1. Introduction
Tropical cyclones (TCs) are rapidly rotating atmospheric vortices that originate over warm tropical oceans. About 80–90 TCs form annually (e.g., Frank and Young 2007), and some TCs make landfall in heavily populated regions, causing significant damage from dangerously gusty winds and torrential rainfall; storm surges can cause catastrophic damage (e.g., Blake 2018). To fully recognize the damage potential of a TC, both the wind- and rainfall-induced damages need to be considered.
With continued global warming expected from increasing greenhouse gases, how global TC activity will change in the future has been of great interest. Many previous studies have focused on changes in global TC frequency and intensity, often utilizing TC simulations from global climate models (GCMs), which have been invaluable in advancing our understanding of global TC activity (e.g., Knutson et al. 2010; Walsh et al. 2015; Knutson et al. 2020).
Compared to early trailblazing works that examined GCM TC simulations with horizontal resolutions of around 2°, contemporary GCM simulations can now be performed with a much higher resolution of 0.5° or better [e.g., Manabe et al. 1970; Zhao et al. 2009; Roberts et al. 2018; see Camargo and Wing (2016) for an overview of GCM simulations of TCs]. Using a higher resolution helps GCMs more realistically reproduce the observed TC activity (e.g., Zhang et al. 2016; Roberts et al. 2018). However, GCMs, including their high-resolution versions, still have difficulties correctly simulating the observed TC intensity distributions. It is well documented that GCMs tend to produce intense TCs less frequently in comparison to the observations (e.g., Shaevitz et al. 2014; Wehner et al. 2014). Previous studies found that in addition to the model resolution, details of GCM configurations, such as the physics parameterizations and dynamical cores, could contribute to the commonly noted low-intensity bias of GCM-simulated TCs (e.g., Reed and Jablonowski 2011; Kim et al. 2012; Zarzycki 2016).
To identify possible root causes for the biases in TC simulations, many previous studies examined the structures of GCM-simulated TCs (e.g., Manganello et al. 2012; Yamada et al. 2017; Gao et al. 2019; Wing et al. 2019), including the amount and distribution of rain rates around TCs (e.g., Kim et al. 2018, hereafter K18; Moon et al. 2020a, hereafter M20). For example, K18 and M20 examined TC rainfall structures in eight GCM simulations and found that simulations producing greater rainfall in the TC inner-core region tended to simulate stronger TCs more frequently. However, comparisons were made among the simulations only, rather than against observations, so it was difficult to determine which TC simulations were more realistic.
In the meantime, efforts have been made to evaluate TC rainfall structures against observations. (e.g., Villarini et al. 2014; Knutson et al. 2015; Liu et al. 2018). The model evaluation studies have shown that while the simulations were able to reasonably capture the TC rainfall distribution that decreases with increasing radius from the center, there were discrepancies on the radial location and magnitude of the maximum rainfall in the composites, with the simulations producing more pronounced peak rainfall that is located farther outward from the center.
The studies mentioned above speak to the need to examine TC rain rate structure in the climate models and comparing it against observations as new generation of GCM simulations becomes available. The High Resolution Model Intercomparison Project (HighResMIP; Haarsma et al. 2016), which is one of the Model Intercomparison Projects (MIPs) endorsed by phase 6 of the Coupled Model Intercomparison Project (CMIP6; Eyring et al. 2016), offers a unique opportunity toward that direction by providing of a coordinated set of experiments made with GCMs at different horizontal resolution. The main objective of HighResMIP is to examine how improved representation of smaller-scale processes arising from increased model horizontal resolution influences the projections and predictions of climate variability and change, including extreme weather events such as TCs (Haarsma et al. 2016).
This study—part of the efforts to develop process-oriented diagnostics developed for TCs in GCMs (e.g., K18; Wing et al. 2019; M20)—performs an evaluation of TC rainfall structures in the HighResMIP models against multiple satellite rainfall measurements. Our work expands the recent work by Vannière et al. (2020), Huang et al. (2021), and Zhang et al. (2021), who also examined simulated TCs in the HighResMIP multimodel ensemble in three aspects. First, our investigation includes the coupled HighResMIP simulations as well as the AMIP simulations. Vannière et al. (2020) and Zhang et al. (2021) examined only the AMIP simulations of the HighResMIP ensemble, and hence were unable to investigate the role of air–sea coupling, which is shown to have substantial impacts on simulated TCs (e.g., Zarzycki 2016; Scoccimarro et al. 2017a). Second, in contrast to the previous studies that evaluated the daily rainfall structures of TCs having the top 10% average daily rainfall (e.g., Villarini et al. 2014; Knutson et al. 2015) or the lifetime mean 3-hourly rainfall of the 200 strongest TCs (e.g., Vannière et al. 2020), we use a different but complementary analysis method that compares the rainfall composites of individual TC snapshots that have the same intensity (measured by maximum 10-m wind speed) both in the simulations and observations. Last, this study evaluates how realistically the HighResMIP simulations can represent the shear-induced precipitation asymmetries that are prominent in observed TCs.
The rainfall-based process-oriented diagnostic in this work could reveal which GCMs produce more realistic TCs by quantifying how closely GCM-simulated TC rainfall structures resemble those constructed from satellite observations and provide physical insights into the source of the model biases. In addition, our rainfall diagnostic can prove to be useful in increasing the fidelity of future projections in TC rainfall-induced damage potential by facilitating future model developments. How TC rainfall would change under global warming scenarios was investigated by many studies (e.g., Knutson and Tuleya 2004; Bengtsson et al. 2007; Knutson et al. 2010, 2013; Kim et al. 2014; Villarini et al. 2014; Yoshida et al. 2017; Patricola and Wehner 2018; Knutson et al. 2020), which typically inferred changes in TC activity by calculating the differences between present and future climate simulations using the same model. Identifying where the models struggle to realistically reproduce observed TC rainfall structures in the present climate simulations can aid future model development efforts and subsequently enhance the models’ capability to faithfully simulate TCs and their response to the future warming conditions.
