Thresholds for operational agro-climatic monitoring and early warning against high impact rainfall events in the Sudan-Sahel region, West Africa.

Abstract

High impact rainfall events (HIRE) are among the most challenging intra-seasonal climate variability components which threaten human security and natural resources in the West African Sudan-Sahel region (WASS). The exposure and vulnerability of rural communities and farming systems to random onset of rainy seasons, long dry spells, heavy rainfall events, droughts and floods can subsequently increase food insecurity, disasters risks on life and property. The identification and use of thresholds can improve the provision of weather/climate information to people and smallholder farming systems in order to alleviate food crisis and reduce disaster risks in WASS. In-situ observations data (weather, maize cultivars and soil datasets), collected from some reference stations, are combined with crop model simulations data (DSSATV4.6, www.dssat.net), to generate dates of occurrence and amplitudes of first efficient rainfall (FER), extreme dry spells (ExDS), intense rainfall event (IRE) and water requirement satisfaction index (WRSI). The threshold values defining these agro-climatic HIRE as rainfall extremes are identified and analysed, at the station level and upscaled to the WASS level, with respect to observed dry (wet) regime of the cropping seasons. The thresholds’ operational rating scales and warning flag colours are suggested for both crop-climate related indices (i.e. FER, ExDS, WRSI) and the disaster reduction related indices (i.e. IRE). Further predictability potentials, at 10-day (dekad) lead time, are investigated for WRSI, using a binary logistic regression (BLR) model developed based on observed candidate predictors and tested using prefect prognostics (PP) forecasting approach. Forecast verification indices show an uneven performance of the PP approach, in predicting WRSI extremes, across reference stations with high probability of detection and bias. From these results, the study demonstrates that thresholds profiling can improve the quality of agro-meteorological information delivery to operational maize monitoring and early warning services against rainfall extremes in the fields of disaster risk reduction and food security in this region.