Probabilistic Subseasonal-to-Seasonal Forecasts of Drought in the United States

Drought is among the costliest natural disasters to affect the United States (U.S.). Accurate subseasonal-to-seasonal (S2S) drought forecasts are essential for mitigating drought-related impacts and costs. To address evolving user needs, NOAA Climate Prediction Center (CPC) has been conducting research and development to improve its drought forecast products and capability. Here, we discuss the challenges in S2S drought forecasts and our methodologies to address them within the context of our ongoing development of CPC probabilistic S2S drought outlooks. Specifically, the limited skill of dynamical forecast systems in predicting precipitation and temperature - the key meteorological drivers for drought evolution (e.g., flash drought development) - is improved by applying statistical post-processing and developing and incorporating statistical prediction models. To improve land surface forecasts, offline land surface modeling approach is employed to use more accurate land initial conditions, post-processed meteorological forecasts, and more advanced land surface models. To predict drought as a multivariate and multifaceted phenomenon, forecasts for indices representing meteorological, agricultural and hydrological drought conditions are integrated through objective drought blends. Here to optimize the forecast skill of the integrated drought conditions, objective blends are developed separately for short-term and long-term drought conditions, where the weights of input short-term drought indices are determined based on their hindcast skills, and the weights of input long-term drought indices are determined based on their Fractional Information (FRACI) obtained from a Mutual Information (MI) analysis against the U.S. Drought Monitor (USDM). The skill performance of the drought outlooks is then investigated by assessing the extent to which known sources of S2S drought predictability, such as recurring modes of S2S climate variability (e.g., ENSO), are captured. Finally, to overcome user challenges in understanding probabilistic drought forecasts, social science is incorporated in finalizing the visualization of our probabilistic drought outlook products. The applications of CPC drought outlooks to address regional and sectoral needs are achieved through engagement with NIDIS and its regional partners, focusing on their key sectors of interest (e.g., agriculture, water resource management).