We describe in sections 2 and 3 the satellite rainfall and TC best-track datasets used to construct the reference rainfall diagnostics in this study. Section 4 describes the HighResMIP simulations. Section 5 examines the azimuthally averaged and asymmetric TC rainfall structures and explores the relationship between the TC inner-core rainfall and intensification. The paper concludes with a summary in section 6.
2. Satellite rainfall retrievals
This study uses two satellite-based rain rate estimates to evaluate TC rainfall structures: the NOAA CPC Morphing Technique (CMORPH: Joyce et al. 2004; Xie et al. 2017) version 1 and Tropical Rainfall Measuring Mission (TRMM) 3B42 Multisatellite Precipitation Analysis (TMPA) version 7. Both datasets provide 3-hourly, 0.25° precipitation analysis. The CMORPH ingests precipitation retrievals from passive microwave overpasses [e.g., Special Sensor Microwave Imager (SSM/I; Wentz et al. 2012a), Special Sensor Microwave Imager Sounder (SSMIS; Wentz et al. 2012b), TRMM Microwave Imager (TMI; Wentz et al. 2015)] and propagates microwave-derived precipitation features with atmospheric motion vectors produced from geostationary infrared images to produce precipitation estimates at the analysis time. The CMORPH also includes bias corrections from calibrating against the CPC daily gauge analysis over land and against the GPCP-combined precipitation analysis over ocean. The TMPA combines and calibrates the available passive microwave rainfall estimates (e.g., TMI, SSM/I, SSMIS) and creates microwave-calibrated rainfall estimates from geostationary infrared datasets. Then, the microwave-based and infrared-based rainfall estimates are merged, and the combined dataset is adjusted with available rain gauges to produce the TMPA precipitation analysis. The use of two different rain rate products is to account for the uncertainty in the reference datasets.
It is possible that the satellite rainfall measurements, especially in the TC inner-core regions, might be somewhat underestimated. Several studies have evaluated the TRMM-3B42 rainfall against the ground-based observations and found that heavy TC rainfall points might be underestimated, but the total rainfall associated with TCs and heavy TC rain rates over ocean are more reasonable (e.g., Tuleya et al. 2007; Habib et al. 2009; Yu et al. 2009; Chang et al. 2013; Chen et al. 2013). Since most TCs are located over the ocean, our analysis results, especially those of the area-averaged rainfall, might not be severely affected by possible underestimation of heavy TC rainfall in the satellite measurements.
3. TC best-track data
For the TC center position and intensity, we use the TC best-track datasets prepared by the National Hurricane Center (NHC) for North Atlantic and eastern North Pacific TCs and by the Joint Typhoon Warning Center (JTWC) for the other basins. The NHC and JTWC data are from the International Best Track Archive for Climate Stewardship (IBTrACS; Knapp et al. 2010) version 4, which provides 6-hourly TC information. We also use the radius of maximum wind (RMW) estimated for TC snapshots in the IBTrACS.
4. HighResMIP/PRIMAVERA simulations
This study examines TC rainfall structures in five HighResMIP models: Centro Euro-Mediterraneo sui Cambiamenti Climatici Climate Model version 2 (CMCC-CM2; Cherchi et al. 2019; Scoccimarro et al. 2020), Centre National de Recherches Météorologiques Coupled Global Climate Model version 6 (CNRM-CM6; Voldoire et al. 2019), European Consortium Earth System Model 3 (EC-Earth3P; Haarsma et al. 2020), Hadley Centre Global Environment Model 3–Global Coupled version 3.1 (HadGEM3-GC31; Roberts et al. 2019), and Geophysical Fluid Dynamics Laboratory Coupled Model version 4 (GFDL-CM4; Zhao et al. 2018a,b; Adcroft et al. 2019; Held et al. 2019). More details of the first four models can be found in appendix A of Roberts et al. (2020a), which were performed as part of the European Union Horizon 2020 project PRIMAVERA (Process-based climate simulation: Advances in high-resolution modeling and European climate risk assessments; www.climateurope.eu/primavera). These HighResMIP/PRIMAVERA models have completed the Tier-1 atmosphere-only (“highresSST-present”) simulations using at least two different horizontal resolutions, and Roberts et al. (2020a) and Vannière et al. (2020) have recently examined their TC simulations. Roberts et al. (2020b) discussed TC activity in the future climate simulations. The GFDL-CM4 model simulations are available only for one horizontal resolution at the time of the analysis. We build on these previous studies by including the Tier-2 coupled (“hist-1950”) simulations. The highresSST-present simulations are AMIP simulations for 1950–2014 forced with observed historical forcing datasets [see Haarsma et al. (2016) for details]. The hist-1950 simulations are coupled simulations for the same period that are initialized from 50-yr spinup integrations that start from the 1950 initial state under 1950 conditions. Haarsma et al. (2016) provides more details of the HighResMIP models and simulation design. Hereafter, the highresSST-present and hist-1950 simulations will be referred to as the AMIP and coupled simulations, respectively.
This study analyzes a total of 18 HighResMIP simulations (see Table 1 for the models and simulations examined in this study). For each simulation type, each HighResMIP model typically includes lower- and higher-resolution simulations. However, at the time of the analysis, the EC-Earth3P model had only the AMIP simulations and the GFDL-CM4 model had one horizontal resolution only. This study labels each simulation as L (low), M (medium), or H (high) if the nominal model horizontal resolution is 1° or coarser, between 0.5° and 0.7°, or 0.35° or better, respectively.
Descriptions of the CMIP6 HighResMIP simulations examined in this study. The “highresSST-present” simulations refer to the atmosphere-only (AMIP) simulations, and the “hist-1950” refer to the coupled GCM simulations. Nominal horizontal resolution [high (H), medium (M), or low (L)] is assigned as the average of the latitude and longitude grid spacing.
