GRADGRID4: an advanced spatial interpolation tool combining GRASS and R functions

Presentation | Presented

  • Martin Mergili, Institute of Geography, University of Innsbruck, Austria

Standard interpolation algorithms (Kriging, Spline, IDW, etc.) are valuable tools for many purposes when points have to be interpolated to a continuous space. They are less applicable when a spatially highly variable descriptor determines the variability within the target data. GRADGRID4, combining GRASS and R functionalities, was designed for solving this problem. The spatial interpolation of point values of the target variable is based on one or more descriptors: local gradients are determined for each pixel using a regression approach, in order to cope with spatially variable dependencies between the target variable and the descriptors. The gradients are then combined with weighted local averages. The algorithm was applied for producing spatially continuous maps of various climatological variables from point data (meteorological stations) and a digital elevation model. The validation indicated a high accuracy of the results, outperforming the related method of Co-Kriging. The tool and its applicability will be presented with the examples of mean annual and seasonal sums of precipitation for the region of Tyrol (Austria and Italy).

Supporting Files