Premium
Performance of State‐of‐the‐Art C3S European Seasonal Climate Forecast Models for Mean and Extreme Precipitation Over Africa
Author(s) -
Gebrechorkos S. H.,
Pan M.,
Beck H. E.,
Sheffield J.
Publication year - 2022
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2021wr031480
Subject(s) - precipitation , climatology , environmental science , anomaly (physics) , forecast skill , quantitative precipitation forecast , climate model , seasonality , meteorology , lead time , ensemble average , climate change , geography , statistics , mathematics , ecology , physics , condensed matter physics , marketing , biology , geology , business
Seasonal hydrological forecasts at high spatial and temporal resolution can help manage water resources and mitigate impacts of extreme events but are dependent on skillful and operational seasonal forecasts from climate models. In this study, we evaluate precipitation forecasts from five operational climate models with a potential to drive hydrological forecasts: European Centre for Medium‐Range Weather Forecasts (ECMWF), UK Met Office (UK‐Met), Météo France, Deutscher Wetterdienst, and Centro Euro‐Mediterraneo sui Cambiamenti Climatici. The Multi‐Source Weighted‐Ensemble Precipitation is used as a reference data set to evaluate the model skill. The performance of individual models is evaluated on daily, weekly, monthly, seasonal, and climatological periods, and for selected target months, lead‐times and drought events, and compared to unweighted and skill‐weighted multi‐model ensemble mean forecast. For all models, the lead 1‐month forecast can replicate the climatological mean, monthly mean, and monthly anomaly precipitation, although much of this skill originates from the first week of the forecast. The skill drops rapidly for lead 2‐month and longer and is highest in drier regions and seasons. The forecast skill of monthly meteorological drought events at lead 1‐month is modest. All models represent the monthly variation in the length of wet and dry spell days at lead 1‐month, but the skill is weak for heavy and very heavy precipitation days. ECMWF is found to be the most skillful model, followed by the UK‐Met, although the multi‐model weighted average provides the highest performance compared to individual models and the un‐weighted multi‐model mean.