18.30. Interpolation and contouring

Module contributed by Paolo Cavallini - Faunalia

Catatan

This chapter shows how to use different backends to calculate different interpolations.

18.30.1. Interpolation

The project shows a gradient in rainfall, from south to north. Let’s use different methods for interpolation, all based on vector points.shp, parameter RAIN:

Peringatan

Set cell size to 500 for all analyses.

  • GRASS ‣ v.surf.rst
  • SAGA ‣ Multilevel B-Spline Interpolation
  • SAGA ‣ Inverse Distance Weighted [Inverse distance to a power; Power: 4; Search radius: Global; Search range: all points]
  • GDAL ‣ Grid (Inverse Distance to a power) [Power:4]
  • GDAL ‣ Grid (Moving average) [Radius1&2: 50000]

Then measure variation among methods and correlate it with distance to points:

  • GRASS ‣ r.series [Unselect Propagate NULLs, Aggregate operation: stddev]
  • GRASS ‣ v.to.rast.value on points.shp
  • GDAL ‣ Proximity
  • GRASS ‣ r.covar to show the correlation matrix; check the significance of the correlation e.g. with http://vassarstats.net/rsig.html.

Thus, areas far from points will have less accurate interpolation.

18.30.2. Contour

Various methods to draw contour lines [always step= 10] on the stddev raster:

  • GRASS ‣ r.contour.step
  • GDAL ‣ Contour
  • SAGA ‣ Contour lines from grid [NB: in some older SAGA versions, output shp is not valid, known bug]