Assimilation of Wind Speed and Direction with Real Observations

The assimilation of wind observations in the form of wind speed and direction (asm_sd) was firstly developed for Weather Research and Forecasting Data Assimilation system (WRFDA). To explore the benefit of asm_sd in real observation experiments, a set of full-cycle assimilation experiments for two-week period were carried out in August, 2013 over continental United States domain (CONUS). Meteorological Assimilation Data Ingest System (MADIS) dataset were used in this study. Corresponding to observation vector in asm_sd, the observation errors of spd and dir were estimated. The verifications presented are the differences of analyses and forecasts from radiosondes. Results demonstrated that the new assimilation methodology produces better analyses and forecasts of spd and dir than the standard methodology in the meaning of root-mean-square error, and it's also impressive of the dramatically decreased spd bias in analyses and forecasts generated by asm_sd. The improvements from asm_sd benefit partly from the quality control screening wind observations in a direct way, but mainly from the independent impacts of observation errors on assimilation variables, which is the essential distinction from asm_uv. A case developing during experiments was performed to present the indirect impacts of the new assimilation method on other variables except wind. The improved wind forecasts in asm_sd contribute to more accurate intensities and locations of wind shear line and warm and wet advection, which are the key factors in this case affecting the intensity and location of precipitation.