Bias correction and characterization of climate forecast system re‐analysis daily precipitation in Ethiopia using fuzzy overlay

Abstract

ABSTRACTKnowledge of spatiotemporal variability of rainfall magnitude, pattern and trend is fundamental for understanding hydrological systems and runoff prediction for both gauged and ungauged catchments. These variables can be derived from rainfall‐monitoring programmes with adequate spatial distribution and temporal coverage. However, rainfall‐gauging stations in most developing countries are distributed sparsely. Remotely sensed rainfall datasets are becoming alternative rainfall data sources for larger area applications and are proven to have adequate spatiotemporal resolutions. Climate forecast system re‐analysis (CFSR) is one such dataset provided by the National Center for Environmental Prediction (NCEP). This dataset captures the rainfall pattern in Ethiopia but with s magnitude bias of over‐ and underestimations. In this study, magnitude bias correction of the CFSR dataset with a linear scaling technique resulted in a rainfall grid of the country with ∼38 km spatial resolution of a 32 year (1979–2010) daily rainfall dataset. For the bias correction, observed annual rainfall from 930 and daily rainfall from 195 rain gauges were used. The study also attempted to understand the space and time variability of the rainfall through the construction of shape, magnitude and composite rainfall regimes for the entire country. The rainfall regimes of the country were developed using the fuzzy overlay technique with multi‐indices of rainfall. The rainfall regimes address the frequency, duration, timing and magnitude variability of rainfall. The performance of the dataset generation and rainfall regime classification was evaluated using Nash–Sutcliffe Efficiency (NSE) and percent bias (PBIAS) values, which were found to be 0.8 and 1.3, respectively.