Modelling and application of geostatistical techniques and remotely sensed data to estimate grid-bases rainfall of Pangani Basin (Tanzania)

摘要

The purpose of this study is to estimate the spatial rainfall in Pangani River using techniques called geostatistics. Three geostatistical techniques were applied in the estimation of rainfall: Simple Kriging (SM), Ordinary Kriging (OK) and Kriging with trend model or external drift (KT). Estimation results show that Ordinary Kriging produced reasonably better estimation than others single variable kriging methods (taking rainfall alone) with R~2 value 58 % for annual rainfall. An encouraging improvement was observed by including CCD and DEM data as secondary variable in the kriging with external drift (KT) model. The R~2 between the observed and estimated rainfall was improved from 58.0 to 73% when CCD is used as a secondary variable and 76% for DEM for annual data. On monthly basis, the estimation performance is even better than the annual case during the wet seasons from April to June.