Abstract ID: 126
Skill assessment and sources of predictability for sub-seasonal rainfall forecasts in Africa
Lead Author: Felipe M. de Andrade
National Institute for Space Research, Brazil
Keywords: Sub-seasonal Prediction, African rainfall, S2S models, forecast verification, variability modes
Abstract: Skilful sub-seasonal rainfall forecasts with intervals from weeks to months into the future may help mitigate the impacts of extreme weather conditions on lives, infrastructure, and socioeconomic activities. Therefore, it is essential to assess the skill of sub-seasonal ensemble prediction systems and explore sources of predictability for improving scientific understanding and driving model developments. As part of the African Science for Weather Information and Forecasting Techniques (African SWIFT) project, we evaluate the quality of weekly rainfall forecasts across Africa using hindcasts from three state-of-the-art operational models (ECMWF, NCEP, UKMO) and a set of verification metrics. In particular, we also examine the models’ ability to represent the leading modes of weekly rainfall variability during an important rainy season in Africa, i.e., the East African short rains from October to December. Lastly, we investigate forecast skill by assessing how well models capture the connections between climate drivers and African rainfall variability, along with analyzing how this affects the skill of forecasts. Our results can improve our knowledge regarding the strengths and weaknesses of modeling rainfall variability patterns in East Africa, in addition to assisting forecasters in identifying the African regions and forecast horizons with actionable sub-seasonal forecast skill, as well as supporting better interpreting regime-dependent skill.
Co-authors:
Linda Hirons (University of Reading-UK)
Matthew Young (Oldbaum Services-UK)
David Macleod (University of Bristol-UK)
Emily Black (University of Reading-UK)
Steve Woolnough (University of Reading-UK)
