Coupled Land–Atmosphere Data Assimilation in the NOAA Operational Weather Prediction Models—Rapid Refresh (RAP) and High-Resolution Rapid Refresh (HRRR)

The operational weather prediction models Rapid Refresh (RAP) and High-Resolution Rapid Refresh (HRRR) are frequently updating systems assimilating all available observations hourly. Not well-observed on the regional scales, soil and snow states are cycled within these systems, constrained by hourly assimilation of various atmospheric observations including satellite cloud and radar reflectivity data. Surface observations of 2-m temperature and dewpoint are also assimilated to affect both the atmospheric boundary layer but also the land state (soil moisture, soil temperature) using a simple loose atmosphere-land coupling within Global Statistical Interpolation (GSI) analysis system. Snow temperature is also modified in GSI using the same method as soil temperature, while snow cover area is modified daily from the NESDIS IMS snow data at 4-km horizontal resolution. Overall, analysis increments of lowest-level atmospheric temperature and water vapor are used, using estimated coupling between atmospheric and soil errors under certain conditions, to diagnose changes to soil moisture and temperature.

The developed method retains the soil/snow–air temperature difference from the forecast background field to the analyzed state. Under the assumption that near-surface analysis increments for temperature and moisture of opposite sign (warm and dry, or alternatively, cold and moist) may be related to soil moisture errors under conditions of strong coupling during the daytime with no precipitation, a technique analogous to that for soil temperature has been also developed to modify soil moisture. This technique is applied only in daytime with no precipitation in the background forecast and with no snow on the ground. Figure.1 presents an example of coupled atmosphere/soil data assimilation in Rapid Refresh (RAP).

Coupled land-atmosphere data assimilation prevents from drifts in the land state caused by model biases. Its effect on the RAP and HRRR surface conditions will be evaluated and presented at the meeting.