Articles | Volume 34, issue 2
Regular paper
09 Feb 2016
Regular paper |  | 09 Feb 2016

Impact of variational assimilation using multivariate background error covariances on the simulation of monsoon depressions over India

M. Dhanya and A. Chandrasekar

Abstract. The background error covariance structure influences a variational data assimilation system immensely. The simulation of a weather phenomenon like monsoon depression can hence be influenced by the background correlation information used in the analysis formulation. The Weather Research and Forecasting Model Data assimilation (WRFDA) system includes an option for formulating multivariate background correlations for its three-dimensional variational (3DVar) system (cv6 option). The impact of using such a formulation in the simulation of three monsoon depressions over India is investigated in this study. Analysis and forecast fields generated using this option are compared with those obtained using the default formulation for regional background error correlations (cv5) in WRFDA and with a base run without any assimilation. The model rainfall forecasts are compared with rainfall observations from the Tropical Rainfall Measurement Mission (TRMM) and the other model forecast fields are compared with a high-resolution analysis as well as with European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim reanalysis. The results of the study indicate that inclusion of additional correlation information in background error statistics has a moderate impact on the vertical profiles of relative humidity, moisture convergence, horizontal divergence and the temperature structure at the depression centre at the analysis time of the cv5/cv6 sensitivity experiments. Moderate improvements are seen in two of the three depressions investigated in this study. An improved thermodynamic and moisture structure at the initial time is expected to provide for improved rainfall simulation. The results of the study indicate that the skill scores of accumulated rainfall are somewhat better for the cv6 option as compared to the cv5 option for at least two of the three depression cases studied, especially at the higher threshold levels. Considering the importance of utilising improved flow-dependent correlation structures for efficient data assimilation, the need for more studies on the impact of background error covariances is obvious.

Short summary
The three-dimensional variational technique (3DVar) is a popular technique used for data assimilation. Background error covariance (BEC) influences the performance of 3DVar technique. In this study, two formulations of BEC, namely the cv5 and cv6 options in the 3DVar assimilation in the Weather Research and Forecasting (WRF) model, are compared. It is found that the formulation of BEC impacts the analysis. Utilising the cv6 option moderately improves the simulation of monsoon depressions.