Inclusion of Model Error in the 3DVAR Radar Assimilation system using Ensemble based Background Error Covariance.

The Indian subcontinent experiences an average of three to six floods every year due to extreme rainfall events. To warn the public about the occurrence of such intense rainfall events beforehand, a reliable weather forecasting system is essential. Generally, Numerical Weather Prediction (NWP) models are used to predict the current and future state of the atmosphere. However, these NWP models become unreliable when it comes to forecasting short duration heavy convective rainfall events. One of the major reason for such forecast skill reduction is the improper representation of convective storm structure in the initial condition supplied to the NWP model. Assimilation of high-resolution Doppler Weather Radar (DWR) through variational method is one of the most commonly used methods for representing storm structure in the initial conditions. However, these variational radar assimilation methods are sensitive to the error in the forecast model known as Background Error Statistics (BES) [Thiruvengadam et al., 2019]. The BES plays a major role in a variational radar assimilation system by spreading the assimilated DWR observations to nearby grid points and correlated variables. Generally, the methods used to model the background error statistics (BES) in variational systems often fail to represent the model error in BES. In this study, we have proposed a new ensemble-based method using Stochastically Perturbed Parameterization Tendency to represent the model error in the BES. In addition to that, we have also analyzed the effect of different control variable options in improving the structure of BES. The performance of the proposed ensemble-based BES in improving short term precipitation forecasts is further tested for an extreme rainfall event occurred over Chennai city in December 2015. Results demonstrate that the use of ensemble-based BES improves the precipitation forecast skill in terms of both position and intensity than traditional National Meteorological Centre based BES.

Reference:

Thiruvengadam, P., J. Indu, and S. Ghosh (2019), Assimilation of Doppler Weather Radar data with a regional WRF-3DVAR system: Impact of control variables on forecasts of a heavy rainfall case, Adv. Water Resour., 126, doi:10.1016/j.advwatres.2019.02.004.