Thesis (M.Sc.) - Cairo University - Faculty of Science - Department of Astronomy and Meteorology

A seasonal forecasting system that is capable of skillfully predicting rainfall on a regional scale over Ethiopia would be crucial for Ethiopia and the downstream countries (Egypt and Sudan). The main rainy season over Ethiopia Kiremt (Jun-Sep) contributes with the major mount of the Nile flow, about 70 %of the total amount of water resources reaches downstream countries through the Nile River. This study examines potential predictors for June{u2013}September rainfall over the Blue Nile basin in northwest Ethiopia using mainly statistical approaches. Due to the spatial variability in rainfall, the rain gauge stations was grouped into three homogenous rainfall regions. The study focus on the region of the Blue Nile basin in northwestern Ethiopia refered as cluster two. Oceanic and atmospheric climate indices with different lead times are very useful as potentially predictors of seasonal forecast models. Testing teleconnection between study area and regularly updated climate indices using correlation analysis with different lead times to identify potential predictor for Multiple Linear Regression model (MLR). Climate indices with significant correlation were chosen to develop MLR which was verified using a leave- one-out cross validation technique with stepwise regression. Our experiments show that the strongest correlated climate indices with Ethiopian rainfall are North Atlantic Oscillation (NAO), West Paci{uFB01}c Pattern (WP), Southern Oscillation Index (SOI), Scandinavia Index pattern (SCA), Paci{uFB01}c/North American Pattern (PNA), Monthly mean SST anomalies in the Niño 3.4 region, oceanic Niño index (ONI) and Western Hemisphere Warm Pool (WHWP) relative to other indices used over different lead time

Issued also as CD