Regionalization and association with global climate drivers of rainfall in the Rift Valley Lakes Basin of Ethiopia

Regionalization and evaluation of associations of hydro-climatic variables with indices of global climate drivers are helpful for local-scale seasonal forecasting of weather patterns and planning water resources management. The main objective was to regionalize the Ethiopian Rift Valley Lakes Basin into homogenous sub-regions based on monthly and seasonal rainfalls and investigate the influence of some global climate drivers at the scale of the sub-regions to devise predictive tools. The dataset used monthly rainfall and global and regional climate indices anomalies from 1983 to 2014. Principal component analysis (PCA), Pearson correlation, and multivariate linear regression methods were applied using SPSS and R software. Based on PCA analysis, three principal components were identified which have a significant association with global climate indices. Over the study period, there were nine moderates to strong El Nino and six La Nina events; the warming phases received more rainfall and less in the cooling phase. Lagged sea surface temperature (SST) and atmospheric variables were selected as predictors based on significant associations with regional rainfall. The multiple linear regression analysis revealed the possibilities of deriving seasonal forecasts at the local level. The study showed that the model derived an excellent and scientifically robust seasonal rainfall prediction skill in a short lead time of the different seasons at a range of 30 to 80% in the sub-regional level. The capabilities of rainfall prediction skills help reduce climate-induced hazards in planning and decision-making processes by providing timely and specific climate information.