The center and intensity of TCs in the CNRM, EC-Earth3P, and HadGEM3-GC31 simulations are obtained using the TempestExtremes (Ullrich and Zarzycki 2017; Zarzycki and Ullrich 2017; Roberts 2019b) TC tracking algorithm, while the TRACK (Hodges et al. 2017; Roberts 2019a) TC tracker is used for the CMCC-CM2 simulations. The TC tracking algorithm by Harris et al. (2016) and Murakami et al. (2015) is used for the GFDL-CM4 simulations. These TC trackers use different variables and threshold values for tracking TCs in GCM simulations. TRACK identifies cyclonic vorticity maxima in the average of the relative vorticity fields at 850, 700, and 600 hPa that are spectrally filtered to a common T63 grid. Then, it further checks to ensure that 850-hPa cyclonic relative vorticity is at least 6 × 10−5 s−1 and there is an upper-level warm core as well as coherent vorticity structures at 850, 700, 600, 500, and 250 hPa. TempestExtremes searches for local minima in sea level pressure and local maxima in the upper-level warm-core anomalies. In addition, sea level pressure needs to increase at least 2 hPa within 5.5° of the candidate, and geopotential height difference (250–500 hPa) needs to decrease by 6 m within 6.5° of the candidate. The tracking algorithm by Harris et al. (2016) and Murakami et al. (2015) searches for local maxima in 850-hPa cyclonic relative vorticity exceeding 1.6 × 10−4 s−1 that also have a local minimum of sea level pressure and local maximum in the averaged 300–500-hPa temperature. Appendix B of Roberts et al. (2020a) provides details of the TRACK and TempestExtremes trackers, while Harris et al. (2016) and Murakami et al. (2015) describe the TC tracker used for the GFDL-CM4 simulations. Using different TC trackers for different simulations does not appear to significantly influence the conclusions presented in this study. For example, changing the TC tracker for HadGCM3-GC31-HM from TempestExtremes to TRACK yielded only about a 1%–2% difference in TC rainfall within 500 km from TC centers (not shown). M20 also employed the composite method to examine TCs in GCM simulations using different tracking algorithms, but their results were not qualitatively sensitive to the choice of the tracking algorithm. Horn et al. (2014) suggested that tracking algorithm differences are more important for low-resolution simulations and weak TCs. For all trackers, TC intensity is measured by maximum 10-m wind speed obtained from the 6-hourly instantaneous wind fields.
Only TCs located between 25°S/N during 2000–14 are included in our analysis to focus on TC structures in the deep tropics. The rain rates are 6-h averages centered at the analysis times every 6 h (i.e., 0000, 0600, 1200, and 1800 UTC). To evaluate GCM-simulated TC rainfall structures against the satellite observations, we interpolate the rain rates onto a TC-centered 2000-km square grid of 5-km grid spacing. Note that the finest grid used is 0.27° and the use of a 5-km grid for interpolation is to aid visualization of the azimuthally averaged plots (e.g., Fig. 2) and not to resolve fine-scale features such as the eyewall. As TCs can be approximated in the zeroth order as axisymmetric vortices (e.g., Emanuel 1986, 2018), we compute the azimuthally averaged rainfall structures of HighResMIP-simulated TCs and compare them to those calculated from the satellite observations. Azimuthal averages are computed using 5-km radial increments out to r = 500 km from the center. Since HighResMIP TCs do not correspond to observed TCs, we create composites and compare with TCs of the same intensity.
5. Results
a. TC intensity distributions
Figure 1a shows the TC intensity distributions of the HighResMIP simulations and best-track observation. The black line in Fig. 1a displays the observed TC intensity distribution, and the colored lines show the TC intensity distributions of the simulations. Solid and dashed color lines are for the AMIP and coupled simulations, respectively. It is clear from Fig. 1a that in comparison to observations, the simulations tend to produce weaker TCs more frequently and stronger TCs less frequently, as discussed in Roberts et al. (2020a) and Vannière et al. (2020), which is also consistent with previous studies that examined GCM TC simulations of comparable resolution (e.g., Shaevitz et al. 2014; Roberts et al. 2015; Camargo et al. 2020). Figures 1b–d show the TC intensity distributions for the high-, medium-, and low-resolution simulations. As resolution increases (going from Fig. 1d to Fig. 1b), the simulations tend to produce stronger TCs more frequently. However, even for the high-resolution simulations in Fig. 1b, the fraction of strong TCs is still lower than in the observations. This low occurrence of the strongest TCs can be expected for these model resolutions, as discussed in Davis (2018) and Moon et al. (2020b). In this paper, we perform the TC rainfall analysis over three intensity intervals where the HighResMIP simulations produce abundant TC snapshots: 35–45 kt (i.e., weak tropical storms), 50–60 kt (i.e., strong tropical storms), and 65–75 kt (i.e., weak category-1 hurricanes) (1 kt ≈ 0.51 m s−1).
