Why do climate models give different predictions for East African rainfall?

Sponsored content: Accurate predictions for future rainfall are essential for planning cities and agriculture

(Photos: Oxfam and Dfid)


Researchers from the UK Met Office are improving understanding of why different climate models produce different predictions of rainfall over East Africa’s two wet seasons.

This important advance works towards assisting researchers and practitioners in selecting the best climate models for various applications over East Africa, and may ultimately assist decision-makers in the region with more skilful and reliable predictions to inform climate resilient decisions.

Evidence suggests that rising greenhouse gas emissions and other human activities have already led to large changes in extreme rainfall events over West Africa. Scientists suspect that climate change may be to blame for similar extreme events over other parts of the continent. This will worsen impacts on vulnerable communities already suffering from droughts and floods. Although climate models agree that extreme rainfall will increase, they disagree on the pattern of future changes in seasonal rainfall. This makes rainfall predictions for specific locations highly uncertain, presenting significant challenges to people who need reliable information to make decisions resilient to climate change impacts.

A new study, published in the Journal of Climate, seeks to address this issue by understanding why 40 different climate models produce different future seasonal totals in rainfall over East Africa’s two wet seasons. The paper presents three important findings.

Firstly, uncertainty in different models’ predictions of East African rainfall is largely caused by the way the models simulate the effects of global warming on local East African climate as well as climate in other areas of the world that is connected to East Africa’s climate through atmospheric chains of events (referred to as “teleconnections” in the scientific literature).

Secondly, during the long rains season (from March to May), one particular model predicts an exceptionally large warming of the southern Indian Ocean, which triggers a chain of events causing it to predict approximately a doubling of the seasonal rainfall total over East Africa. This is by far the largest rainfall increase suggested by any of the models. If this behaviour can be related to something we can observe in today’s climate, then we can establish whether this model’s prediction should carry less importance or be retained as a most-extreme scenario. Either way, this will be useful information, particularly for researchers trying to understand the impact of future climate change on various economic sectors, and indirectly to communities who may benefit from improved decision making based on this information.

Lastly, the study gives more evidence that global warming causes a chain of responses in the atmosphere that is often very different from the chain of responses that cause year-to-year variability in rainfall. This is important because it suggests that we need more sophisticated ways to evaluate climate model predictions. The research helps lay the groundwork for a better discrimination of the best and worst models employed for various applications in the East African region.

“Extremes of tropical rainfall, whether they be droughts or floods such as the devastating events recently experienced in Kenya, can be catastrophic in developing countries. Decision-makers need more robust information about changes in climate over the coming decades but are hampered by disagreements amongst climate models. We hope this study will contribute to reducing that spread amongst model predictions, and inspire further work on this important topic” says Dr Dave Rowell of the UK Met Office.

“Understanding and predicting future changes in tropical rainfall patterns is one of the greatest and most important challenges facing climate science today” says Dr John Marsham of the National Centre for Atmospheric Science and the University of Leeds. “This is an important study that sheds new light on why predictions differ over East Africa. Through our work in East Africa, we are communicating it directly to a broad range of researchers, practitioners and decision-makers, to enable feedback from users on the usefulness of such new science for decision making, and on their priorities for future research”.

This research was conducted as part of the HyCRISTAL project. HyCRISTAL is an interdisciplinary research partnership between institutions in the UK, USA, Kenya, Tanzania and Uganda. It aims to improve understanding of climate change and its impacts in East Africa and works with the region’s decision-makers to manage water for a more climate-resilient future. HyCRISTAL is supported by the East African Community and is funded by the UK Department for International Development and the Natural Environment Research Council through the Future Climate for Africa programme.

This paper is immediately available under open access: https://journals.ametsoc.org/doi/full/10.1175/JCLI-D-17-0830.1