A comparison of gridded datasets of precipitation and temperature over the Eastern Nile Basin region

Abstract

Three gridded datasets of precipitation and surface air temperature over the Eastern Nile Basin (ENB) were compared and evaluated for further use in studies of the basin, such as for the optimization and validation of regional climate models for the basin and for hydrologic studies of the ENB. The Tropical Rainfall Measuring Mission (TRMM), Climate Research Unit (CRU), and Global Precipitation Climatology Center (GPCC) gridded datasets for precipitation were assessed and compared to available ground-based observations of the precipitation in the Ethiopian Highlands, which receive most of the precipitation that falls in the basin. Pearson’s correlation coefficient (COR), the modified index of agreement (MIA), and the mean absolute error (MAE) were used in statistical analyses of the gridded datasets. TRMM and GPCC were found to agree well with each other and with the GHCN station data, although GPCC showed the closest agreement with the ground-based observations. This advantage of GPCC was clearly apparent in the calculated statistical metrics, where GPCC gave the best performance (the highest COR and MIA and the lowest MAE) among the three datasets when they were compared with the ground-based observations. CRU indicated drier conditions than the other two datasets, especially over the highlands, while it also yielded the poorest agreement with ground-based observations. Temperature datasets for the ENB from the University of Delaware (UDel) and CRU were also evaluated. Both datasets performed well, and the bias between them was small. When compared to ground-based observations, both showed high COR and MIA and low MAE for most stations, but UDel presented slightly better calculated statistical metrics than CRU. Based on these results, we recommend the use of GPCC when modeling the precipitation and UDel when modeling the air temperature over the ENB.