Consistent with recent studies, the post-1998 increase in drought frequency during the March-May (MAM) “long rains” is strongly associated with decadal variability in the Pacific Ocean. The results reveal that both the Pacific Decadal Oscillation (PDO) and Interdecadal Pacific Oscillation (IPO) show a positive and significant correlation to long rains. The percent variance explained (PVE) ranges from 25 to 64% and the association stems from low-frequency variability (>8 years) for both PDO and IPO. The Eastern Mediterranean SSTs (EMS) are found to be significantly related to long rains on the interannual time scale with PVEs between 10 and 36% in the early 20th century, but this teleconnection disappeared in the recent decades. The interannual variability of the October-December (OND) “short rains” is dominantly influenced by the SST anomalies in the Indian Ocean (Indian Ocean Dipole, IOD), particularly in recent years, with PVEs increasing to 80%. However, no significant connection could be derived between the IOD and the short rains for time series less than 20 for starting years between 1901 and 1951, with an exception for starting year near 1918. A relatively strong coupling between the two occurred after starting year 1951 and thereafter until 1987, when another abrupt shift in correlation occurred to the recently observed even stronger coupling. These abrupt changes have been identified by previous studies and confirmed as times when real shifts occurred in the Indian Ocean. El Niño-Southern Oscillation (ENSO) is associated with the short rains year-to year variability, yet the ENSO influence is mediated by the warming in the western Indian Ocean. This has been extracted by partial correlation analysis.
On the other hand, low-frequency variations of short rains are mostly explained by SST anomalies in the Pacific Ocean, as it is demonstrated by significant correlations to the low-pass filtered time series of ENSO and IPO indices. ENSO is also strongly associated with June-September (JJAS) “Kiremt rains” over the Ethiopian Highland. This correlation is significant, stable with PVEs larger than 50%, and mainly based on high-frequency variability (<8 years). However, the correlation between Kiremt rains and Atlantic Niño 3 (ATL3) changed from significantly negative to positive after the 1960s. Corresponding to decadal regime shift of the relation between the Atlantic Niño and the West African monsoon.
In summary, the present study revealed both stable and variable teleconnections between SST-based indices of remote climate variability and indices of seasonal rainfall over the GHA during 20th century. Three novel findings are worthy of note. On interannual time scales, (A) an early 20th century impact of the EMS on long rains has been discovered and (B) a regime shift within the West African monsoon system might have simultaneously changed the influence of the Atlantic Niño on Kiremt rains. Potential physical explanations will be discussed. Lastly (C), the study sheds more light on the link between decadal variability of the long rains and two well-known indices of Pacific decadal variability, the PDO and IPO.