Improving the Simulations of Indian Extreme Precipitation Events through a Rotational Mixing Framework in the WRF Model

Accurate simulation of extreme precipitation remains a major challenge for weather prediction. Even in convection-permitting simulations at kilometer-scale resolution, biases such as underestimation of light rainfall frequency, overestimation of heavy rainfall events, and poor representation of localized convective extremes persist. These deficiencies are partly attributed to under-resolved lateral mixing between intense convective updrafts and their surrounding environment. Recently developed rotational mixing framework [1], which apply rotation to the convective-scale flow to enhance sub-grid-scale lateral mixing without artificially strengthening updrafts or downdrafts showed promising improvement for a month-long simulation. However, its applicability to moisture-rich convective environment of the tropics, particularly for short-duration, high-impact extreme rainfall events, has not yet been assessed. This study evaluates this rotational framework over India during monsoon in convection-permitting 4 km Weather Research and Forecasting (WRF) model simulations for five high-impact extreme rainfall events: Mumbai (July 2019), Kerala (August 2019), Delhi (July 2023), Gujarat (August 2024), and Andhra Pradesh (September 2024). We apply a 10° rotation (ROTMIX10) to the convective-scale flow to enhance sub-grid-scale lateral mixing while preserving the dynamical integrity of convective updrafts and downdrafts. We evaluate the model performance using the GPM precipitation and brightness temperature estimates. For each case, we compare the simulations using the ROTMIX10 framework against a standard CONTROL (no rotation) configuration. Across all cases, ROTMIX10 consistently improves the simulation of extreme precipitation, more accurately capturing rainfall intensities (RMSE reduced from 25.83 to 25.18) and enhancing spatial coherence of convective cores (correlation coefficient increasing from 0.42 to 0.51). Forecast skill also improves substantially, with the probability of detection rising from 0.20 to 0.37. We also find that these extreme events are primarily associated with mesoscale convective systems (MCSs), whose lifetimes, spatial extents, convective intensities, and accumulated rainfalls are more realistically represented in the ROTMIX10 simulations. Analysis of the underlying physical processes reveals that rotational mixing broadens the vertical velocity spectrum and enhances detrainment of moisture and condensate from strong updrafts between 4 and 12 km, moistening the mid-troposphere and increasing condensate loading in downdraft regions. This preconditions the atmosphere for subsequent convection, allowing rainfall to initiate under relatively drier conditions. Additionally, ROTMIX10 mitigates the typical bias of excessively cold cloud tops, yielding brightness temperature and reflectivity distributions closer to satellite observations. Overall, this work provides the first investigation of the performance of the rotational mixing framework for simulating high-impact convective extremes in the tropics. This approach demonstrates that rotational mixing framework enhances the physical realism of convective processes and improves extreme precipitation statistics in convection-permitting models.

 

[1] Hagos, S., Feng, Z., Varble, A. C., Tai, S. L., & Chen, J. (2025). The impacts of rotational mixing on the precipitation simulated by a convection permitting model. Journal of Advances in Modeling Earth Systems, 17(5), e2024MS004524.