Here, building on recent analyses of East African October-December (OND) and March-June droughts, we suggest that an event-based analysis framework can help resolve the East African Climate Paradox. Examining trends, we face a discrepancy between the observed intensification of the Walker Circulation and the weakening predicted by climate models. Examining ENSO events, however, we suggest that the models and observations converge on more frequent strong El Nino events, interspersed with Neutral/La Nina conditions enhanced by anthropogenic warming in the western Pacific.
Interestingly, the East African response to these patterns differs substantially by season. During boreal spring we find a weak response to El Nino events and a strong response to La Nina-like SST patterns. We suggest that this latter sensitivity has emerged due to warming in the West Pacific and an intensification of the Indian Ocean branch of the Walker Circulation. During boreal autumn, however, rising upper level heights over the Indian Ocean may have prevented such an intensification. For OND, La Nina-related droughts are not enhanced. Thus the ‘paradox’ seems linked to seasonal variations in and over the Indian Ocean, and the difficulty that climate models have in recreating these complex circulations.
Our discussion also briefly examines potential biases in tropical warm pool vertical warming patterns. Climate change models anticipate upper level warming (weakening the Walker Circulation), while observations indicate cooling (strengthening the Walker Circulation). This discrepancy may help explain part of the paradox.
Figure 1. a. Standardized EA rainfall from CenTrends/CHIRPS, stars denote droughts. b. 20-yr EA rainfall with bootstrapped confidence intervals and standard errors. Vertical error bars denote CenTrends kriging standard errors. Horizontal lines indicate deviations significant at p=0.05, based on 10,000 bootstrapped samples. Red line indicates West Pacific Warming Mode regression estimates. Ethiopia time series averaged over Ethiopia southeast of 35°E, 9°N. c. EA decadal rainfall decadal variability, based on WPWM regression residuals d-e. Dry season composites of MAMJ SST and Reanalysis 1 200 hPa heights. Screened for significance at p=0.1. f-g. MAMJ SST time series. h-i. MAMJ 200 hPa heights from multiple reanalyses. j. Changes in CAM5 200 hPa heights and winds during 1981-2016 versus 1921-1980 WNP warm events, screened for significance at p=0.1 (Funk et al. 2018).