(a) Intensity distributions of TCs simulated in the HighResMIP simulations in color, showing (b)–(d) the intensity distributions for the high (H), medium (M), and low (L) resolution simulations only. The frequency TC snapshots at each intensity bin (x axis) is normalized by the number of all snapshots with the wind speed greater than 34 kt to yield frequency of occurrence (y axis). Black lines show the IBTrACS intensity distribution. Solid and dashed color lines are for the AMIP and coupled simulations, respectively. Numbers in the parentheses in (a) are the number of TC snapshots in the simulations and observations. The analysis is restricted to TCs between 25°S/N during 2000–14.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
b. Azimuthally averaged rainfall structures
Figures 2a–c show the composites of azimuthally averaged rain rates for the HighResMIP simulations and observations at the aforementioned three different TC intensities, while Fig. 3a shows the magnitude of the peak rain rates in the composites. Figure 3b shows the RMW in the composited azimuthally averaged surface wind fields. A first look at the 35–45-kt composites in Fig. 2a suggests that although these TCs have the same intensity, their azimuthally averaged rainfall profiles are quite different. However, in comparison to the satellite observations (black lines), many simulations produce more rainfall and their peak rain rate magnitudes are substantially higher, as noted in previous studies (e.g., Villarini et al. 2014; Knutson et al. 2015). Note that part of the high bias in the peak TC rain rate could be due to that our measure of TC intensity (maximum 10-m wind speed) is underestimated for a realistic central minimum pressure, which is often the case in GCMs (Manganello et al. 2012; Roberts et al. 2015; Yamada et al. 2017; K18). In addition, a closer look at these TC rainfall profiles reveals some horizontal resolution-dependent variations. Figures 2d–f show the 35–45-kt composites for the high-, medium-, and low-resolution simulations. As resolution increases from Fig. 2f to Fig. 2d, the peak rain rate magnitudes in the composites typically increase on average, but there are considerable differences in the peak magnitude among models with similar resolution (Fig. 3a). Furthermore, the TC rainfall fields tend to be more compact at higher resolution (cf. Figs. 2d–f), which is due to the smaller RMWs in the higher-resolution simulations as in previous studies (e.g., Manganello et al. 2012; Roberts et al. 2015; Gao et al. 2019; M20; Vannière et al. 2020). The average RMW of the composited azimuthally averaged 35–45-kt surface wind fields decreases from 265 km in the low-resolution simulations to 95 km in the high-resolution ones (Fig. 3b). The estimated RMW in the IBTrACS for the same-intensity TCs is about 70 km. Outside the RMW, the TC rainfall appears to decay more slowly in the lower-resolution simulations, more in agreement with observations.
(a)–(c) Composited azimuthally averaged rain rates for TC snapshots with an intensity of 35–45, 50–60, and 65–75 kt in the HighResMIP simulations in color lines. (d)–(f) Rain rate composites for the high (H), medium (M), and low (L) resolution simulations in (a). Solid black and dashed lines show the rain rate composites of the CMORPH and TRMM datasets. Solid and dashed color lines are for the AMIP and coupled simulations, respectively. Numbers in the parentheses in (a)–(c) are the number of TC snapshots in the simulations and observations.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
(a) Peak magnitude of the composited azimuthally averaged rain rates for 35–45-kt (circles), 50–60-kt (triangles), and 65–75-kt (squares) TC snapshots. (b) RMW of the composited azimuthally averaged surface wind fields at the same intensity intervals. Filled and open markers are the AMIP and coupled simulations. Black lines show the observations.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
The 50–60- and 65–75-kt rain rate composites are presented in Figs. 2b and 2c. As shown in Fig. 1, not all HighResMIP simulations have TCs reaching these intensities and the number of TC snapshots is also lower (see the figure legends). As in the 35–45-kt composites (Fig. 2a), the simulations produce more rainfall than the satellite observations for 50–60- and 65–75-kt TCs. In particular, the peak rain rate magnitude in the HighResMIP composites is substantially greater than that of the observations, and the degree of the model-to-observation differences appears to increase with intensity (Fig. 3a). The radial extent of the TC rainfall fields in Figs. 2b and 2c also decreases with increasing resolution for stronger TCs, as the RMW of the composited surface wind fields decreases with increasing resolution (Fig. 3b). It is evident from Fig. 2 that many HighResMIP simulations produce substantially more rainfall near the TC center than the satellite estimates for same-intensity TCs, especially when TCs are stronger, and that using higher resolution tends to produce the peak rain rate of greater magnitude.
Now we examine the difference in the TC rainfall structures between AMIP and coupled simulations in Fig. 4. Enabled air–sea coupling in the coupled simulations can affect TC rain via local SST changes under TCs or through causing changes in the model’s mean climate. Figures 4a–c show the differences in the azimuthally averaged rain rates between the coupled and AMIP simulations for 35–45-, 50–60-, and 65–75-kt TCs. Figures 4d–f show the RMW-normalized versions of Figs. 4a–c, which will be discussed in the next paragraph. In Fig. 4, differences are plotted as the percent change from the AMIP rainfall structures. Consistent with previous studies, the overall effect of ocean coupling is to lower the rain rates (e.g., Scoccimarro et al. 2017a; K18; M20; Huang et al. 2021). However, the effect of the coupling varies across the models. For example, for 35–45-kt TCs, the coupled-AMIP differences are greatest near the center for the CMCC and HadGEM3 simulations, but the opposite occurs for the CNRM simulations with the greatest differences found far from the center. However, as TC intensity increases, the largest coupled-AMIP differences in the CMCC and HadGEM3 simulations are also found away from the center as in the CNRM simulations. Comparing the AMIP and coupled simulations in Fig. 3b indicates that the ocean coupling tends to slightly increase the RMW in the CMCC and HadGEM3 simulations but decrease in the CNRM simulations. Ocean coupling tends to decrease the peak rain rate magnitude across all models and simulations (Fig. 3a) and lower the azimuthally averaged rainfall profiles, closer to the observations (Fig. 2).
