Soil temperature and moisture errors in Eta model analyses

Forecast models require accurate soil temperature and soil moisture conditions to be able to properly partition the surface heat fluxes that drive the evolution of the planetary boundary layer. The National Centers for Environmental Prediction (NCEP) operational Eta model produces land surface analyses by continuously cycling soil temperature and moisture fields. In the past, these fields evolved only in response to radiation budget constraints and modeled precipitation, but NCEP recently upgraded the self-cycling process to assimilate observed precipitation. This study highlights potential problems with the land surface analysis from the Eta model by comparing 00 UTC and 12 UTC Eta model analyses of soil temperature and moisture at several depths with observations from the Oklahoma Mesonet. There are strong biases in soil temperature and there is a severe underestimation of soil moisture at all depths. There is notable improvement in the analyzed soil moisture fields after the change to a new assimilation scheme. While this change reduced the magnitude of the errors, a strong dry bias persists in the soil moisture field. A simple one-layer slab soil model reveals that these soil moisture errors alone may account for 1.0-1.5 degrees Celsius increases in maximum soil temperatures during the day and reductions in soil temperatures at night of 0.3-0.8 degrees Celsius. The remaining soil temperature errors likely stem from documented problems with the solar radiation and longwave parameterizations within the Eta model.