Upload
geographical-analysis-urban-modeling-spatial-statistics
View
538
Download
0
Tags:
Embed Size (px)
DESCRIPTION
"Geographical Analysis of Foreign Immigration and Spatial Patterns in Urban Areas. Density Estimation and Spatial Segregation" Third International Workshop on "Geographical Analysis, Urban Modeling, Spatial Statistics"
Citation preview
Geographical Analysis of Foreign Immigration and Spatial Patterns in Urban Areas.
Density Estimation and Spatial Segregation
Dr. Giuseppe Borruso
Department of Geographical and Historical SciencesUniversity of TriesteEmail. [email protected]. +39 040 558 7008Fax. +39 040 558 7009 / 7005
GeogAnMod ‘08The International Conference on Computer Science and its Applications
ICCSA 2008Perugia 30 June – 03 July 2008
Abstract
The paper is focused on the analysis of immigrant population with particular reference to their spatial distribution and the tendency to cluster in some parts of a city, with the risk of generating ethnic enclaves or ghettoes.
Methods used in the past to measure segregation and other characteristics of immigrants have long been aspatial, therefore not considering relationships between people within a city.
In this paper the attention is dedicated to methods to analyse the immigrant residential distribution spatially, with particular reference to density-based method.
The analysis is focused on the Municipality of Trieste (Italy) as a case study to test different methods for the analysis of immigration, and particularly to compare traditional indices, as Location Quotients and the Index of Segregation, to different, spatial ones, both based on Kernel Density Estimation functions, as the S index and the first version of an Index of Diversity.
Topics
Qualitative and quantitative methods for the analysis of immigrants at urban level;
Measures of segregation; Spatial indices of dissimilarity; The spatial distribution of migrant population in Trieste; Conclusions and discussion.
Qualitative and quantitative methods for the analysis of immigrants at urban level
The analysis on migrations can rely on a mixed combination of methods and tools, both quantitative and qualitative ones
diffusion of spatial analytical instruments and information systems
scholars involved in migration research should therefore rely also on qualitative methods in order to integrate their studies, with the difficult task of interpreting correctly what is happening over space
Researchers have focused their attention on different indicators in order to examine the characters of the spatial distribution of migrant groups, particularly in order to highlight the trends towards concentration rather than dispersion or homogeneity, or, still, the preferences for central rather than peripheral areas.
summary indices are useful to portray the level of segregation of a region and for comparing the results obtained for different regions, but they say little about some spatial aspects of segregation
Varies between 0 and 100 (or 0-1) Represent major or minor
dispersion or concetration of an ethnic group
with xi = # of residents of a national group in a sub-area i;
X number of residents in the study region (municipality);
yi the population in area i; Y the population of the study
region.
xi = number of residents of a national group in sub-area i;
X = number of residents in the study region (municipality),
yi = foreign population in area i Y = the overall foreign population QL = 1 => the group in the sub-
area present same characterisics of the distribution in the overall study region considered;
QL > 1 the group is over-represented in the sub-area
QL < 1 the group is under-represented in the sub-area
Y
y
X
xD ii
2
1
1005,0 Yy
XxD ii Y
Xy
xLQi
i
Measures of segregationSegregation
Index(Duncan & Duncan, 1955)
Location Quotient (Cristaldi,
2002)
Segregation index
Generally a-spatial The zoning system of the study
region affect the final result Higher disaggregation of data
=> higher value of segregation; Higher aggregation of data =>
lower value of segregaion Figures: segregation index on
foreign nationals from address point data aggregated to:
a) census blocks; b) urban districts
a)
b)
Segregation Index for selected ethnic groups (census blocks)
71,82
83,86
74,49
91,50
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
80,00
90,00
100,00
SI Albania SI China SI Romania SI Senegal
Seg
reg
atio
n In
dex
Segregation Index for selected ethnic groups (urban districts)
4,70
9,90
18,27
30,55
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
80,00
90,00
100,00
SI Albania SI China SI Romania SI Senegal
Seg
reg
atio
n In
dex
Location Quotient
Spatial index Represent ‘specialization’ or
representativeness of a phenomenon in a given place (area) with respect to the overall study region
Figure: foreign residents in Trieste census blocks > 5 % (mean municipal data)
The function creates a density surface from a distribution of points (events) in space, providing an estimate of events withn its searching function according to their distance from the point where the estimate is computed.
= estimate of intensity of the spatial distribution of events, measured in point s;
si = the ith event;k( ) = kernel function = bandwidth. Varying the bandwidth it is possible to
obtain smoother or sharper surfaces and analyze the phenomena at different scales.
Diversity Index (IDiv) for area i consider the number of countries in each sub-area (census block). The value is multiplied by the % of residents in the census block
Ni = # of countries in sub-area i; yi foreign population in sub-area
i and xi the overall population in sub-area i.
The index is computed by means of a KDE (from values assigned tocensus blocks’ centoids) in order to transform it into a density surface and visualize its variation in space.
n
i
issks
12
1
s
100*
ii
ii xyNIDiv
Spatial indices of dissimilarityKernel Density Estimation
Diversity Index
Segregation index S (O’ Sullivan e Wong, 2007) spatial modification of the Duncan’s index D comparing, at a very local level, the space deriving from the intersection of the extents
occupied by two sub-groups of an overall population and the total extent of the union of such areas
involves the computation of probability density functions by means of KDE for the different population sub-groups of interest.