(a)–(c) Differences between the coupled and AMIP simulations in the azimuthally averaged rain rates for TCs with an intensity of 35–45, 50–60, and 65–75 kt previously shown in Figs. 2a–c. (d)–(f) Differences between the coupled and AMIP simulations in the normalized azimuthally averaged rain rates for TCs that have the intensity of 35–45, 50–60, and 65–75 kt shown in Figs. 5a–c.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
As shown in Fig. 3, a major difference in the simulated TCs presented in Fig. 2 is the RMW that decreases with increasing resolution. While Fig. 2 can illustrate the differences in the azimuthally averaged rainfall profiles among the simulations, many of such differences are somewhat difficult to be recognized, especially outside the RMW. To help see them better, Fig. 5 shows the TC rainfall profiles of Fig. 2 but on the RMW-normalized radial coordinate, and their profiles are also normalized by the rain rate value at the RMW. It is evident in Fig. 5 that the normalized HighResMIP rainfall profiles decrease with radius outside the RMW at substantially faster rates than the observations, although the satellite rain rate products use infrared-based rainfall estimates and hence might not be suitable for examining fine-scale structures inside the eyewall. Figures 5d–f show the normalized rainfall profiles stratified by resolution for 35–45-kt TCs. As resolution increases (going from Fig. 5f to Fig. 5d), the normalized rain rates outside the RMW (i.e., the RMW-normalized radius > 1) decay more slowly. Similar patterns are present for stronger 50–60- and 65–75-kt TCs in Figs. 5b and 5c. In all rainfall profiles, the maximum normalized rainfall is found inside the RMW, indicating that the eyewall is located radially inward of the RMW, consistent with observations (e.g., Kimball and Mulekar 2004). Moving radially inward from the RMW, the rain rates in some simulations reach the maximum just inside the RMW and drop off quickly, while the rain rates in other simulations reach the maximum closer to the center and do not exhibit steep decline. These behaviors appear to be more a function of model-specific features (thus more influenced by the model parameterization physics) and less of model resolution. Judt et al. (2021) showed that TC inner core structure varies substantially across models even at storm-revolving resolution, suggesting that it is strongly affected by model physics and not constrained by resolved dynamics. For comparison, the coupled-AMIP differences in the RMW-normalized rainfall composites are shown as Figs. 4d–f. For the RMW-normalized radius > 1, these differences are negative in most simulations, meaning that ocean coupling mostly leads to faster decay of the normalized rain rates outward from the RMW. It appears that this faster rainfall decay in the coupled simulations increases with TC intensity.
As in Fig. 2, but on the RMW-normalized radial coordinate and the rain rate is normalized by the rain rate value at the RMW. Solid and dashed color lines are for the AMIP and coupled simulations, respectively.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
c. Area-averaged rainfall
Another helpful metric is to evaluate how much of area-averaged rainfall is associated with TCs. Figure 6a show the area-averaged rain rates within r = 500 km from the center for the 35–45-kt rainfall composites. Black bars of Fig. 6a are the area-averaged rain rates within r = 500 km for the CMORPH and TRMM composites at the same intensity, with the horizontal lines showing 1 and 2 times the mean of the satellite rain rates. The area-averaged rain rates in Fig. 6 are computed by weighting the azimuthally averaged rain rates in Fig. 2 with the area of a 5-km width annulus centered at the radial grid used in computing the azimuthal averages. Since the area of such annulus increases with radius [i.e.,
(a) Area-averaged rain rates within r = 500 km from the center for the 35–45-kt composites in the simulations (in color bars) with the black bars showing the same area-averaged rain rates for the same-intensity satellite composites. (b) As in (a), but shown by the model horizontal resolution. (c),(d) AMIP and coupled simulations, respectively, of (b). (e),(f) Area-averaged rain rates within r = 500 km for the 50–60- and 65–75-kt composites. Black horizontal lines show (lower line) the average of the two satellite rain rates and (upper line) 2 times the satellite estimates. Filled and open circles are the AMIP and coupled simulations.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
Comparing the AMIP and coupled simulations (solid vs dashed outlines of the same color) indicates that the amount of the area-averaged rainfall decreases and closer to the satellite mean when there is ocean coupling, consistent with Figs. 2 and 4. Using the area-averaged rainfall within r = 500 km does not qualitatively affect the results (not shown). Our results are consistent with those of Huang et al. (2021), who showed that TC rainfall in the coupled simulations is systematically lower than that in the AMIP simulations. Kreussler et al. (2021) found that TC integrated kinetic energy is reduced in the coupled simulation, while being not much affected by horizontal resolution. Huang et al. (2021) also showed that both the negative mean state SST bias and the local TC–ocean interactions via SST cold wakes lead to the weakening of TC rainfall in the coupled simulations. Further work is needed to quantify the relative importance between the mean state SST bias and the local SST feedback, which is beyond the scope of our study.
Another interesting feature in Fig. 6a is that for the HighResMIP models with multiple resolutions, the higher-resolution versions produce less rainfall than their lower-resolution counterparts. This horizontal resolution dependency can be better highlighted when the area-averaged rainfall is sorted by the absolute resolution of the simulations (Fig. 6b). As resolution improves to the left along the x axis, there is a tendency for the area-averaged rainfall to decrease. Plotting the AMIP and coupled simulations separately makes this resolution dependency even more evident, especially in the AMIP simulations (Figs. 6c,d). It is also interesting to note that the area-averaged rain rates are closer to the mean satellite retrievals in the higher-resolution versions of the models (left side of Figs. 6b–d). Figures 6e and 6f show the area-averaged r ≤ 500 km rain rates for 50–60- and 65–75-kt TCs. Similar to Figs. 6a–d, Figs. 6e and 6f indicate that many simulations produce more rainfall than the satellite estimates, with some simulations producing more than 2 times the mean satellite rain rates for 50–60-kt TCs. As in Figs. 6a–d, the area-averaged rain rates in the higher-resolution simulations are closer to the satellite observations, but this resolution-dependent trend is less evident than in Figs. 6b–d. It is interesting to note that the area-averaged rainfall of the coupled CNRM-M simulation produces the rainfall amount that is closest to the satellite observations at all three TC intensity intervals examined in this study.
As mentioned earlier, the peak rain rate magnitude tends to be greater and the RMW is smaller in the higher-resolution simulations. This might yield an impression that TCs in the higher-resolution simulations have a greater amount of the area-averaged rainfall than in the lower-resolution simulations. However, the area-averaged rainfall is a radius-weighted azimuthal average, so it is also strongly modulated by the rainfall occurring far from the center. To illustrate the importance of the outer-core rainfall in the area-averaged rainfall amount, Fig. 7 shows the composite rainfall profiles of Fig. 2 but weighted by the area of a 5-km width annulus centered at the corresponding radial grid points. In the lower-resolution simulations that have the larger RMW (Fig. 3b), the peak rain rates of smaller magnitude are often located farther radially outward than in the higher-resolution simulations. Therefore, the radius-weighted rainfall of the lower-resolution simulations in Fig. 7 is greater between the center and peak rainfall location, due to the fact that the increase in the annulus area up to the maximum rainfall location is greater than the decrease in the rainfall magnitude. The rain rate far from the center is higher in the lower-resolution simulations (Fig. 2) and so are their radius-weighted rain rates. Their combined contributions result in the greater amount of the area-averaged TC rainfall in the lower-resolution simulations.
