Temesgen Gebremariam ASFAW and Jing-Jia LUO

1Institute for Climate and Application Research (ICAR)/CICFEM/KLME/ILCEC, Nanjing University of Information Science and Technology, Nanjing 210044, China

2Institute of Geophysics Space Science and Astronomy, Addis Ababa University, Addis Ababa 1176, Ethiopia

Fig. S1. Average fraction (%) of annual precipitation occurring during July-September.

Fig. S2. The percentage of grid cells over each cluster in East Africa (Fig. 1) where NUIST-CFS1.0 has positive skill (i.e., ROCSS and RPSS ＞0, GDS＞0.5).

Fig. S3. JJAS seasonal precipitation (mm season-1) during 1982-2019 aggregated from 20°E to 50°E [(a) and(b)] during La Ni?a (blue line), El Ni?o (red line), and neutral (green line) years, based on (a) the ensemble mean forecast of NUIST-CFS1.0 and (b) the CHIRPS observations, and [(c) and (d)] during negative (blue line), positive (red line), and neutral (green line) IOD years, based on (c) the ensemble mean forecast of NUIST-CFS1.0 and (d) the CHIRPS observations. The dashed and solid lines represent the individual and mean of those events, respectively.

Fig. S4. JJAS seasonal precipitation (mm season-1) differences [(a) and (b)] between El Ni?o and La Ni?a years and[(c) and (d)] between positive and negative IOD years based on the ensemble mean forecasts of NUIST-CFS1.0 (left column) and the CHIRPS observations (right column). Dots represent significance at the 10% level.

Fig. S5. Composites of JJAS 850-hPa wind (m s-1, vectors) and geopotential height (m, shaded) from [(a) and (b)]NUIST-CFS1.0 and [(c) and (d)] ERA5.