Street blocks features computation

This repository contain the python script (in a jupyter notebook) used for computing landscapes metrics in street blocks or any other landscape unit defined as a shapefile to be provided by the user.

This code was published belong to the following paper:

Grippa & al. Mapping Urban Land Use at Street Block Level Using OpenStreetMap, Remote Sensing Data, and Spatial Metrics. ISPRS Int. J. Geo-Inf. 2018, 7, 246. doi:10.3390/ijgi7070246

Cite this code

Please use the following DOI for citing this code:

Related code

The code provided in this repository compute spatial metrics for a layer of polygons. When working on urban environment, these polygons could be, for example, street blocks in order to classify land use. Another repository provide a computer code to create street block geometries from OpenStreetMap data => https://github.com/ANAGEO/OSM_Streetblocks_extraction.

Outputs

The code relies on GRASS GIS and mainly on the r.li suite. It enable for automated creation of the r.li configuration files which otherwise should be created in the graphical user interface more info.

The script is supposed to work with a user-provided land cover map, NDVI, NDWI, nDSM. If any of them is missing, the user would need to adapt the code.

The script will compute the following landscape metrics

Spatial metrics at the "landscape" level:

• Dominance
• Pielou
• Renyi
• Richness
• Shannon
• Simpson

Spatial metrics at the "class" level:

• "patchnum" : Patch number
• "patchdensity" : Patch density
• "mps" : Mean patch size
• "padsd" : Stand. dev. of patch size
• "padcv" : Patch size coef. of variation
• "padrange" : Range of patch size
• "shape" : Shape index
• "prop_xx" : Proportion of the class

Street blocks morphology metrics:

• "area" : Area
• "perimeter" : Perimeter
• "compact_circle" : Compactness relative to a circle
• "compact_square" : Compactness relative to a square
• "fd" : Fractal dimention

Spectral metrics:

• "ndvi_stddev" and "ndvi_median" : Std. dev. and median of NDVI
• "ndwi_stddev" and "ndwi_median" : Std. dev. and median of NDWI

Other metrics:

• "mean_build_height" : Mean nDSM value of built pixels
• "count_buildpixels" : Number of built pixels in the block

Example

Here after are presented few spatial metrics computed on a land cover map and used as main features for land use classification at the streetblock level.

Land cover map Shannon index (landscape level) Patch density on "low elevated building" class (class level) Landuse classification