Impacts of 3DVAR Data Assimilation of Humidity and Scattered Rainfall Prediction over Central Nepal Himalaya

Accurate humidity prediction in weather models is crucial for reliably predicting weather phenomena like the development of storms, precipitation, etc. The Weather Research and Forecasting model (WRF), widely used in Nepal, has been systematically underpredicting near-surface humidity and rainfall over the Himalayan complex terrain of Nepal in recent years. As an assessment to correct the humidity prediction for a case of scattered precipitation over the Kathmandu Valley, we performed the WRF three-dimensional variational data assimilation (WRF-3DVAR) using the publicly available data of surface and upper air instruments. The data for the assimilation was available from the Research Data Archive of the National Center for Atmospheric Research. The WRF simulation domain consisted of a triply nested two-way interacting domain system centered in the Kathmandu Valley with the first, second, and third domain in 9 km × 9 km, 3 km × 3 km, and 1 km × 1 km horizontal grid resolution, respectively. In the series of assimilation experiments, the case of assimilating both surface and upper air data in all three domains resulted in the reduction of humidity bias from 40% to 20%. The assimilation could also capture the scattered rainfall over the Kathmandu Valley that was missing in the control simulation. However, the amount of rain was one-third compared with the observation. Since the amount of scattered rainfall during spring determines the quantity of the cash-crop production in Nepal, the improved prediction would help ensure agricultural/food security in the region and the economic stability of its low-income people. Assimilating a more structured observation in numerical weather prediction may significantly improve weather forecasting in Nepal.