This study compares the performance of four data assimilation (DA) systems: Ensemble Adjustment Kalman Filter (EAKF), Variational (3DVAR/4DVAR), and Hybrid ensemble-3DVAR (HYBRID) in the Weather Research and Forecast (WRF) model. A heavy rainfall event that produced notorious floods in the Uttarakhand over the Himalayan region is considered. Observations are assimilated at every 6 h interval and all the conventional observations including cloud tracked-wind from the satellite are used. The forecast initialized from the analysis of four DA systems at different lead times is evaluated. A non-cycled nested assimilation strategy that provides advantages of increased resolution in the DA system is tested. The results indicate that 4DVAR experiments produce more skillful forecasts for wind while both 4DVAR and EAKF experiments show improvement for upper tropospheric temperature forecasts as compared to the other experiments. The evaluation of rainfall forecast depicts that the 4DVAR DA system has outperformed the other DA systems when the effect of high-resolution assimilation is mimicked in the system using the nested assimilation strategy. Further analysis of the event indicates that an early merging of the southward protruding trough with the westward-moving monsoon depression has resulted in stronger southeastward flow in EAKF and HYBRID experiments, which is suggested as a potential reason for enhanced precipitation over the Uttarakhand in both the experiments.