Evaluation of Two Global Rainfall Models for Water Resources Planning and Management in Nigeria

Abstract

Given the sparseness of weather stations in Nigeria, there is an increasing need for alternative sources of rainfall data such as satellite measurements, numerical models and reanalysis. Nevertheless, the complexity of such data requires proper evaluation and validation. Therefore, this study evaluated two globally available rainfall products from Climate Research Unit (CRU) and University of Delaware (UNIDEL) using rain gauge data obtained from Nigerian Meteorological Agency (NIMET), over a period of twenty years (1980-1999) covering 24 stations. Time series plot and statistical tools were used to evaluate the products on annual, seasonal and zonal basis. The results show that the two products demonstrated comparable ability and sufficiently captured the spatial and temporal patterns of rainfall over the country. However, the products overestimated and underestimated during the dry and rainy seasons, respectively. Although, correlation was comparatively high between 0.3 and 0.8, but negative in few instances, mean bias error (MBE) and root mean square error (RMSE) were generally high depicting high random error. The performance of the products was best in the Sahel, followed by the Savannah and Forest zones, with UNIDEL showing better performance in most cases. Consequently, we recommend further studies to validate the present results on the use of gridded data in the country.Keywords— evaluation, CRU, UNIDEL, rain gauge, rainfall products