Improvements in the prediction of extreme rainfall events with nested high-resolution rapid refresh modelling system over the Indian Himalayan region

This study investigates the advantages of employing a nested High-Resolution Rapid Refresh (HRRR) model for simulating highly localized heavy rainfall events over the Indian Himalayan region, characterized by a complex terrain. Accurate weather prediction in such a region necessitates a model configuration that resolves the underlying terrain features, incorporates optimal physics combinations, and assimilates the most precise atmospheric state. A nested HRRR system with 5 km and 1 km resolutions was developed, incorporating all available quality observations through an hourly assimilation cycle. Three rainfall episodes from July–August 2023 were analyzed, focusing on spatial rainfall, diurnal variations, dynamic, thermodynamic, and microphysical properties. This nested HRRR system could effectively capture the rainfall intensity and location, as it is better tuned to the realistic atmospheric conditions through frequent assimilation of observations, and thereby, the dynamics and thermodynamics are modified. However, a lead or lag of 1–2 h is observed in the diurnal rainfall variation in both domains. The quantitative model evaluation for rainfall, using various statistical skill scores, demonstrates the better performance of the 1 km domain, emphasizing the impact of higher resolution and frequent updates to initial conditions on the high-impact weather simulation accuracy. The spatial verification with the Contiguous Rain Area (CRA) analysis method reveals that pattern errors dominate over displacement errors and highlights improved spatial accuracy. Overall, the comparison between the 5 km and 1 km domains underscores the importance of high-resolution models, combined with frequent updating of initial conditions, for accurately predicting highly localized, high-impact rainfall events over the Indian Himalayan region.