As in Fig. 2, but the rain rates are weighted with the area of a 5-km-wide annulus centered at the radial grid used in computing the azimuthal averages in Fig. 2. Solid and dashed color lines are for the AMIP and coupled simulations, respectively. See section 5c for details.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
In Figs. 2–7, we have evaluated the simulated rainfall structures in terms of 1) the peak rain rate magnitude and 2) area-averaged rain rates within r = 500 km for TCs that have the intensity of 35–45, 50–60, and 65–75 kt. This analysis showed that the simulated rain rates of TCs in many HighResMIP simulations are greater than those from the satellite retrievals, which indicates that GCMs might need to generate more heating in order to produce a TC of the same intensity as observed. In addition, it is interesting to note that while both metrics exhibit resolution-dependent tendency, its direction is opposite. As model resolution increases, the magnitude of the peak rain rates increases, but the area-averaged rainfall amount decreases.
d. Rainfall, precipitable water, surface heat fluxes
To better understand the differences in the simulated TC rainfall distributions, we examine additional moisture-related variables from the HadGEM simulations that could help provide further insight. Figures 8a–c show composites of azimuthally averaged precipitable water for 35–45-, 50–60-, and 65–75-kt TCs. Precipitable water in Figs. 8a–c is greatest near the center in all composites, and it decreases with increasing radius from the inner-core region. At the same resolution, precipitable water increases with TC intensity especially near the center. Coupled simulations produce less precipitable water than their AMIP counterparts. When the composites have the same TC intensity, the lower-resolution simulations produce more precipitable water than their higher-resolution simulations. This suggests that the greater amount of the area-averaged rainfall in the lower-resolution simulations (Fig. 6) is associated with more precipitable water. However, the greater peak rain rate magnitude in the higher-resolution simulations (Fig. 2) cannot be attributed to precipitable water, which is actually lower in the higher-resolution simulations.
(left) The composites of azimuthally averaged precipitable water for TC snapshots having the intensity of (a) 35–45, (b) 50–60, and (c) 65–75 kt in the HadGEM simulations. Solid and dashed color lines are for the AMIP and coupled simulations, respectively. (right) The scatterplots between precipitation vs precipitable water for TC snapshots that have an intensity of (d) 35–45, (e) 50–60, and (f) 65–75 kt in the HadGEM simulations using the composited values shown in Figs. 2a–c and 8a–c. Solid dots are for the AMIP simulations, and open circles are for the coupled simulations.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
Figures 8d–f show scatterplots between the composites of azimuthally averaged rain rates and precipitable water from Figs. 2a–c and 8a–c. The scatterplots exhibit a tight coupling between precipitable water and rain rates, with rain rates starting to increase exponentially with precipitable water when precipitable water becomes greater than about 54–56 mm, which is qualitatively similar to previous studies that examined the relationship between precipitation and precipitable water using the satellite observations (e.g., Bretherton et al. 2004; Peters and Neelin 2006; Holloway and Neelin 2010). The exponential increase in rain rates with precipitable water appears to be steeper in the coupled and higher-resolution simulations than in their uncoupled and lower-resolution counterparts. As shown in Figs. 2a–c, the maximum rain rates in the HadGEM simulations are greater in the higher-resolution and AMIP simulations than their counterparts. However, these rain rate maxima are not associated the precipitable water maxima. After reaching the maximum rain rates, additional increases in precipitable water actually lead to decreases in rain rates, which occurs very close to the center (cf. Figs. 2a–c vs Figs. 8a–c), likely due to the upper-level warm-core positive temperature anomalies and the associated increase in static stability. The degree of such decrease in rain rates with increasing precipitable water appears to be greater in higher-resolution simulations, likely due to the smaller RMW in the higher-resolution simulations, which results in the warm-core anomalies of greater magnitude. Before rain rates start to decrease with increasing precipitable water, coupled simulations tend to produce more rain rates at given precipitable water than their counterparts.
Figures 9a–c show the surface latent heat flux composites for 35–45-, 50–60-, and 65–75-kt TCs. The composite surface latent heat flux, which accounts for more than 80% of the total surface heat flux (not shown), is greatest near the RMW in the corresponding horizontal wind fields, and decreases away from the RMW in both increasing and decreasing radial directions. As TC intensity increases, the magnitude of the surface latent heat flux increases, especially closer to the center. Not surprisingly, the surface latent heat flux in the coupled simulations is lower than that in the AMIP simulations due to the effect of the wind-induced reduction of SST beneath TCs (e.g., Price 1981). The difference in the surface latent heat flux due to the ocean coupling appears to increase with TC intensity. For TCs of the same intensity, the maximum surface latent heat flux increases with resolution. However, this increase in the maximum surface heat flux magnitude is very modest, and the amount of the area-averaged surface latent heat flux within r = 500 km (the bars in Figs. 9a–c) is instead greater for the lower-resolution simulations. This is because the radius-weighted, area-averaged surface heat flux amount is more strongly modulated by the surface heat flux occurring far from the center, as in Figs. 6 and 7. The higher peak rain rate magnitude in the higher-resolution simulations appears to be aided by the slightly greater surface latent heat flux from below.
