A188-0005
Boreal summer subseasonal predictability of rainfall and monsoon onset over Senegal

Tuesday, 15 December 2020
Poster
Bohar Singh1,2, Andrew William Robertson2, Sylwia Trzaska3, Ousmane Ndiaye4 and Oumar Konté4, (1)Columbia University of New York, International Research Institute for Climate and Society, Columbia University, Palisades, NY, United States, (2)Columbia University, International Research Institute for Climate and Society, Palisades, NY, United States, (3)Center for International Earth Science Information Network, Palisades, NY, United States, (4)Agence Nationale de l'Aviation Civile et de la Météorologie (ANACIM), Dakar, Senegal
Abstract:
Rainfall over Senegal is concentrated in a unimodal season, from June to September, starting in early June in the southeast part of the country then slowly progressing towards northwest. Similarly to the seasonal rainfall amounts, rainfall onset presents large interannual variability and the season can be shortened by as much as one month. Thus, reliable information on when the rains will start is one of the foremost demands in climate forecast, especially from the agricultural sector, as it is critical for planning agricultural and pastoral activities, such as preparing the fields and securing appropriate seeds and fertilizer, or starting the migration of herds towards new pastures and water points

This study analyzes the predictability of monsoon onset over Senegal using subseasonal model hindcast suites from the S2S and SubX databases, and observed rainfall estimates from the Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) data. Local scale rainfall onset and demise dates are defined by precipitation and dry spell thresholds (to avoid false onsets), using both the Agence Nationale de l'Aviation Civile et de la Météorologie (ANACIM) operational definition, and a “regime shift” definition, which depends on precipitation accumulation before onset/after demise. Both definitions yield similar results in terms of onset dates and their interannual variability.

Onset/demise forecasts are based on a hybrid method that combines observations and calibrated subseasonal model forecasts by appending the model predictions to rainfall observed prior to the forecast initialization date, and evaluating the hybrid past/future rainfall against onset/demise criteria. Forecast skill is assessed using both deterministic and probabilistic metrics, and as a function of both the lead time and spatial scale.