Abstract
Extreme precipitation events over the Asian monsoon region have profound societal and ecological impacts; however, accurate simulations of these events remain a major challenge in current climate modeling. Increasing horizontal resolution in atmospheric general circulation models (AGCMs) has emerged as a promising pathway for enhancing model fidelity in capturing small-scale convective processes and orographic effects. In this study, we assess the influence of resolution refinement on extreme precipitation simulations using two advanced AGCMs: the finite-volume atmospheric model of the IAP LASG (FAMIL), a component of the Chinese FGOALS-f climate system model, and the non-hydrostatic icosahedral atmospheric model (NICAM16), which was developed in Japan. Each model performs three AMIP simulations at various horizontal resolutions –100, 25, and 12.5 km for FGOALS-f and 56, 28, and 14 km for NICAM16. The analysis focuses on climatological precipitation patterns, frequency distributions of extreme rainfall, and three diagnostic indices–consecutive wet days (CWDs), total precipitation amount (TPA), and daily precipitation intensity (DPI)–with a particular focus on the Asian summer monsoon region and the mainland of China. Our findings indicate that increasing the resolution markedly improves the simulations of the spatial structure and intensity distribution of extreme precipitation. The FGOALS-f model substantially reduces the long-standing positive rainfall bias over the western Pacific warm pool as the resolution increases from 100 to 12.5 km and shows an improved ability to simulate events in the 300–500 mm d–1 range. In contrast, NICAM16 tends to overestimate extreme precipitation at its finest resolution (10 km), especially over the Indo-Pacific and Chinese subregions, suggesting strong sensitivity to the model’s non-hydrostatic core and physical parameterizations. Over China, both models capture the spatial pattern of persistent heavy rainfall, but the 12.5-km version of FGOALS-f consistently demonstrates greater skill across all three indices. This includes better spatial correlation and lower root-mean-square errors than coarser configurations. NICAM16 shows improvements by increasing the resolution across all three extreme precipitation indices, with the 25-km version exhibiting the best overall performance, surpassing even the 10-km version. These discrepancies are likely linked to model dynamics: NICAM16’s non-hydrostatic dynamics yield stronger vertical velocities and higher low-level moisture convergence, amplifying extreme rainfall intensity, whereas FGOALS-f, which is based on a hydrostatic framework, presents more conservative estimates. These findings indicate that increasing the model horizontal resolution is an effective approach for improving the long-term simulation and projection accuracy of extreme precipitation in the Chinese region. Notably, NICAM16 tends to overestimate the intensity of extreme precipitation, whereas FGOALS-f tends to underestimate it, especially at high resolution. This contrast highlights the need for continued development of resolution-aware physical schemes tailored to different model dynamics, which is crucial for improving the ability to simulate extreme rainfall events. This study provides a comprehensive evaluation of how horizontal resolution affects extreme precipitation simulations in FGOALS-f and NICAM16. Although increased resolution improves model performance, both FGOALS-f and NICAM16 still exhibit biases in the context of simulating climatological means and extreme precipitation, indicating that refined physics is essential for credible projections. Future efforts should prioritize the development of resolution-aware physical schemes that are suitable for the convective “grey zone” emerging at kilometer-scale resolutions. Additionally, improving the simulation of weather-to-subseasonal processes–particularly tropical cyclones–is vital, as typhoon track and intensity directly influence extreme rainfall over the western Pacific and East Asia. An enhanced representation of these systems and their interactions with monsoonal circulations will benefit regional prediction efforts. Finally, moving beyond atmosphere-only simulations by employing high-resolution coupled ocean-atmosphere models is crucial for capturing key processes such as the Madden-Julian oscillation and tropical cyclone genesis. The incorporation of air-sea interactions will significantly improve the simulation and forecasting of extreme precipitation in a changing climate. In summary, the findings of this study provide empirical support for the necessity of high-resolution climate modeling in the upcoming CMIP7 framework. This study also provides practical insights for improving extreme precipitation projections and risk assessments in monsoon-affected regions.

This publication has 30 references indexed in Scilit: