-
The two high-resolution AGCMs, FGOALS-f3-H and MRI-AGCM3-2-S, and the model simulations used in this study, are all derived from the HighResMIP, which is a part of the Coupled Model Intercomparison Project (CMIP6). Currently, CMIP6-HighResMIP has incorporated GCMs with grid spacings ranging from the CMIP6 resolution (250-km atmosphere and 100-km ocean) to higher resolutions (25-km atmosphere and 10-km ocean). A large ensemble of atmosphere-only and coupled model experiments are conducted for these GCMs using consistent forcings covering the present-day (1950–2014) and future period (2015–50) under the high-emission SSP585 scenario. More specific details about CMIP6-HighResMIP can be found in Haarsma et al. (2016). This study focuses on the simulations of highresSST-present experiment of their first ensemble member (r1i1p1f1) under the same external conditions with prescribed SST forcing. This methodology helps to clarify the responses to identical forcing changes across models with higher resolutions.
FGOALS-f3-H is the high-resolution version of the Chinese Academy of Sciences Flexible Global Ocean-Atmosphere-Land System, Finite-Volume Version 3 model. It includes two sets of configurations with horizontal grid spacings of 100 km (C96) and 25 km (C384) for HighResMIP Tier 1, and this study uses the higher resolution simulation of 25 km. The atmospheric component of FGOALS-f3-H is FAMIL2.2 has been largely developed in terms of its dynamic core and physics parameterizations, especially the resolving convective precipitation (RCP) scheme (Bao and Li, 2020), which greatly benefits the simulation of moist processes and associated tropical variabilities (He et al., 2019, 2020; Bao et al., 2020). Additional details about the oceanic, land surface, and sea-ice components can be found in Li et al. (2021).
MRI-AGCM3-2-S is a spectral model with a very high resolution of T959L60, equivalent to a 20-km grid spacing with 60 vertical layers. It is version 3.2 developed by the Japan Meteorological Agency (JMA) and MRI from the previous version 3.1. An important update was to replace the prognostic Arakawa-Schubert convection scheme in Version 3.1 with a modified Tiedtke scheme by incorporating two convective updrafts with different entrainment and detrainment rates. This modification has been recognized in MRI-AGCM V3.2 to effectively improve the simulation of tropical precipitation and TC activities (Mizuta et al., 2012; Murakami et al., 2012). More details about its physics parameterization schemes can be found in Mizuta et al. (2012) and Yoshimura et al. (2015).
-
The TC observations are derived from the International Best Track Archive for Climate Stewardship (IBTrACS v04r00), which is a global tropical cyclone multisource dataset (Knapp et al., 2010). Among them, the Regional Specialized Meteorological Centre Tokyo (RSMC) and Chinese Meteorological Administration-Shanghai Typhoon Institute (CMA_BST) in IBTrACS from 1985–2014 are selected to evaluate WNP TC activities (Ying et al., 2014; Lu et al., 2021). Here, the time period before 1985 is excluded due to the large data uncertainty for the lack of satellite observations. As suggested by Li et al. (2021), the 1-min-average maximum wind is converted to a 10-min-average maximum wind average by a reduction factor of 0.88. In addition, the Global Precipitation Measurement (GPM) IMERG-V6 product (resolution 0.1°) for the time period 2000–14 is used as the TC precipitation observations (Hou et al., 2014; Huffman et al., 2019). The fifth generation European Centre for Medium-Range Weather Forecasts atmospheric reanalysis dataset (ERA5, resolution 0.25°) is used to evaluate the simulation of large-scale atmospheric circulations (Hersbach et al., 2020).
-
TempestExtremes v2.1, a framework for scale-insensitive pointwise feature tracking introduced by Ullrich and Zarzycki (2017), is used for both AGCMs to detect the simulated TCs based on 6-h model output. This TC tracking algorithm essentially searches for a local minimum of sea-level pressure that is collocated with a local maximum of low-level cyclonic vorticity at 850 hPa and a warm core aloft at middle-to-upper pressure levels) and defines a minimum threshold for the maximum surface wind speed. Given that the 6-h surface wind for these two models is not released on the HighResMIP website, we use the 850-hPa wind speed as a detection criterion following the study of Murakami et al. (2012). In addition, the lifetime of the TC must be greater than 72 hours to filter out short-lived TCs. These key parameters in TempestExtremes v2.1 are determined from a sensitivity analysis (the Morris Method) based on four reanalysis data for TC tracking (Zarzycki and Ullrich, 2017). It is more reliable and flexible than traditional TC tracking algorithms (Ullrich et al., 2021) and has been widely used for TC detection in the HighResMIP models (Roberts et al., 2020a, b). Table 1 lists the TC identification criterion in TempestExtremes for FGOALS-f3-H and MRI-AGCM3-2-S, referring to Li et al. (2019) and Murakami et al. (2012), respectively.
