Estimating Daily Regional Rainfall Fields by Multiquadratic Functions: Accuracy of Interpolation and Decision Making
Annegret H. Thieken
Journal of Geographic Information and Decision Analysis, Vol. 2, No. 2, pp. 168-181, 1998.
ABSTRACT: It is reported how far multiquadratic functions, a specific type of radial basis functions, are capable of estimating daily precipitation on a regional scale. This is done with the software package SURFER, Version 6 (Golden Software, Inc. 1995). Beforehand, spatial correlation and anisotropy within the data was analysed with VARIOWIN 2.2 (Pannatier 1996). A geometric anisotropy along the NE-SW axes was identified. Anisotropy was reflected in the interpolation model by anisotropy ratios that were associated to the ranges of the directional variogram models. The performance of the estimation was assessed by traditional statistical methods. In order to highlight possible consequences of errors on decision making, a typecasting of errors was carried out by simplified decision rules with relevance to environmental objectives.
KEYWORDS: multiquadratic function, classification of errors, decision making