Abstract

Droughts are a worldwide concern, thus assessment efforts are conducted by many centers around the world, mainly through simple drought indices, which usually neglect important hydrometeorological processes or require variables available only from complex land surface models (LSMs). The U.S. Climate Prediction Center (CPC) uses the Leaky Bucket (LB) water-balance model to postprocess temperature and precipitation, providing soil moisture (SM) anomalies to assess drought conditions. However, despite its crucial role in the water cycle, snowpack has been neglected by LB and most drought indices. Taking advantage of the high-quality snow water equivalent (SWE) data from The University of Arizona (UA), a single-layer snow scheme, forced by daily temperature and precipitation only, is developed for LB implementation and tested with two independent forcing datasets. Compared against the UA and SNOTEL SWE data over CONUS, LB outperforms a sophisticated LSM (Noah/NLDAS-2), with the median LB versus SNOTEL correlation (RMSE) about 40% (26%) higher (lower) than that from Noah/NLDAS-2, with only slight differences due to different forcing datasets. The changes in the temporal variability of SM due to the snowpack treatment lead to improved temporal and spatial distribution of drought conditions in the LB simulations compared to the reference U.S. Drought Monitor maps, highlighting the importance of snowpack inclusion in drought assessment. The simplicity but reasonable reliability of the LB with snowpack treatment makes it suitable for drought monitoring and forecasting in both snow-covered and snow-free areas, while only requiring precipitation and temperature data (markedly less than other water-balance-based indices).