Global Meteorological Drought Prediction Using the North American Multi-Model Ensemble

Abstract

Abstract Precipitation forecasts from six climate models in the North American Multi-Model Ensemble (NMME) are combined with observed precipitation data to generate forecasts of the standardized precipitation index (SPI) for global land areas, and their skill was evaluated over the period 1982–2010. The skill of monthly precipitation forecasts from the NMME is also assessed. The value-added utility in using the NMME models to predict the SPI is identified by comparing the skill of its forecasts with a baseline skill based solely on the inherent persistence characteristics of the SPI itself. As expected, skill of the NMME-generated SPI forecasts depends on the season, location, and specific index considered (the 3- and 6-month SPI were evaluated). In virtually all locations and seasons, statistically significant skill is found at lead times of 1–2 months, although the skill comes largely from initial conditions. Added skill from the NMME is primarily in regions exhibiting El Niño–Southern Oscillation (ENSO) teleconnections. Knowledge of the initial drought state is critical in SPI prediction, and there are considerable differences in observed SPI values between different datasets. Root-mean-square differences between datasets can exceed typical thresholds for drought, particularly in the tropics. This is particularly problematic for precipitation products available in near–real time. Thus, in the near term, the largest advances in the global prediction of meteorological drought are obtainable from improvements in near-real-time precipitation observations for the globe. In the longer term, improvements in precipitation forecast skill from dynamical models will be essential in this effort.