Outdated version of the documentation. Find the latest one here.

18.36. Predicting landslides

Module contributed by Paolo Cavallini - Faunalia

Muista

This chapter shows how to create an oversimplified model to predict the probability of landslides.

First, we calculate slope (choose among various backends; the interested reader can calculate the difference between the outputs):

  • GRASS ‣ r.slope
  • SAGA ‣ Slope, Aspect, Curvature
  • GDAL Slope

Then we create a model of predicted rainfall, based on the interpolation of rainfall values at meteo stations:

  • GRASS ‣ v.surf.rst (resolution: 500 m)

The probability of a landslide will be very roughly related to both rainfall and slope (of course a real model will use more layers, and appropriate parameters), let’s say (rainfall * slope )/100:

  • SAGA ‣ Raster calculator rain, slope: (a*b)/100 (or: GRASS ‣ r.mapcalc)
  • then let’s calculate what are the municipalities with the greatest predicted risk of rainfall: SAGA ‣ Raster statistics with polygons (the parameters of interest are Maximum and Mean)