(left) Composites of azimuthally averaged surface latent heat flux for TC snapshots with an intensity of (a) 35–45, (b) 50–60, and (c) 65–75 kt in the HadGEM simulations. Solid and dashed color lines are for the AMIP and coupled simulations, respectively. The inset bar graphs show the area-averaged surface heat flux within r = 500 km. (right) Differences between the composited rain rate and evaporation for TCs with an intensity of (d) 35–45, (e) 50–60, and (f) 65–75 kt. The inset bar graphs show the area- averaged P − E within r = 500 km.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
Figures 9d–f show the difference between the composited precipitation and evaporation for 35–45-, 50–60-, and 65–75-kt TCs. If the tendency of precipitable water is small, which is the case as shown in Vannière et al. (2020), this P − E difference is mostly related to the vertically integrated horizontal moisture flux convergence. If evaporation plays the most dominant role in modulating the TC rainfall amount, P − E should be close to zero. Instead, the comparison of Figs. 9a–c with Figs. 9d–f suggests that horizontal moisture flux convergence appears to play a bigger role in modulating area-averaged TC rainfall than local evaporation. Quantitatively, the inferred vertically averaged horizontal moisture flux convergence accounts for about 57%–69% of the area-averaged rain rates (Table 2). Our result is largely consistent with that of Vannière et al. (2020), who showed that vertically integrated horizontal moisture flux convergence balances precipitation in HighResMIP-simulated TCs. While the area-averaged surface evaporation decreases as grid spacing becomes smaller (Figs. 9a–c), it explains a greater fraction of area-averaged rain rate as the horizontal resolution increases. Also, when the AMIP simulations are compared to their corresponding coupled simulations, water vapor used for rain comes more from horizontal moisture flux convergence when air–sea coupling is enabled, likely due to the reduced surface evaporation around TCs (Figs. 9a–c). It should be noted that the water budget analysis conducted in this study cannot consider possible interactions among surface evaporation, convection, and TC circulation near TC centers. Understanding the relative role of surface evaporation and horizontal moisture flux convergence on modulating large-scale TC rain warrants further work.
The ratio of vertically averaged horizontal moisture flux convergence to the area-averaged rain rates within r = 500 km from the center.
e. Asymmetric TC rainfall distributions
A prominent feature in the observed TC precipitation is the vertical wind shear–induced asymmetries. The presence of environmental vertical wind shear is evident in TC rainfall snapshots, showing up as the azimuthal wavenumber-1 (n = 1) asymmetries with the positive anomalies located in the downshear regions (e.g., Corbosiero and Molinari 2003; Lonfat et al. 2004; Chen et al. 2006). Vertical wind shear is known to exert a strong influence on TC intensity (e.g., DeMaria and Kaplan 1994). Therefore, it is important to examine whether GCM-simulated TCs can correctly reproduce these vertical wind shear–induced precipitation asymmetries as seen in the observations. Since the HighResMIP simulations report horizontal wind variables at the 250-hPa (not 200 hPa) level, the vertical wind shear vector is computed as the 250–850-hPa difference in the simulations and observations.
We first examine the influence of vertical wind shear in the satellite rainfall observations. Figure 10 shows the composite horizontal maps of rain rates for 35–45-, 50–60-, and 65–75-kt TCs. Averages of horizontal winds over an annulus of r = 200–800 km are used to compute the vertical wind shear vector around the observed TCs as in DeMaria et al. (2005) using the ERA5 reanalysis. Before compositing, individual rainfall snapshots have been rotated by the vertical wind shear vector, such that the shear vector is always directed toward the positive y axis and that the positive cyclonic direction is counterclockwise as in the Northern Hemisphere in all snapshots. In all composites, the shear-rotated rainfall is greater in the downshear regions (y > 0) than in the upshear regions (y < 0). The maximum of the shear-rotated rainfall is located off the center in the downshear-left quadrant (x < 0, y > 0). In addition, as TC intensity increases, the magnitude of the shear-rotated rainfall maximum increases and the peak rainfall location appears to move cyclonically and inward toward the center. These features in the shear-rotated rainfall composites are consistent with those noted in the previous studies (e.g., Corbosiero and Molinari 2003; Lonfat et al. 2004; Chen et al. 2006). The CMORPH and TRMM composites are qualitatively and quantitatively similar to each other.
The composited horizontal maps of rain rates for TCs with an intensity of (top) 35–45, (middle) 50–60, and (bottom) 65–75 kt in the (left) CMORPH and (right) TRMM datasets. Dashed concentric circles are for every 50-km radius. The plotting boxes are 700-km squares. The shear vector is always directed toward the positive y axis.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
Figure 11 shows the shear-rotated rain rate composites for the HighResMIP-simulated TCs that have an intensity of 35–45 kt. Because the RMW is larger in the HighResMIP-simulated TCs, the environmental horizontal winds are averaged over r = 400–500 km. It is clear in Fig. 11 that the shear-rotated rainfall is greater in the downshear regions in all of the composites, with the maximum rainfall being located in the downshear-left quadrant, which is consistent with the satellite composites in Fig. 10. As in the azimuthally averaged profiles, the main difference between the simulations and satellite observations is the radial location of the maximum rain rates, with the maximum in most simulations being located further radially outward from the center, which is due to the larger RMW in the HighResMIP-simulated TCs. However, some simulations (CMCC-H, HadGEM-H, and GFDL-M) produce the maximum rainfall closer to the center than the satellite composites. In addition, the maximum rain rates increase with increasing resolution, except in the AMIP CMCC simulations, in which the maximum rain rate is actually ower in its higher-resolution version (cf. Figs. 11a and 11c). The maximum rain rates in most of the simulations (i.e., 16 out of 18 in Fig. 11) are greater than those in the satellite observations (Figs. 10a,d). Comparisons between the AMIP and coupled simulations indicate that the maximum rain rates are greater in the AMIP versions of the CMCC-L, CNRM, HadGEM-L, HadGEM-H, and GFDL-M simulations, but the opposite is true for the CMCC-H and HadGEM-M simulations.
