An Observational Constraint on CMIP5 Projections of the East African Long Rains and Southern Indian Ocean Warming

• Published in: Geophysical research letters 2019-06, Vol.46 (11), p.6050-6058
• Main Author: Rowell, David P.
• Format: Article
• Language: English
• Subjects: 21st century; Climate; climate change; Climate models; Cloud cover; cloud feedbacks; Clouds; CMIP5; East Africa; emergent constraint; Indian Ocean; Ocean temperature; Ocean warming; Oceans; Rain; Rainfall; Rainfall forecasting; Sea surface; Sea surface temperature; Seasonal rainfall; Sensitivity; Surface temperature
• ISSN: 0094-8276; 1944-8007
• DOI: 10.1029/2019GL082847

Two outlying projections of the East African Long Rains suggest the seasonal rainfall may double by the late 21st century. Previous work has linked these extremes—found in the IPSL‐CM5A model—to an exceptional March to May warming of the southern Indian Ocean. The current study shows a strong feedback between sea surface temperature (SST) increases and reduced low‐level cloud cover (with similar behavior in other southern subtropical oceans). An observational constraint is developed by demonstrating a correlation across 28 models between the strength of present‐day interannual SST‐cloud sensitivity and future SST response. Verification of the present‐day sensitivity finds that IPSL‐CM5A's feedbacks are very likely overestimated. It is therefore suggested its projections should be discounted for the March to May southern Indian Ocean and East African Long Rains. This narrows the CMIP5 plausible range of Long Rains totals by a third. Plain Language Summary Predictions of seasonal mean rainfall for the middle to late 21st century vary substantially between climate models. This is problematic for those making policy and infrastructure decisions that must remain resilient to a range of possible futures. Consequently, climate scientists are often requested to identify the most and least reliable models, in order to narrow this range. This study focuses on the Long Rains of East Africa, for which one unusual model suggests a doubling of seasonal totals by the late 21st century. The primary physical causes of this exceptional prediction are examined and then linked to the strength of two‐way interactions between clouds and surface temperatures in the southern Indian Ocean for a comparison with observational data. It is concluded that this model's extreme predictions for both the southern Indian Ocean and the East African Long Rains are likely unreliable. The plausible range of the full ensemble of Long Rains predictions is narrowed by about a third. Key Points Outlying projections of the East African Long Rains suggest seasonal totals may double; judging their reliability is critical for users These outliers exhibit an unusually strong feedback between clouds and surface warming in the southern subtropical Indian Ocean An observational constraint founded on interannual surface cloud sensitivities reduces projection uncertainty of the Long Rains by a third