Performance of 4D-Var Data Assimilation on Extreme Snowfall Forecasts over the Western Himalaya Using WRF Model

The accurate predictions of extreme precipitation/snowfall events are very helpful in identifying the severe avalanche/landslide prone hazard areas in advance over high mountainous regions. The Weather Research and Forecasting model (WRF) version 3.9 has been used to investigate the performance of Four-Dimensional Variational data assimilation (4D-Var) on Three-Dimensional Variational data assimilation (3D-Var) by considering two extreme snowfall events (23–26 January 2017 and 05–08 February 2019) over the Western Himalaya (WH). The result shows that the 4D-Var performed better than the 3D-Var for both the events by analyzing domain-averaged error and sensitivity parameter analysis. The initial state model variable’s domain-averaged error analysis revealed that 4D-Var has great potential to improve the initial conditions than the 3D-Var from lower to the upper atmosphere. Sensitivity parameter analysis also supports 4D-Var has more sensitive than the 3D-var especially in the lower and upper atmosphere by changing temperature and moisture fields along with winds circulations. From statistical skill scores analysis, 4D-Var performed well to reproduce the extreme snowfall events than the 3D-Var over WH.