New SST indices and strength of their correlation with long and short rains over East Africa at different lead times relative to some existing indices

• Published in: International journal of climatology Vol. 44; no. 10; pp. 3214 - 3239
• Main Authors: Assamnew, Abera Debebe, Tsidu, Gizaw Mengistu
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01-08-2024; Wiley Subscription Services, Inc
• Subjects: Anomalies; Atlantic Oscillation; Bivariate analysis; composite; Correlation; CRU; Dipole moments; El Nino; El Nino phenomena; El Nino-Southern Oscillation event; indices; Oceans; Performance prediction; Precipitation; Rain; Rainfall; Rainfall forecasting; Regression analysis; Regression models; Sea surface temperature; Sea surface temperature anomalies; season; Southern Oscillation; Southern Oscillation Index; SST; Surface temperature; Indian Ocean; Pacific Ocean; East Africa
• ISSN: 0899-8418; 1097-0088
• DOI: 10.1002/joc.8520

This study assesses the skill of existing indices and newly introduced sea surface temperature (SST) indices in predicting rainfall over East Africa (EA) during short rains from September to December (SOND) and long rains from March to May (MAM). The existing and newly introduced indices were selected based on grid point correlation and spatial mean correlation. West–east SST gradient over the Indian Ocean (WEIO), north–south SST gradient over the Pacific Ocean (NSPO), anomalies over the western Indian Ocean (WIO) and anomalies over the eastern Pacific Ocean (EPO) from newly developed indices, and bivariate ENSO time series (BEST), Oceanic Niño Index (ONI), Niño3.4, western Pacific gradient (WPG) and dipole index moment (DMI) from existing indices have a positive correlation with SOND rainfall at 5% (p<0.05) significance level. West–east SST gradient over the Pacific Ocean (WEPO) and SST anomalies over the southern Pacific Ocean (SPO) from new indices and Southern Oscillation Index (SOI) from existing indices have a significant negative correlation with SOND rainfall over EA. SST anomalies over central and eastern Pacific and Indian oceans significantly correlate with SOND rainfall over southern EA (SEA) and equatorial EA (EEA). Rainfall over northern EA (NEA) has a significant negative correlation with SST over the east and central Pacific oceans. Negative correlation extends northeast and southeast from North of Australia over the Pacific Ocean during MAM. WEIO, WIO, NSPO, Niño3.4, Atlantic Oscillation (AO) and WPG have shown strong and significant (p<0.05) relations at least over one subregion. The new SST and existing oceanic indices have a stronger and more significant correlation with short rains than long rains over EA, except in the northern part of EA. Correlation, RMSE and skill score of regression models revealed that new indices showed higher performance than existing indices over EEA and SEA in predicting rainfall, unlike over NEA during both seasons.
Rainfall over East Africa has displayed a mainly opposing pattern throughout the boreal spring and boreal autumn seasons, according to the sign sum of correlation and composite relation from concurrent to 3‐month lag. The northern parts of East Africa (NEA) during the boreal spring season and the eastern and southern parts of EA (EEA and SEA) during the boreal autumn season have shown a more direct relationship, whereas the EEA and SEA during the boreal spring season and the NEA during the boreal autumn season have shown an inverse relationship.