(s) according to specific needs

• Basis for more operational land cover validation activities of GOFC-GOLD

GOFC-GOLD reference data web portal

http://www.gofcgold.wur.nl/

GOFC-GOLD reference data web portal

https://cartodb.com/

Promoting Standards:

Land Cover Classification system

• Developed by FAO and UNEP as comprehensive and standardized classification system for mapping purposes.

• Independent from mapping scale

• Allows dynamic creation of classes using combination of LC diagnostic attributes called classifiers.

• Last version of the LCCS: LC Metadata Language (LCML – LCCS v.3) proposed as standard by the International Organization for Standardization (ISO): ISO 19144-1.

• Complementary specifications under development (reference WI 19144-2).

• Herold, M., Hubald, R., & Di Gregorio, A. (2008). Translating and evaluating the land cover legends using the UN Land Cover Classification System (LCCS). Network (p. 189). Jena, Germany. http://nofc.cfs.nrcan.gc.ca/gofc-gold/Report%20Series/GOLD_43.pdf

Available datasets

Name Sampling design Sample size Sample

unit/size

Source or reference

data

Legend

protocol Legend Rreference

GLC 2000 2 stage stratified

cluster sampling

1265

253 PSU

5SSU in each

PSU

3by3 pixels

Landsat 2000, aerial

photographs, thematic

maps, NDVI profile

If many LC types

are there, 2

main covers

were recorded

>80%

LCCS 22

class

Mayaux et al

2006

GlobCover Stratified random

sampling

4258

3167 certain

5by5 pixel SPOT VGT-NDVI profile,

Google Earth,

>75%

dominance,

Record more

classes if there is

LCCS 22

class

Defourny et al

2009

STEP stratified 1780

Landsat, high and low

resolution images

(Google Earth)

Google Earth

IGBP 17

classes +

other classes

Friedl et al.,

2000

Sulla-Menashe

et al., 2011

VIIRS

stratified random

sampling

500 5by5 km

blocks VHSR (<2-m)

manual

interpretation

aided by Google

Earth and MODIS

time-series

IGBP legend,

and LCCS in

the future

Olofsson et al.,

2012

Stehman et al.

2012

GLCNMO

International Steering

Committee for Global Mapping

600 Training/Vali

dation Worldwide

Tateishi, et al., 2011

GLCNMO

International Steering

Committee for Global

Mapping

Global urban

ground truth data

University of Tokyo

3734 Validation Worldwide

Global urban ground truth

data

University of Tokyo

Upcoming datasets and future updates

Dataset name

Provider Sample size Suitable for Coverage Reference

VHSR Boston U. 500 Validation Worldwide Olofsson, et al., 2012

GlobCover 2009

ESA/UCL Validation Worldwide Bontemps, et al., 2011

Landsat GLC map

China 38664 Validation Worldwide Gong, et al., 2013

LC CCI ESA 13000 Validation Worldwide Achard et al., 2011

NELDA dataset

NELDA 11 Validation Northern Eurasia

Clark & Aide,

2011b

• Regular updates coming from: VIIRS, STEP

Distribution License

Datasets distributed under Creative Commons License: Attribution-Non Commercial-No Derivatives

http://creativecommons.org/

Providing guidance: Metadata

Difficult or impossible to use data without proper metadata

See ISO 19115 and ISO 19139 standards

Providing Guidance: Suitability of GLCR datasets for

different users Metadata of GLCR

datasets User requirements for

reference datasets

User requirement criteria

Evaluation of GLCR datasets for each user requirement criteria

Combining criteria performance using multi-criteria approach

Criteria Probability

sampling

Class

representatio

n

Easily combined

and augmented

sampling

scheme

Classification

scheme and

thematic details

of the legend

Hierarchical

classifiers

provided

Temporal

coverage

Stable

multi- date

sample

Spatial

resolution Verified

Interpreta

tion

confidenc

e

recorded

IGBP-DIS ++ - + +/- ++ - -- ++ + ++

GlobCov5 ++ +/- + ++ ++ + - ++ ++ ++

GlobCov9 ++ + + ++ ++ + - ++ - ++

GLC2000 ++ +/- + ++ ++ + -- - ++ -

GLCNMO-

val ++ +/- + ++ - + -- + - -

GLCNMO-tr -- +/- -- +/- - + -- +/- +/- -

MODIS-tr -- +/- -- +/- ++ + - +/- + -

FAO-FRA ++ ++ +/- +/- - + ++ ++ + -

LC-CCI ++ ++ +/- ++ ++ ++ ++ ++ + ++

GOFC-GOLD ++ + ++ +/- ++ + - ++ + ++

GEO-WIKI -- + -- + +/- + - + - ++

VIEW-IT +/- + - - - + - -- + ++

(++ Highly suitable, + Very suitable, +/- Moderately suitable, - Marginally suitable, -- Not suitable )

Criteria performance of GLCR datasets for each user requirement criteria of Climate modelling users

Tsendbazar et al., 2013

Suitability of GLCR

datasets for

different users

The LC-CCI, GOFC-GOLD, FAO-

FRA and Geo-Wiki datasets were

generally more suitable for re-

use than the other datasets.

