Information content of downscaled GCM precipitation variables for crop simulations

Abstract:

A simple statistical downscaling procedure for transforming daily global climate model (GCM) rainfall was applied at the local scale in Katumani, Kenya. We corrected the rainfall frequency bias of the GCM by truncating its daily rainfall cumulative distribution into the station's distribution using a wet-day threshold. Then, we corrected the GCM’s rainfall intensity bias by mapping its truncated rainfall distribution into the station's truncated distribution. Additional tailoring was made to the bias corrected GCM rainfall by linking it with a stochastic disaggregation scheme based on a conditional stochastic weather generator to correct the temporal structure inherent with daily GCM rainfall. Results of the simple and hybridized GCM downscaled precipitation variables (total, probability of occurrence, intensity and dry spell length) were linked with a crop model. An objective evaluation of the tailored GCM data was done using entropy. This study is useful for the identification of the most suitable downscaling technique, as well as the most effective precipitation variables for forecasting crop yields.