Analysis of Historical and Projected Future Climate of Mali, West African Sahel

The West African country of Mali experienced devastating drought from the late 1960s to the mid 1980s, followed by partial rainfall recovery that remains below pre-1960s climatology. Here, we examine the historical and projected temperature and rainfall variability across the Malian Sahel in an effort to assist future planning for food security. Particular emphasis is placed on clarifying the teleconnection between sea-surface temperatures (SST) expressed by the Atlantic Multidecadal Oscillation (AMO) and the West African monsoon (WAM). Using gridded observations and reanalysis, we show that cool/wet (1950-1967) and warm/dry (1980-1997) end-member climates over Mali correspond to warm and cool phases of the AMO, respectively, with associated atmospheric patterns consistent with negative (slow winds, shallow poleward pressure gradient) and positive (fast winds, steep poleward gradient) modes of the North Atlantic Oscillation (NAO). These opposing climate regimes (AMO-warm/NAO-negative and AMO-cool/NAO-positive) are coupled to strength of the Sahara Low and latitudinal position of the Intertropical Convergence Zone (ITCZ), thereby accounting for changes in strength of the WAM. A case is made that multi-decade North Atlantic SST variability arises naturally from volcanic forcing and resultant changes in strength and position of the westerly winds. However, greenhouse-gas warming, stratospheric ozone depletion, and Arctic sea-ice loss is likely modifying the system behavior. In an examination of CMIP5 general circulation model (GCM) output for Mali, we find that temperature and precipitation in historical simulations fail to validate against observations, and notably do not show multi-decade variability. Atmospheric circulation across the North Atlantic is furthermore anomalously strong in CMIP5 due to an unrealistically steep pressure gradient between Azores and Iceland. These shortcomings undermine the fidelity and meaning of CMIP5 future rainfall projections across the western Sahel. We conclude here that downscale studies should target reanalysis, and investigate extreme years (e.g., cool/wet or hot/dry) as future climate analogs rather than depend on CMIP5. In all from this work, we suggest five plausible future climate scenarios for 2030-2050: 1) standard CMIP5 projection of 2 °C warming with slight rainfall decline; 2) Warming > 1 °C with rainfall remaining at present norm, or increasingly slightly due to poleward displacement of the ITCZ; 3) Warming > 1 °C with diminished rainfall or drought from southward displacement of the ITCZ; 4) Warming > 1 °C with onset of severe drought arising from renewal of high volcanic activity, and subsequent development of strong NAO-positive circulation and cool-AMO sea-surface temperature distribution; 5) Abrupt climate shift in response to collapse of summer Arctic sea ice, wherein any of the scenarios above could develop within a decade.