A multivariate approach for persistence-based drought prediction: Application to the 2010-2011 East Africa drought

- Improves drought prediction using a multi-index integrated approach.
- Provides probability of drought occurrence for different lead times.
- Extends current persistence-based drought prediction to multiple variables.

SummaryThe 2011 East Africa drought caused dire situations across several countries and led to a widespread and costly famine in the region. Numerous dynamic and statistical drought prediction models have been used for providing drought information and/or early warning. The concept of Ensemble Streamflow Prediction (ESP) has been successfully applied to univariate drought indicators (e.g., the Standardized Precipitation Index) for seasonal drought prediction. In this study, we outline a framework for using the ESP concept for multivariate, multi-index drought prediction. We employ the recently developed Multivariate Standardized Drought Index (MSDI), which integrates precipitation and soil moisture for describing drought. In this approach, the ESP concept is first used to predict the seasonal changes to precipitation and soil moisture. Then, the MSDI is estimated based on the joint probability of the predicted accumulated precipitation and soil moisture as composite (multi-index) drought information. Given its probabilistic nature, the presented model offers both a measure of drought severity and probability of drought occurrence. The suggested model is tested for part of the 2011 East Africa drought using monthly precipitation and soil moisture data obtained from the NASA Modern-Era Retrospective Analysis for Research and Applications (MERRA-Land). The results indicate that the suggested multi-index predictions are consistent with the observation. Furthermore, the results emphasize the potential application of the model for probabilistic drought early warning in East Africa.