From seasonal climate forecasts to a crop yield early warning system for rainfed agriculture

Seasonal climate forecasts can support anticipatory decision-making in rainfed agricultural systems. However, their translation into actionable information remains limited. This is especially the case in Sub-Saharan Africa, where food systems are highly exposed to climate variability and depend on timely information for risk management. Here, we develop a seasonal crop yield forecasting system that translates ensemble climate predictions into decision-relevant indicators for food-system planning.

The system combines bias-corrected and downscaled SEAS5 seasonal forecasts with the LPJmL process-based crop model to generate ensemble yield forecasts across multiple crops and agro-climatic regions in Ghana, Kenya, and Zimbabwe. We evaluate forecast performance over a 30-year hindcast period using probabilistic verification metrics to assess reliability and event discrimination relative to climatology.

Results show that forecast skill varies substantially across crops, regions, and lead times. While continuous probabilistic skill is often limited, forecasts retain discriminatory ability in several crop–region combinations, indicating potential for early warning of adverse yields. This highlights the importance of matching forecast products with decision contexts.

To bridge the gap between forecast generation and use, we implement the forecasting system within an information service (interactive dashboard) that delivers spatially explicit yield forecasts, probabilistic information, and comparisons to climatological baselines. This demonstrator illustrates how ensemble-based yield forecasts can be operationalised and communicated to support anticipatory actions in food systems.

As seasonal climate forecasts continue to improve in accuracy and resolution, and process-based crop models advance to capture more complex representations of crop physiology and phenology, the potential for more actionable forecasts increases. In this context, contextualising forecasts within decision-making frameworks becomes critical to realise their uptake and impact for food security in climate-sensitive regions.