Impact of background error statistics on 3D-Var assimilation: case study over the Indian region

Abstract

The quality of background error statistics (BES) is one of the key components for successful assimilation of observations in a numerical model. Considerable uncertainties and non-uniqueness exist, however, in prescribing BES; in particular, the prescription and impact of BES can also depend on the weather regime and not much is known in this regard over the Indian region. We have conducted a series of assimilation experiments using the WRF three-dimensional variational data assimilation (3D-Var) system with different BES to asses the relative improvement in model forecast due to different BES over the Indian region. The forecasted wind, temperature, and humidity are verified against NCEP analysis and conventional radiosondes, while the predicted rainfall is verified against Tropical Rainfall Measuring Mission (TRMM) observations. Using a number of parameters to quantify impact of BES, it is shown that the use of regional BES (RBES) in WRF 3D-Var significantly improves model forecast as compared to the control experiment (no assimilation) and Global BES (GBES). The use of RBES from National Meteorological Center (NMC) and ensemble perturbation (ENS) method in WRF 3D-Var produced similar impact on model forecasts except slight differences in wind speed. This study highlights the importance of domain-dependent and region-specific BES in WRF 3D-Var assimilation system; for the selected events, results obtained using RBES is found to be significantly better than GBES.