Assessing the subseasonal forecasting skill of extreme agrometeorologically relevant dry spells over Southern Africa

Abstract

Compared to prolonged seasonal droughts, impactful within-season agrometeorologically relevant dry spells of a relatively shorter duration that lie within the subseasonal-to-seasonal (S2S) timescale can have significant negative impacts on agriculture and pose severe threats to food security, particularly in agro-based economies such as those in Southern Africa. This study assesses the skill of the ECMWF ensemble subseasonal forecasting system in predicting these impactful events during the austral summer maize growing season (October to March, ONDJFM). We utilise two definitions of agrometeorologically relevant dry spells: (1) a generic index based on accumulated daily precipitation falling below optimal thresholds for the maize crop across the broader Southern Africa domain, and (2) case-specific events where a crop yield proxy derived over Zimbabwe’s primary maize-growing region falls below critical thresholds. Our findings reveal that predictive skill for the generic index declines with longer lead times but remains higher during the OND sub-season compared to JFM. Furthermore, the ECMWF subseasonal forecasting system demonstrates predictive skill for these events with a 10-to-30-day lead-time, particularly in northern Zimbabwe, central Zambia, Malawi, and northern Mozambique. For the case-specific events, findings reveal that prediction accuracy is conditioned by the model’s ability to simulate key atmospheric circulation patterns that modulate such extreme events. Overall, these results underscore the potential of the ECMWF subseasonal forecasting system to improve drought early warning systems and support anticipatory action initiatives that are still in their infancy in Southern Africa.