Process-based assessment of an ensemble of climate projections for West Africa

Determining the level of confidence in regional climate model projections could be very useful for designing climate change adaptation, particularly for vulnerable regions. The majority of previous research to evaluate models has been based on the mean state, but for confidence in projections the plausibility of the mechanisms for change is just as, if not more, important. In this study we demonstrate a methodology for process-based assessment of projections, whereby circulation changes accompanying future responses are examined and then compared to atmospheric dynamics during historical years in models and reanalyses. We apply this methodology to an ensemble of five global and regional model experiments and focus on West Africa, where these models project a strong drying trend. The analysis reveals that this drying is associated with anomalous subsidence in the upper atmosphere, and large warming of the Saharan heat low region, with potential feedback effects via the African easterly jet and West African monsoon. This mode occurs during dry years in the historical period, and dominates in the future experiments. However, the same mode is not found in dry years in reanalysis data, which casts doubt on the reasons for strong drying in these models. The regional models show a very similar response to their driving global models, and are therefore no more trustworthy in this case. This result underlines the importance of assessing model credibility on a case-by-case basis and implies that process-based methodologies should be applied to other model projections before their outputs are used to inform decision making.