Algorithms for Computerized Estimation of Thiessen Weights

Abstract

The Thiessen polygon method is commonly used in hydrological modeling to estimate average areal rainfall from point measurements. This paper presents and compares two fast methods for computerized estimation of Thiessen weights. The first method, termed as the “modified raster method” uses a grid approach with selective fine resolution rasterization of the catchment to enhance the accuracy. The second method, termed as the “vector method,” uses the principles of analytical geometry to estimate Thiessen weights accurately. The accuracy and time requirement of the modified raster method at different resolutions is compared for rain-gauge networks of four catchments. It is observed that excessive fine resolution of rasterization does not significantly improve the accuracy of estimation of Thiessen weights relative to an increase in computation time. Two case studies are presented as the application of the vector method with uncertainty in data availability. It was found that the vector method can be used for high speed automated generation of Thiessen polygons in real time and also for generation of a large number of combinations for simulation studies.