Reliable projections of tropical cyclone (TC) frequency, tracks, and intensity are essential for assessing climate risks and facilitating human adaptation to climate change. Systematic biases in general circulation models (GCMs) can be transferred into regional climate models (RCMs) via initial conditions (ICs) and lateral boundary conditions (LBCs), thereby lowering the reliability of dynamical downscaling simulations for TCs. To mitigate this issue, this study explores using a novel bias-corrected CMIP6 dataset generated using the Mean-Variance-Trend (MVT) bias correction method for dynamical downscaling, focusing on northwestern Pacific (NWP) TC simulations. We carried out three Weather Research and Forecasting (WRF) simulations over the Asian-Western Pacific region at a resolution of 25 kilometers during the period 1980-2014, to assess the improvement in modeling TC activities using this dataset. For the experiments, three distinct pairs of ICs and LBCs were prescribed: one utilizing the original GCM outputs (hereafter referred to as WRF_MPI), another employing the bias-corrected GCM data (referred to as WRF_MPIbc), and a third based on ERA5 from the European Centre for Medium-Range Weather Forecasts (referred to as WRF_ERA). We performed a comprehensive comparison and evaluation of these three simulations. The results indicate that the environmental conditions conducive to TC activity, such as wind shear, mid-level humidity, and monsoon troughs, were notably improved by the MVT bias correction in the WRF_MPIbc, particularly in critical regions for TC genesis and development. Consequently, WRF_MPIbc exhibited marked improvements in predicting TC genesis locations, the number of TCs, seasonal variations,tracks, and landfall occurrences. This suggests that bias correction in ICs and LBCs from GCMs is an effective method for simulating and projecting TCs and other high-impact weather events using dynamical downscaling.

tropical cyclones, bias-correction, dynamical downscaling