Reliability of climate models for prediction of high impact climate events
For the first time the world leading UK Met Office modelling suite has been evaluated with an African Lens to assess its reliability in representing high-impact weather events
For the first time, the world-leading UK Met Office modelling suite has been evaluated with an African Lens to assess its reliability in representing high-impact weather events. The method is now being published and is being shared with other modelling centres.
Results show that the Met Office models provide good reliability in predicting hot temperature extremes (heatwaves) for seasonal and multiannual prediction. Reliability for dry extremes (droughts) is lower and is more variable with region. Relatively good reliability is found for the Sahel region while poor reliability was found for some other regions e.g. the Greater Horn of Africa and Guinea Coast regions in the March-May season.
Results demonstrate the potential for developing operational predictions for the Sahel out to several years ahead, to complement the current seasonal forecasting. On this basis experimental predictions have been generated for the 2014-2017 period in the Sahel and indicate an enhanced risk of drier than normal conditions over the next years.
The results also have practical implications for interpretation of model scenarios in adaptation planning. Additional caution should be exercised in such studies if the models used show low reliability for the region and variable of interest - though the general topic of assessing model "trustworthiness" requires further research.