This work compares future projections of rainfall over the Pra River Basin (Ghana) using data from five climate models for the period 2020–2049, as referenced to the control period 1981–2010. Bias-correction methods were applied where necessary and models' performances were evaluated with Nash–Sutcliffe Efficiency, root-mean-square error and coefficient of determination. Standardised Anomaly Index (SAI) was used to determine variability. The onset and cessation dates and length of the rainy season were determined by modifying the Walter–Olaniran method. The ensemble means of the models projected a 1.77% decrease in rainfall. The SAI showed that there would be drier than normal years with the likelihood of drought occurrence in 2021, 2023, 2031 and 2036. The findings showed that high-resolution models (≤25 km) were more capable of simulating rainfall at the basin scale than mid-resolution models (26–150 km) and projected a 20.13% increase. Therefore, the rainfall amount is expected to increase in the future. However, the projected increase in the length of the dry season by the ensemble of the models suggested that alternative sources of water would be necessary to supplement rainfed crop production for food security.

这项工作使用参考1981-2010年控制期的2020-2049年期间的五个气候模式的数据，比较了Pra河盆地（加纳）的未来降雨预测。偏倚校正方法在必要时应用，并使用Nash-Sutcliffe效率，均方根误差和确定系数来评估模型的性能。使用标准化异常指数（SAI）确定变异性。雨季的开始和结束日期以及雨季的长度是通过修改Walter–Olaniran方法确定的。这些模型的整体均值预计降雨量将减少1.77％。SAI显示，到2021年，2023年，2031年和2036年，干旱年份可能比正常年份干燥。研究结果表明，高分辨率模型（≤25km）比中分辨率模型（26–150 km）更有能力模拟流域尺度的降雨，并预计增加20.13％。因此，预计将来的降雨量将会增加。但是，由于模型的综合，预计旱季时长的增加表明，为保障粮食安全，需要补充水源来补充雨育作物。