Variational Assimilation of VAS Data into a Mesoscale Model; Assimilation Method and Sensitivity Experiments

Abstract

A variational method has been developed to assimilate VAS temperature and moisture gradient information into a mesoscale model. A series of experiments were conducted to test the sensitivity of both adiabatic and diabatic versions of the model to VAS data assimilations for the 20?21 July 1981 case. The VAS data for this case are compared to the rawinsonde data and VAS moisture imagery. The retrieved VAS temperature fields captured the asynoptic development of strong mesoscale temperature gradients although the VAS relative humidity fields were generally too smooth. The synoptic-scale effects of the assimilation of VAS data were negligible. The greatest impact was on the mesoscale forecasts of the patterns of convective instability. The assimilation of the strong VAS temperature gradients resulted in the short-term forecast of greater convective instabilities across Oklahoma, where observed convection subsequently developed. The additional assimilation of relative humidity gradients did not significantly change the patterns of the forecast instabilities. Increasing the number of successive assimilations improved the subsequent forecasts of convective instability. For this case, the greatest improvements from assimilation resulted from the resolution of the strong mesoscale temperature gradients by the asynoptic VAS data. The assimilation of this structure into the model resulted in forecasts of convective instability and precipitation more closely resembling the patterns of the observed convection.