The Use of Seasonal Performance Probability in Monitoring and Predicting Drought in Africa (Invited Presentation)

Drought is one of the greatest challenges in Africa due to its impacts on access to sanitary water and food. This paper presents the CPC developed Seasonal Performance Probability (SPP) to improve drought early warning in Africa. SPP is a statistical tool that quantitatively evaluates as early as within the first month of the rainfall season, the season-to-date performance of the seasonal rainfall and the probability for seasonal precipitation to finish at predefined varying percent of normal thresholds, less than 60% of normal rainfall, for example. Kernel Density Estimation (KDE) is employed to compute smoothed, continuous density functions based on more than 30 years of historical precipitation data from the Africa Rainfall Climatology Version 2 (ARC2) dataset. Various KDE parameterizations tests are performed to determine the optimal density estimates that produce the most statistically significant results. Verification results using Heidke Hit Proportion (HHP) scores for the period from 2006-2017 suggest that SPP provides reliable probabilistic drought outcome by early to mid-stages of the rainfall seasons. We apply SPP in the monitoring and the prediction of the southern Africa drought in 2015-16, which ranked as one of the most severe droughts in Africa in 30 years. According to the United Nations, by 2017, the Southern Africa Development Community (SADC) was still battling to recover from the severe drought, which affected about 41 million people in the region. The worst affected areas were the crop and pasture areas encompassing the southern two third of Mozambique and Malawi, southern Zambia, Zimbabwe, southern Angola, northern Botswana, and the eastern half of South Africa. The World Food program estimated a decline in crop production to 38% below the previous 5-year average. Southern Africa rainfall season spans the period from October to April. By the end of December 2015, SPP already projected that portions of eastern southern Africa were likely to finish the Dec-Feb rainfall season significantly below average. The detection of the development, monitoring, and forecasting of this drought through the analysis of SPP is presented. We further discuss the use of SPP in combination with other drought indicators such as SPI, NDVI, soil moisture and runoff, and the predictions of precipitation and temperature anomalies, to provide timely and reliable drought early warnings to humanitarian relief agencies.