Abstract
The crop yield modeling remains a central input in the assessment of climate change risks. This study aims at investigating to what extent that rainfall, temperature and runoff could ably transmit significant information needed for analyzing and predicting crop yields under extreme unmitigated climate change conditions (RCP 8.5) in the Nile basin, using mutual information and stepwise regression models. Monthly climate change outputs (2040-2079) were derived from the fifth phase of the coupled inter-comparison project (CMIP5) and six regional climate models of the CORDEX-Africa. Yield time series (1970-2016) of selected crops (sorghum, groundnut, wheat, potato, and sugarcane) were retrieved from the United Nations Food and Agriculture Organization (FAO). The analysis was carried out using the R package, QGIS, and Excel sheets. The MI results stated the significant interdependency of crop yields on weather variables, especially temperature. Developed stepwise crop yield models managed to explain 9-97% of inter-annual variability in crop yields, depending on location and crop concerned. Unlike predominating results, most crop yields in the Nile basin will be benefited from climate change by 2079. Generally, the crop farming would remain more resilient to global warming in Egypt and vice versa holds for the basin's tropical countries, especially Rwanda and Burundi. Yields of sugarcane, groundnut, and sorghum, respectively, showed the highest sensitivity to global warming. There is a compelling need to analysis of uncertainty in crop yields and climate change using finer spatial and temporal datasets in the Nile basin.