Flexible spatial models on the example of temperature in China

Abstract

Spatial modeling of temperature is of crucial importance for agriculture, industry and ecology. This work presents interpolation methods for the daily average temperature in China in the time period from 1957 to 2009. Due to complex topography and diverse climate of the country flexibility of the spatial models is of great importance. This study attempts to develop techniques which are able to minimize the spatial prediction error and to capture temperature extremes. The current research extends copula-based interpolation method and proposes the innovative IDW-GEV model. Spatial regression, kriging and inverse distance interpolation are used as a benchmark to evaluate the performance of suggested techniques.