The analysed datasets are

generally more suitable for

agricultural monitoring and

improving GLC maps.

Climate modelling community

and forest change analysis

require stable multi-date sample

which couldn’t be met many of

the GLCR datasets.

Data Storage, Retrieval, and Visualisation

Web-GIS application

• For data storage, retrieval,

and visualization

Relational Model in PostgreSQL

Following BOYCE-CODD Normal Form underlined: primary/secondary key(s), #: foreign key, table names in capital letters INSTITUTION (Iinstitution, Iname)

TITLE (Ttitle, Tname)

PERSON (Pperson, Pinstitution#, Ptitle#, Pfirstname, Plastname, Pemail, Pprofile#, Plogin, Ppwd)

DATASET (DAdataset, DAname)

ACCESS (Aaccess, Aname)

DATA (Ddata, Daccesstype#, Duploaddate, Dthemayear#, Dcountry#, Ddataset#, Ddatalink, Dmetadatalink)

COUNTRY (Ccountry, Ccharid, Cname)

OWN (Operson#, Odata#)

BVHRIM (BVImage, BVIdata#, BVIsite#, BVItype#)

SITE (Ssite, Slatcentroid, Slongcentroid, Secoclimate#)

ECOCLIM (Ecoclim, Ename)

BVHRPROCES (BVPimage#, BVPimageori#, BVPperson#, BVPdatebegin, BVPdatend, BVPsoft#, BVPtrain#, BVPaccuracy, BVPcomments)

SOFTWARE (SOsoft, SOname) Theory and practice example

https://www.youtube.com/watch?v=hTFyG5o8-EA

Database: Forms to upload and retrieve

information

Form to enter new information on

persons, affiliations, sites, software

not referenced in database yet

Based on SQL (Structured Query Language)

=> Enables creation and manipulation of tables in a database

AND

=> Query the tables to retrieve information

SQL query example:

https://www.youtube.com/watch?v=aZekk0udLYg

Geoserver

http://geoserver.org/

Needs from the user communities

Better guidance on how to use reference datasets - thematic applications (biodiversity, crop monitoring, ...)

- best land cover map accuracy assessment practices

Online data portal enabling upload of map products for accuracy assessment.

Independent reference dataset as benchmark for GLC maps

Data Stream Management in the Context of Forest Monitoring

Credits: Arun Pratihast, Post-doc researcher at WUR

Distribution of Aboveground Forest Biomass

Source: Avitabile et al. 2015

Forest Change Patterns 2000 – 2005

Source: FAO/FRA RSS, 2012

Climate Mitigation Mechanism: REDD+

United Nations Framework Convention on Climate Change – Cancun agreements on REDD+ (UNFCCC, 2010)

Following activities are included:

● Reducing emissions from deforestation

● Reducing emissions from forest degradation

● Conservation of forest carbon stocks

● Sustainable management of forest

● Enhancement of forest carbon stocks

REDD

+

Need for a Monitoring System:

Complementarity of Data Streams

The relative strength of contribution of each data stream to the REDD+ MRV objectives is indicated by shade (dark = strong; light = limited)

BFAST Spatial: in-Migration (Kafa, Ethiopia)

Image background: SPOT5 (Feb 2011,

2.5m)

Technical Setup for Forest Monitoring

Data collector : local expert

Systematic form design : decision based form design for Mobile

device

Interactive Forest Monitoring System

Web Resources

• Change detection and monitoring (BFAST):

http://bfast.r-forge.r-project.org/

https://github.com/dutri001/bfastSpatial

http://www.wageningenur.nl/en/Expertise-Services/Chair-

groups/Environmental-Sciences/Laboratory-of-Geoinformation-Science-and-

Remote-Sensing/Research/Integrated-land-

monitoring/Change_detection_and_monitoring.htm

• GOFC-GOLD REDD sourcebook:

http://www.gofcgold.wur.nl/redd

• Organizing community level forest surveys for REDD+: Manual for Community

Technicians

• http://redd.ciga.unam.mx/files/CommunityManual.pdf

brice.mora@wur.nl

Follow us:

Twitter: @gofcgold_lc Facebook: www.facebook.com/gofcgold.lc.po Newsletter: www.gofcgold.wur.nl