Severe weather in hilly regions like the Eastern Himalayas poses significant risks to life and natural resources due to complex and varied topography within small geographic areas. This study explores the integration of Dual-Polarization Weather Radar (DWR) data into the Weather Research and Forecasting (WRF) model to evaluate the performance of two data assimilation techniques (3DVar and 4DVar) in predicting heavy rainfall events. Reflectivity and radial velocity measurements from an X-Band DWR were assimilated to quantitatively assess their impact on precipitation forecasts. Simulations were conducted for two heavy rainfall events in August 2023 at spatial resolutions of 9-km and 3-km, each over three days. Forecast accuracy was measured using Root Mean Square Error (RMSE) and RSR, where lower values indicate better performance. Results show that decreasing grid spacing from 9-km to 3-km improves prediction accuracy. RMSE for RH at 9 km_CTRL is 8.09 % and for 3 km_CTRL is 7.10 %. Similarly, for T2 at 9 km_CTRL, RMSE is 3.38oC, and for 3 km_CTRL is 2.71oC. RMSE for precipitation at 9 km_CTRL is 20.80 mm and for 3 km_CTRL is 12.78 mm. The lowest RMSE is observed for the 4DVar experiment for all three variables, where RMSE is 8.50 %, 2.54oC, and 2.66 mm for RH, T2 and precipitation. RSR value is lowest for the 3 km_4DVar experiment as compared to other experiments for all the variables. These findings highlight that integrating radar data via 4DVar assimilation markedly improves heavy rainfall forecasting, especially at higher resolutions. This approach holds strong potential to support disaster management and mitigation efforts in the vulnerable Himalayan region.