Advancing Drought Prediction Capabilities
The overarching goals of drought prediction research have been to improve our understanding of physical mechanisms of drought, sources of predictability, and the nature and magnitude of unpredictable variability (i.e., noise). Goals also include improving operational drought prediction skill through the full utilization of sources of predictability and the development of improved models, and the observations and data assimilation systems needed to initialize and validate the models. Specifically, research seeks to better understand the physical mechanisms and advance the ability to predict various aspects of drought, including its onset, duration, severity, and recovery. To facilitate progress toward these objectives, the MAPP DTF developed a research framework and the drought capability assessment protocol (Wood et al. 2015), which proposes performance metrics, test cases, and verification datasets to guide individual researchers in testing and evaluating their methods and ideas against the operational or state-of-the-art capabilities. The framework, as originally developed, focuses on the analysis of four major historical drought events over North America to standardize evaluations over particular reference periods. To provide a more general evaluation of prediction skills, the DTF has also embraced the North American multimodel ensemble (NMME) seasonal prediction protocol for evaluation of capabilities over a standard 30-year (1981-2010) period (Kirtman et al. 2014).
Section 6.3.1 describes the current drought prediction capability and prediction research advances, and Section 6.3.2 highlights key research results related to drought mechanisms and predictability.