Calculating continuous raster representing land cover proportion from categorical raster using QGIS?

Question

I have a raster (5 x 5 m resolution) representing land cover. Each cell contains a integer value (from 1 to 10) representing a total of ten land cover classes (e.g. 1-forest; 2-agricultural areas; 3-urban areas;(...)10-scrublands).

Based on this categorical land cover raster, I would like to produce ten rasters with a resolution of 10 x 10 m representing the proportion of each land cover class per cell. In another words, I would like to produce a raster representing the proportion of forest, another raster representing the proportion of agricultural areas, etc.

This figure explains exactly what I want to do (for illustrative purposes, I only accounted for 3 classes in the figure).

I have been looking for similar questions in the internet, but usually people had slightly different questions (e.g. calculate land cover proportion within a polygon).

Does anyone know how to do this in QGIS?

Answer 1

I haven't got access to QGIS on this laptop, but here's roughly how I would approach it:

1. Take your 5m land use raster and use Raster calculator to produce a new boolean raster that has a value of 1 for a given class, and 0 for all other values. For example for the forest class: 5m_landuse = 1
2. Use the r.resamp.stats tool from the GRASS GIS/Raster toolbox to aggregate the new boolean raster to 10m cells. Select the output resolution to 10m and the aggregation method to average. For every 10m pixel, this will take the average of the four contributing 1s (forest) and 0s (non-forest) in your 5m pixels, which ends up being equal to a proportion (e.g. average of 1, 1, 1 and 0 = 0.75).

You will need to repeat this for each of your ten land use classes, but this can probably be automated using the graphical modeller tool. Good luck!

Answer 2

I agree on the method proposed in the first answer (first convert to one binary raster per class, take take the average). However, I think that it would be slighly faster with the gdal tools (from raster > conversion if you don't like command lines).

gdal_translate in_binary_5m.tif out_proportion_10m.tif -r average -outsize 50% 50%

Remark -outsize 50% 50% set the output at a 50% resolution (equivalent to 1/50% (= double) the size of the pixels)