1. New Opportunities for Urban
Analysis
Identification of the urban area by
using population surface models
based on postcode information
2. Urban System Analysis
Generally, urban systems consist of:
• socio-economic functions
– people and what they do
• Built environment
Identifying urban areas and their growth (urban morphology),
therefore, might be through one or more of the following
ways:
• Identification of built form
• Identification of human activity pattern, for example:
residential densities
• Measures of dynamics
3. The disadvantages of RS in
Urban Analysis
RS has been used for a long time in urban analysis (Mesev,
1997). However, it is a rather indirect and inferential
method. Urban surfaces always provide very difficult
challenges for RS, i.e.:
• To determine land use providing
perspective upon the functioning of urban
settlements (Mesev, Batty, &Longley, 1996; Longley, Mesev, 2001),
• To measure change (Brouwer, Valenzuela, Calencia,
&Sijmons, 1990)
4. The New Opportunity in Urban
Analysis
• Code-point data
• Address-point data
The increasing availability of new disaggregate, frequently
updated data pertaining to socio-economic characteristics is
making possible new kinds of urban analysis and modelling
(M. Goodwin, D. Unwin, 2000):
5. Summary of data sources
Code-Point Address-Point
Date vendor
First data collection
Geography
Frequency of updates
Sampling unit
Average number of households
per unit
Data-type
Resolution
Content
Ordnance Survey (GB)
1997 (as Data-Point)
GB
6 months
Unit Postcode
About 12 (GB)
Attributed point
Most postcode centroids
precise to 1m
Total number of residential and
commercial mail delivery
points per postcode
Ordnance Survey (GB)
1996
GB
6 months
Mail delivery point
Should be 1 in the majority of
cases
Attributed point
Most points precise to 0.1m
Full address and National Grid
Reference of individual
delivery points.
Source: New Data and Approaches for Urban Analysis, R. Harris and P. Longley, 2000
6. Some Features of the Above New
socio-economic Data
• More positional precise— non-aggregate/or
low aggregation
• More frequently updated
• Providing a better prospective upon
reflecting human activity patterns more
directly
7. Two Cases
With these new available data, some urban analysis has been done by using
the method of population surface models.
1. Modelling residential densities in Bristol
area (R. Harris, P. Longley, 2000)
Address-point enables the identification of buildings
with considerable precision
8. 2. Defining and delineating the central
areas of towns (Thurstain-Goodwin, Unwin, 2000)
Code-point data makes the spatial resolution of
the result much higher than that achieved using
the ED.
Code-point data affords the possibility of much
greater precision in the statistical aggregations of
town centres (and indeed of any other
geographically identifiable object).
Both of these two researches support the view that identification of urban land area or town centre can be done faster and more
precisely based on the new data-point dataset through the methodology of population surface models.
9. Population Surface Model
• In principle, the PSM approach involves the
redistribution of data points into the cells of
a (raster) output grid. A user-defined search
window is positioned over each data point
in turn. The count associated with the
current point is then distributed into the
cells falling within the window, according
to weighting derived from a distance-decay
function.
10. The Distance-decay Function
put at its simplest, it is a model parameter that influences the weight given to cells close to the
current point within the search window thus controlling the form of the distribution.
Weighting
Distance
Central
Point j
Wij
dij
K
α
The weighting of each cell Wij on the basis of a variable width search window k
11. Benefits of Surface Approach
• Simple to visualise big datasets
– ease of communication
– identification of errors in raw data (Thurstain-Goodwin,
2001)
• Overlay for composite indicators
– Simple addition where attribute metric is the
same (Thurstain-Goodwin, 2001)
• i.e. can be used for overlays in GIS
12. The Problems of Population
Surface Modelling
Barr and Barnsley (1997) argue that the PSM approach has practical and philosophical
problems.
• In practical terms, the optimum size of the search window
must normally be determined experimentally and is not
allowed to vary throughout the study region.
– Use of adaptive kernels (though still dependent on original search
window size)
The original model also has other problems where the data point density is low, consisting of
individual isolated data point ( Martin, et.al, 2000 ).
• In such a situation, population is artificially dispersed in a
symmetrical or isotropic fashion around each data point
location although actually it should be displayed as a
linear or strongly clustered anisotropic form
13. My interests
(I like food and videos. My favourite TV program is The
Simpsons.)
Aim:
Therefore, bearing the RS and PSM issues in mind, my current interest is to experiment with
various types of point-pattern analysis of space-filling raster modelling such as the PSM,
to produce models of urban population density and urban boundary locations based on
the postcode data.
Objectives:
• To measure how good the procedure of identifying urban area by
using PSM is
• To measure how it compares with traditional RS-based methodologies
• To measure how sensitive are the results to the various modelling
parameters
• Improving the model
14. DATA
Current data available:
• Code-point data (Source: Ordnance Survey)
• Urban boundaries data (provided by Prof. John Shepherd,
created based on Ordnance Survey maps)
• Address-Point
16. What I am going to do next?
Rationale: Taken urban boundaries data file as a basic line to measure the
effect of urban area identified by using code-point data and
Population Surface Models.
1. ArcGIS software will be used.
2. Selecting study area---a specific city,based on some
criteria:
– It should be a manageable area
– It should have no policy constraint
– It should belong to fast/fair growing area.
1. Getting the corresponding data from urban
boundaries data and code-pint data
18. 4. Using PSM to identify urban area based on code-
point data
– the distance-decay function - and kernel width K
will be changed to measure how good the procedure
and how sensitive of the results will be.
5. Go to the the pub!
α
Hinweis der Redaktion
Regarding the analysis of socio-economic functions, most urban theories consider human activity patterns as the observable element in urban analysis. As a starting point for understanding human activity-based patterns, models of residential density – i.e. where people live – are important. Meanwhile, as we all are aware, the urban system is not a static system. The scale and function of urban area keeps changing. Repeated measurements of urban system across different time periods are required to generate insights into processes of urban change.
Example from Longley and Mesev (2001), some satellite sensors can provide spatial
information at a level of detail (1m-5m). However, it will not be possible unambiguously to
identify, say, small workshops amidst residential areas from an analysis of their spatial form
alone.
Forster(1985), lack of contextual clues for interpretation and geometrical mis-registration contribute to the difficulties of measuring change.
For example: reflect residential density directly and it has also different categories to distinguish residential and commercial data.
Both of these two researches support the view that identification of urban land area or town centre can be done faster and more precisely based on the new data-point dataset through the methodology of population surface model.
The far distance from cell I to central point j, the less weight will be given the the cell I.