jean-louis weber consultant european environment agency scientific committee
DESCRIPTION
The Natural Capital/Ecosystem Capital Accounting (ECA) project for Mauritius Production of the urban areas land cover layer from high resolution data on buildings, using smoothing ( Gaussian filter ) techniques & Land cover change account 2000 – 2010 / Urban sprawl. - PowerPoint PPT PresentationTRANSCRIPT
The Natural Capital/Ecosystem Capital Accounting (ECA) project for Mauritius
Production of the urban areas land cover layer from high resolution data on buildings, using smoothing (Gaussian filter) techniques
& Land cover change account 2000 – 2010 / Urban sprawlJean-Louis WEBER
ConsultantEuropean Environment Agency Scientific Committee
Honorary Professor, School of Geography, University of Nottingham
Introduction
• The land cover layers for urban areas have been produced using the geo-database of buildings of Statistics Mauritius. It includes data of 2010 and circa 2000.
• The processing consists in data rasterisation at 10 meters followed by smoothing (SAGA Gis Gaussian filter) in order to agglomerate buildings into « urban areas », thus assimilating small holes and streets. Deanse and dispersed urban areas (e,g, in the countryside) can be mapped.
• Accounts compare the stocks and change between two dates.
The buildings Shapefile
The buildings raster (tif) 10 meters x 10 meters
The buildings Shp and Raster 10 m
Smoothing (blurring) with SAGA Gis/ Grid Filters/ User Defined Filter
Input: raster 10 m, values 1 to 101Filter Matrix (for gaussian blur at 10 pixels radius or 100 m, using a kernel of 21 x 21 cells): here Kernel_21_10
Sequence of treatments with SAGA GIS:
Input: shapefile, scale circa 1/5000 or finer
Raster (tif) at 10 meters
Smoothed (Gaussian blur) raster, radius of 100 meters (kernel = 21)
The buildings raster smoothed at 100m (values in the neighbourhood)
Building raster, 10 m and smoothed at 100m (values in the neighbourhood)
Building Shp and smoothed tif (values in the neighbourhood)
Agglomeration/generalisation: cells > 20% of the smoothed value
NB: cells are of 10 x 10 meters
Agglomeration/generalisation: shp and cells > 25% of the smoothed value
NB: cells are of 10 x 10 meters – here, the threshold captures dispersed urban
Agglomeration/generalisation: shp and cells > 50% of the smoothed value
NB: cells are of 10 x 10 meters – here, the threshold eliminates dispersed urban…
Provisional conclusion
• The 20% threshold seems a priori more appropriate for urban areas mapping. The same or different thresholds can be chosen for different classes (e.g. forêts, wetlands…) and in differnt geographical contexts.
• The urban layer will be overlaid and combined with the other layers on agriculture, forêts, natural zones.
• Smaller themes will be given priority to the larger ones in order to minimise the relative errors. Adjustments will be done accordingly.
• The method is to some extent a simulation of visual photo-interpretation.
Land cover change account 2000 – 2010 / Urban sprawl
• Sources: the databases of buildup areas 2010 (LAVIMS) and ~2000
Land Cover / M01 Urban 2000
Land Cover / M01 Urban 2010
Land Cover change / M01 Urban 2000 - 2010
Land Cover stock and change / M01 Urban
Urban density (%) by Districts 2000
Urban density (%) by Districts 2010
Urban density (%) by Districts / Increase 2000-2010
A first account of Land Cover change/ Urban sprawl 2000-2010 by districts
ValuesRow Labels Sum of M01_Urban2000 Sum of M01_Urban2010 Sum of Change M01 2000-2010Black River 3.339511 7.494046 4.199487Flacq 4.045553 9.301728 5.376029Grand Port 4.061176 6.912077 3.054112Moka 2.818719 7.380331 4.676058Pamplemousses 7.049919 16.385576 9.679291Plaines Wilhems 20.597462 28.515584 8.021523Port Louis 26.674336 38.997219 12.421997Rivière du Rempart 7.468229 17.035051 9.664594Savanne 2.663638 4.907965 2.33844Grand Total 78.718543 136.929577 59.431531
A firs account of Land Cover change/ Urban sprawl 2000-2010
Black R
iver
Flacq
Grand Port
Moka
Pamplem
ousses
Plaines
Wilh
ems
Port Louis
Rivière
du Rempart
Savan
ne
0
5
10
15
20
25
30
35
40
Sum of M01_Urban2000Sum of M01_Urban2010Sum of Change M01 2000-2010
M01-Urban 2010 by river catchments