Soil moisture tendencies into the next century for the conterminous United States

A monthly snowpack and soil‐moisture‐accounting model is formulated for application to each of the climate divisions of the conterminous United States for use in climate impacts assessment studies. Statistical downscaling and bias adjustment components complement the model for the assimilation of large‐scale global climate model precipitation and temperature. The model produces monthly streamflow that is broadly consistent with observed streamflow from several drainage basins in the United States for the period 1931–1998. Simulated historical soil moisture fields reproduce several features of the available observed soil moisture in the Midwest. The simulations produce large‐scale coherent seasonal patterns of soil moisture field moments over the conterminous United States, with high soil moisture means over divisions in the Ohio Valley, the northeastern United States, and the Pacific Northwest, and with pronounced low means in most of the western U.S. climate divisions. Characteristically low field standard deviations are produced for the Ohio Valley and northeastern United States, the Pacific Northwest in winter, and the southwestern United States in summer. Differences in extreme standardized anomalies of soil moisture over the historical record possess high values (2.5–3) in the central United States, where the available water capacity of the soils is high. Application of the methodology for future periods using output from the Canadian coupled global climate model (CGCM1) shows that for at least the first few decades of the 21st century, somewhat drier‐than‐present soil conditions are projected, with highest drying trends found in the southeastern United States. The soil moisture deficits in most areas are of the same order of magnitude as the soil moisture field standard deviations arising from historical natural variability. Brumbelow and Georgakakos [this issue] study the implications for crop yield in the United States.