Sensitivity of heatwave simulation to radiation parameterization in WRF and MPAS-A: A case study over Bangladesh

Heatwaves represent one of the most impactful categories of extreme climate events and remain difficult to simulate accurately in numerical weather prediction and regional climate models. In tropical regions like Bangladesh, where strong monsoonal circulations, heterogeneous land-use patterns, and sparse in situ observations limit constraint of model physics and thus remain constant challenge for heatwave representation. This study evaluates the performance of the Weather Research and Forecasting (WRF) model and the Model for Prediction Across Scales-Atmosphere (MPAS-A) in reproducing a documented heatwave event during 26 April - 3 May 2024, to identify the modeling configuration that more reliably represents near-surface thermodynamic conditions. Model-simulated 2-m air temperature (t2) and 2-m specific humidity (q2) were evaluated against the MERRA reference dataset (0.5° × 0.625° spatial resolution) using root mean square error (RMSE) and Pearson correlation coefficients. Both models employed an identical suite of physical parameterizations, including WSM-6 microphysics, the Kain-Fritsch cumulus scheme, the Yonsei University planetary boundary layer scheme, MM5 surface layer physics, and the Noah land surface model, while radiative transfer was represented using the Rapid Radiative Transfer Model for Global Climate Models (RRTMG) and the Community Atmosphere Model (CAM) schemes. WRF was configured with two nested domains at 27 km and 9 km spatial resolution, whereas MPAS-A employed a variable-resolution mesh refined from 46 km globally to 12 km over the study region. Results indicate that WRF with RRTMG achieved the highest skill score in simulating 2-m air temperature (RMSE = 2.52 °C; r = 0.95), outperforming MPAS-A configured with CAM (RMSE = 3.04 °C; r = 0.82). For 2-m specific humidity, WRF-RRTMG minimized overall error (RMSE = 0.003), while WRF-CAM exhibited the strongest temporal correlation (r = 0.899); within the MPAS-A framework, the RRTMG configuration consistently outperformed CAM. Moreover, WRF-RRTMG more accurately captured the timing of heatwave onset, showing smaller temporal displacement relative to the reference dataset than MPAS-A configurations, indicating improved representation of the initiation phase of extreme heat events. Overall, the findings demonstrate that WRF provides more accurate heatwave simulation over Bangladesh under the adopted configuration, while MPAS-A shows competitive performance when configured with radiation transfer schemes, supporting its potential utility for multiscale atmospheric modeling applications.