Operational Seasonal Climate Forecast Methods and Skills over the Sahelian Region of West Africa

Abstract

In the context of increasing climate change and variability, reliable climate seasonal forecasts play critical role in agricultural planning. Recurring droughts, erratic rainfall, and limited adaptive capacity in the Sahel have highlighted the need for timely and usable climate information to enhance resilience among smallholder farmers. The West African Regional Climate Outlook Forums (WARCOF or PRESASS in French) forecasting system, predicts annually the rainy season key parameters (season cumulative rainfall, onset date, cessation date and length of dry spells). The WARCOF often shows significant percentage of forecast mismatches with observation. Thus, it urges to evaluate the performance and practical relevance of the WARCOF/PRESASS approach that has been used since 2011. The research aimed to assess the technical skill of the forecasting system, identify pathways to improve prediction accuracy, and examine how farmers access and utilize forecast information. Specifically, it was achieved through: first by the evaluation of the forecast skills for onset, cessation, and dry spells; then by the research on alternative predictors and statistical methods for forecast enhancement; and finally by the investigation on the dissemination and adoption of forecasts information among smallholders farmers. The guiding research question was: how can seasonal climate forecasts be improved and effectively communicated to strengthen smallholders’ adaptive strategies in West Africa? A mixed-methods approach was adopted. Quantitative forecast verification techniques (such as reliability diagrams, ROC curves, and Brier scores) were used to analyze historical forecasts against observed data from 1991 to 2023. The Climate Predictability Tool (CPT) was applied for statistical modelling using sea surface temperature (SST) anomalies, precipitation hindcasts, and newly tested predictors like April-May-June (AMJ) cumulativerainfall and wet-day frequencies. Additionally, qualitative data obtained from a baseline survey of 619 farmers and quantitative data from two-year demonstration trials in four municipalities of southwest Niger were used. Key findings indicate that in the WARCOF/PRESASS approach, onset forecasts demonstrated higher skill and reliability compared to cessation and dry spells, particularly when using precipitation hindcasts (e.g., NASA and CFSv2) as predictors. Incorporating wet-days and AMJ rainfall totals as alternative predictors significantly improved forecast skill. However, dry spell forecasts exhibited low discrimination ability. The field-level study revealed that only 42.3% of farmers had access to forecasts, and those who fully applied the information achieved measurable yield gains (up to 234 kg/ha) compared to traditional practices, validating the forecast's real-world utility. The positive impact of the forecast information on yield was more noticeable in drought prone areas (trials sites in Sahelian and Sudano-Sahelian zones) compared to wetter areas in the Sudanian zone; showing the relevance of climate information dissemination in the most vulnerable area. The study concludes that enhancing the PRESASS forecasting system requires scientific recalibration using high-performing predictors and user-oriented communication strategies. Co-production of forecasts, capacity building, and integration with local agricultural calendars are essential to maximize uptake and effectiveness. The findings contribute to climate resilience strategies and support SDGs 1 (No Poverty), 2 (Zero Hunger), and 13 (Climate Action), as well as WASCAL's priorities on sustainable agriculture and climate risk reduction.