Statistical predictive models were developed to investigate how global rainfall predictors relate to the March-May (MAM) rainfall over Southern Nations, Nationalities and People's Region (SNNPR) of Ethiopia. Data utilized in this study include station rainfall data, oceanic and atmospheric indices. Because of the spatial variations in the interannual variability and the annual cycle of rainfall, an agglomerative hierarchical cluster analyses were used to delineate a network of 20 stations over study area into three homogeneous rainfall regions in order to derive rainfall indices. Time series generated from the delineated regions were later used in the rainfall/teleconnection indices analyses. The methods employed were correlation analysis and multiple linear regressions. The regression modes were based on the training period from 1987-2007 and the models were validated against observation for the independent verification period of 2008-2012. Results obtained from the analysis revealed that sea surface temperature (SST) variations were the main drivers of seasonal rainfall variability. Although SSTs account for the majority of variance in seasonal rainfall, a moderate improvement of rainfall prediction was achieved with the inclusion of atmospheric indices in prediction models. The techniques clearly indicate that the models were reproducing and describing the pattern of the rainfall for the sites of interest. For the forecast to become useful at an operational level, further development of the model will be necessary to improve skill and to determine the error bounds of the forecast.
|Number of pages||10|
|Publication status||Published - 2015|
All Science Journal Classification (ASJC) codes
- Atmospheric Science