FGOALS-f3-H MRI-AGCM3-2-S 850-hPa wind speed (m s–1) ≥ 15.4 ≥ 10 Absolute vorticity at 850 hPa (s–1) ≥ 3.5×10–5 − Relative vorticity at 850 hPa (s–1) − ≥ 2.0×10–5 Average temperature deviations between 300 and 500 hPa (K) ≥ 1 − Sum of the temperature deviations at 300, 500, and 700 hPa (K) − ≥ 1 Lifetime ≥ 72 h ≥ 72 h Table 1. Parameters of tropical cyclone tracking detection algorithms for FGOALS-f3-H and MRI-AGCM3-2-S
Given that GCMs still have difficulty in reproducing intense wind speeds for strong TCs, using the minimum sea level pressure (MSLP) to classify the TC intensity is considered to be more reliable than using the surface maximum wind speed for different climate models (Knaff and Zehr, 2007; Roberts et al., 2015, 2020b). Following Roberts et al. (2015), the intensity of TCs over the WNP can be classified into five categories, based on the relationship between MSLP and the maximum surface wind speed range in terms of the modified SS scale (Saffir-Simpson hurricane wind scale) as shown in Table 2.
TC Category Wind speed range (m s–1) Minimum sea level pressure range (hPa) TS (Tropical storm) 17.2–24.4 ≥ 985 STS (Severe tropical storm) 24.5–32.6 970–984 TY (Typhoon) 32.7–41.4 955–969 STY (Severe typhoon) 41.5–50.9 935–954 SSTY (Super typhoon) ≥ 51 <935 Table 2. The intensity classification of tropical cyclone over the Western North Pacific
-
The genesis potential index (GPI) applied in this study for analyzing the TCGF bias is based on Emanuel et al. (2004) and modified by Murakami and Wang (2010). Unlike the GPI definition used by Li et al. (2021), this GPI explicitly includes the vertical velocity and especially improves the GPI performance over the WNP. The modified GPI is defined as:
where
$ \eta $ is the 850-hPa absolute vorticity (s–1),$ \mathrm{R}\mathrm{H} $ is the 600-hPa relative humidity (%),$ {V}_{\mathrm{s}} $ is the magnitude of the wind shear between 850 and 200 hPa (m s–1),$ \omega $ is the vertical pressure velocity at 500 hPa (Pa s–1), and$ {V}_{\mathrm{m}\mathrm{a}\mathrm{x}} $ is the maximum potential intensity (MPI, m s-–1), which is defined by its original version in Emanuel (1995) as:where
$ {C}_{k} $ is the exchange coefficient of enthalpy,$ {C}_{\mathrm{d}} $ is the drag coefficient,$ {T}_{\mathrm{s}} $ is the SST, and$ {T}_{0} $ is the mean outflow temperature at 50 hPa (Gilford, 2021).$ {\mathrm{C}\mathrm{A}\mathrm{P}\mathrm{E}}_{*} $ is the convective available potential energy (CAPE) of the air, lifted from saturation at sea level, and$ {\mathrm{C}\mathrm{A}\mathrm{P}\mathrm{E}}_{\mathrm{b}} $ is the CAPE of the boundary layer air. The variables for$ {V}_{\mathrm{m}\mathrm{a}\mathrm{x}} $ are consistent with those of Li et al. (2021).
FGOALS-f3-H | MRI-AGCM3-2-S | |
850-hPa wind speed (m s–1) | ≥ 15.4 | ≥ 10 |
Absolute vorticity at 850 hPa (s–1) | ≥ 3.5×10–5 | − |
Relative vorticity at 850 hPa (s–1) | − | ≥ 2.0×10–5 |
Average temperature deviations between 300 and 500 hPa (K) | ≥ 1 | − |
Sum of the temperature deviations at 300, 500, and 700 hPa (K) | − | ≥ 1 |
Lifetime | ≥ 72 h | ≥ 72 h |