Seasonal predictability of Ethiopian Kiremt rainfall and forecast skill of ECMWF's SEAS5 model

Predictability of Ethiopian Kiremt rainfall (June to September: JJAS) and forecast skill of the European Centre for Medium-Range Weather Forecasts (ECMWF) fifth-generation seasonal forecast system 5 (SEAS5) is explored during 1981-2019. The first empirical orthogonal function of observed rainfall explains 50.6% of the total variability and is characterized by positive rainfall anomalies largely confined over the northwestern and central-western regions of Ethiopia. Consequently, a Kiremt rainfall index (KRI) is defined for this region. The correlation coefficient (CC) between the observed and predicted KRI is 0.68 and 0.53 for May and April starts, respectively. Composite analysis of sea surface temperature (SST) and lower-level circulation based on excess and deficit years of Kiremt rains shows that the El Nino Southern-Oscillation is the main modulator of the Kiremt rainfall variability. The CC between KRI and Nino3.4 index is - 0.62, indicating that El Nino is accompanied by below-normal Kiremt rainfall, while La Nina is accompanied by above-normal amounts. The fifth generation of ECMWF atmospheric reanalysis (ERA5) shows that excess (deficit) Kiremt rainfall anomalies are associated with an anomalous low (high) pressure centered over northeast Arabian Peninsula and an anomalous in-phase (reverse) low-level Somali Jet. SEAS5 reproduces the spatial and temporal components of observed Kiremt rainfall variability, including the main climatic features associated with excess and deficit Kiremt rainfall in May and April starts. However, certain important observed features like above-normal SSTs in the Gulf of Guinea are not well predicted. Results indicate that Kiremt rains has some potential predictability and SEAS5 shows a moderate forecast skill. Probabilistic analysis shows highest values where predictability and deterministic skill are also highest.