Each reference cell i is therefore assigned a probability value for each subgroup, and for each of the subgroups the probability value in that cell contributes to the integration to unity.
i yx
i yx
ii
ii
pp
ppS
),max(
),min(1
For each cell i minimum and maximum values are computed for the true probability of the two subgroups, pxi and pyi, these are summed for all the i cells, obtaining minimum and maximum values under the two surfaces, and their ratio is
subtracted from unity to produce index S. The index S obtained is aspatial as well, as it can be obtained for a study region, but the intermediate values, as the
differences in maximum and minimum values, can be mapped, giving a view of the contribution of each cell to the overall segregation, with lower values indicating areas with some degree of ethnic mixing.
Different bandwidth values produce a decay of the index as bandwidth increases, thus reducing the segregation index overall the study region and still the differences of this behaviour in different regions or for different groups can be
analyzed to explore dynamics proper of territory or group
The spatial distribution of migrant population
in Trieste
The data and the study area Segregation Index Location Quotients Kernel Density Estimation The S index of segregation The Index of Diversity
Data
Spatial component: Areas (zoning systems) Points (addresses; fieldwork; areas centroids)
Attributes Characteristics of migrants (Anagraphical data: sex, age, gender,
country of origin, residence address; working activities.) Key issues related to data:
Availability Format Level of aggregation (results affected) Multidimensional data => difficult to analyze without suitable
instruments and a multicisiplinary approach (qualitative and quantitative methods integrated in the analysis).
Zoning systems in a Municipality
New districts Old districts Census blocks
The study region
The municipality of Trieste
Census blocks Residents’ address points
Indices of spatial distribution of population
Traditional indices -> elaboration/ visualization by GIS Segregation index Location quotient
New indices -> inmplemented thanks to possibilities allowe by GIS enviroment
Density estimators Diversity indices
Indices of spatial distribution of population
Ethnic group of more recent immigration in Trieste (data from Statistical Office, Municipality of Trieste, 2005)
Albania China Romania Senegal
Indices presented: Segregation Index D; Location Quotient LQ ; Kernel Density Estimation Segregation Index S (after O’ Sullivan and Wong) Index of Diversity
Spatial units considered: Urban districts Census blocks Address points
Segregation index census blocks a); urban districts b)
Segregation Index for selected ethnic groups (census blocks)
71,82
83,86
74,49
91,50
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
80,00
90,00
100,00
SI Albania SI China SI Romania SI Senegal
Seg
reg
atio
n In
dex
Segregation Index for selected ethnic groups (urban districts)
4,70
9,90
18,27
30,55
0,00
10,00
20,00
30,00
40,00
50,00
60,00
70,00
80,00
90,00
100,00
SI Albania SI China SI Romania SI Senegal
Seg
reg
atio
n In
dex
a) b)
a) Albania b) China
c) Romania d) SenegalLocation Quotient
for selected ethinc groups
Kernel Density Estimator
Standardize value(comparison with other ethnic groups)
Areas with higher population density
Bandwidth = 300m Grid cell = 50m
a) Albania b) China
c) Romania d) SenegalKDE
On selected ethnic groups
Segregation index S (O’ Sullivan e Wong, 2007)
Kernel Bandwidth (m) Albania Cina Romania Senegal
150 75,66 86,83 77,43 81,38
300 64,14 80,31 62,53 81,46
450 58,31 77,12 54,83 76,60
600 54,56 74,39 49,46 72,94
900 50,15 73,13 43,84 73,13
30,00
35,00
40,00
45,00
50,00
55,00
60,00
65,00
70,00
75,00
80,00
85,00
90,00
95,00
100,00
150 300 450 600 900
Kernel Bandwidth (m)
Seg
reg
atio
n I
nd
ex
Albania China Romania Senegal
KDE Max-min (bandwidth = 300m)
a) Albania b) China
c) Romania d) Senegal
Diversity index (IDiv)
Area with higher density
(# of countries compared to % of foreign population)
bandwidth = 177m(nearest neighbour k=2)
Grid cell = 50m
Diversity index (IDiv)
Area with higher density
(# of countries compared to % of foreign population)
bandwidth = 281m(nearest neighbour k=5)
Grid cell = 50m
Conclusions Indices for measuring segregation or diversity in the distribution of migrant groups at
urban level Considering the spatial aspects of such indices and the need to examine more in depth
the articulated structure and characteristics of the population. Problems still need to be addressed: Availability of disaggregated data, however, if the zoning system produces sufficiently
small areas some analytical methods reduce such problem. The choice of the bandwidth or distances of observation, although efforts in this
direction are under exam Other issues concern the multi-group analysis, therefore not limiting this to two
subgroups but to the overall variety of countries represented in a given study region. Qualitative, multivariate attribute of population data should be considered. Need to explore the opportunity to develop and implement entropy-based diversity
indices, as well as to examining the relations between economic activities, residential locations and segregation as emerging migration issues to analyse.
Methods based on density provide a good starting point for more in depth and local analysis by the researchers, that can focus their attention over a micro scale of analysis, going further than the administrative divisions of space and reducing the minimum distance of observation to examine locally the dynamics at urban scale.