LAUSR

Abstract The background error covariance is a vital component in data assimilation systems, which is usually simplified with the control variable preconditioning procedure. This study aims to assess the impact of the determination of the momentum control variables using ψ and χ (CV_ψχ) or direct x and y components of wind(CV_UV)on the analysis and forecast of a typhoon case for Doppler radar radial velocity (Vr) data assimilation. It is shown that using direct U-wind and V -wind as the control variables is able to reflect the observed wind itself more objectively. Smaller scale increment is found with the use of CV_UV, which is more physical compared to CV_ψχ. Generally, CV_UV yields smaller wind analysis error than CV_ψχ. The vertical structure from CV_UV are more accurate than that from CV_ψχ with enhanced vortex circulation and warm core. The subsequent composite reflectivity, track and intensity forecast using the CV_UV is improved over that using CV_ψχ, due to more accurate wind field that lead to correct convergence for the typhoon system. Sensitivity experiments of the temporal frequency of DA under the determination of the momentum control variable pairs are also conducted. The results show that there are also no clear, positive impacts of increasing the DA frequency on the forecasts especially for the CV_UV experiment.