Tropical cyclones (TCs) simulated at gray-zone resolution (9km) in the Advanced Research version of the Weather Research and Forecasting Model (WRF-ARW) are performed and evaluated over the Western North Pacific (WNP). Two sets of experiments are carried out to test the effect the cumulus parameterization (CPS), driven by the European Center for Medium-Range Weather Forecasts (ECMWF) fifth-generation global atmospheric reanalysis (ERA5), over 11-year (2008-2018) TC seasons (June–November). Through comparisons with the observation and reanalysis, it is shown that the ICPS (involving CPS) experiment yields good skills in simulating TC frequency due to more realistic large-scale mean states, including stronger low-tropospheric circulation, wetter mid-tropospheric environment and more active ascending motion over TC main development region (MDR). On the other hand, the NICPS (not involving CPS) experiment is found to better reproduce intense TCs (Saffir-Simpson hurricane categories 3, 4 and 5) in terms of intensity, accumulated cyclone energy (ACE) and inner-core size. Although NICPS can capture TC inner-core structure well with stronger radial inflow in the boundary layer, rising motion around the eyewall and outflow in the mid to upper troposphere, NICPS significantly underestimates TC outer size due to rapidly decay of TC outer wind field, suggesting limited severe convective activities in the outer region.