The composited horizontal maps of rain rates for TCs with an intensity of 35–45 kt. Dashed concentric circles indicate each successive 50-km radius. The plotting boxes are 700-km squares. The shear vector is always directed toward the positive y axis.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
Figures 12 and 13 show the shear-rotated rain rate composites for 50–60- and 65–75-kt TCs. As in Fig. 11, the composites in Figs. 12 and 13 are in agreement with the satellite observations with the greater rainfall in the downshear regions and the maximum occurring in the downshear-left quadrant. For 50–60-kt TCs, the maximum rain rates increase with increasing resolution in the CNRM and HadGEM models. In addition, the maximum rain rates are greater in the AMIP versions than in their coupled counterparts, except in the CMCC-H simulations at 50–60 kt. As in Fig. 11, the maximum rain rates in the simulations are greater than those in the satellite observations. Comparing Figs. 11–13 indicates that as TC intensity increases, the maximum of the shear-rotated rain rates increases in magnitude and their locations appear to move closer toward the center and turn cyclonically, in agreement with the satellite observations (Fig. 10). Qualitative resemblance of the shear-rotated rainfall composites between the observations and simulations indicates that the effect of vertical wind shear on TCs is properly represented in the HighResMIP simulations.
As in Fig. 11, but for TCs with an intensity of 50–60 kt.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
As in Fig. 11, but for TCs with an intensity of 65–75 kt.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
f. TC inner-core rainfall and intensification
Recently, K18 and M20 showed that GCM simulations that produce stronger TCs tend to produce a greater amount of rainfall—and thus diabatic heating—in the TC inner-core region. This finding is consistent with previous theoretical studies that found that more diabatic heating closer to the center is favorable for TC intensification because the efficiency of the conversion of a heat source into kinetic energy of the TC cyclonic flow is greater near the center (e.g., Schubert and Hack 1982; Shapiro and Willoughby 1982; Hack and Schubert 1986; Nolan et al. 2007). Their analysis was restricted to a small opportunity-based multimodel ensemble that had different simulation years and slightly different forcing and boundary conditions. Therefore, we re-examine the findings of K18 and M20 with the HighResMIP simulations that are performed for the same years using the same forcing and boundary conditions.
Figure 14a shows a scatterplot of the area-averaged inner-core rain rates for TCs that have an intensity of 35–45 kt versus the fraction of TCs intensifying from 35 to 50 kt. Figures 14b and 14c show the same scatterplot of Fig. 14a but for the AMIP and coupled simulations only. The TC inner-core region is defined to be 2 times the RMW in the composited azimuthally averaged surface wind fields (Fig. 3b). The inner-core rain rates are computed using the composites shown in Figs. 2a–c. Diamonds, squares, and circles are for the high-, medium-, and low-resolution simulations, and black triangles show the same quantity as computed with the CMORPH/TRMM and IBTrACS datasets. The scatterplot in Fig. 14a shows a clear positive relationship between the inner-core rainfall and intensification likelihoods across the simulations; that is, GCM simulations that produce stronger TCs more frequently have the greater rainfall close to the center. The correlation coefficient between inner-core rainfall and intensification likelihoods is about 0.69, which is statistically significant at the 95% confidence level. These are in agreement with the aforementioned theoretical and GCM simulation results (e.g., Schubert and Hack 1982; Shapiro and Willoughby 1982; Hack and Schubert 1986; Nolan et al. 2007; K18; M20). Further examination of Fig. 14 suggests that higher-resolution simulations tend to produce a greater amount of the inner-core rainfall and have a higher probability of further TC intensification than the lower-resolution simulations. However, the effect of ocean coupling is not uniform across the models. In the CNRM, HadGEM, and GFDL models, the coupled simulations have a lower amount of the inner-core rainfall (Fig. 4a) and lower probability of further TC intensification. This relationship is not evident in the CMCC simulations.
Scatterplots of the area-averaged inner-core rain rates at 35–45 kt vs the fraction of TCs intensifying from 35 to 50 kt in (a) all simulations, (b) AMIP simulations, and (c) coupled simulations. The area-averaged inner-core rain rates are computed from the composites shown in Figs. 2a–c. The inner-core region is defined to be 2 times the RMW in the composited azimuthally averaged surface wind speed fields (Figs. 3d–f). Black triangles are calculated with the CMORPH/TRMM and IBTrACS datasets.
Citation: Journal of Climate 35, 22; 10.1175/JCLI-D-21-0564.1
It is interesting to note from Fig. 14 that the inner-core rain rates computed from the satellite observations fall within the range of those computed from the simulations. However, the observed TCs have substantially higher intensification likelihoods than the HighResMIP-simulated TCs. Extrapolating the positive relationship exhibited by the models suggest that GCM simulations might need to generate higher inner-core rain rates—and thus more diabatic heating—in order to produce the same intensification likelihoods as observed. This indicates that there is a deficiency in GCMs in their representation of the physical processes responsible for TC intensification, such as the relationship between diabatic heating and rainfall or TC circulation response to the diabatic heating in the models.
6. Summary
This study has evaluated simulated TC rainfall structures in the HighResMIP multimodel ensemble simulations against the satellite observations. Comparing composites of the simulated TC rainfall structures to the satellite observations at the same TC intensity has revealed the following results:
-
In terms of the peak rain rate magnitude and area-averaged rainfall within r = 500 km, the rainfall metrics in many HighResMIP-simulated TCs are greater than the satellite observations.
-
However, these rainfall metrics exhibit opposite trend with model horizontal resolution. As resolution increases, the peak rain rate magnitude increases (greater model bias), but the area-averaged rain rate within r = 500 km decreases (smaller bias). The area-averaged rainfall decreases (smaller bias) with increasing resolution, which is closely related to the TC eyewall being located closer to the center, thus occupying a smaller area and contributing less to the area-averaged rainfall.
-
The effect of ocean coupling is to lower the TC rain rates, bringing them closer to the satellite observations, due to reduced horizontal moisture flux convergence and surface latent heat flux beneath TCs.
-
Examination of horizontal rainfall distributions indicates that vertical wind shear–induced rainfall asymmetries (i.e., higher rainfall in the downshear regions) in HighResMIP-simulated TCs are in good agreement with the observations.
-
A positive relationship is observed between the inner-core rainfall and intensification likelihoods across the HighResMIP simulations, as GCM simulations producing stronger TCs more frequently have the greater rainfall close to the center, which is consistent with previous theoretical and GCM simulation results.
Our analysis indicated that higher-resolution HighResMIP simulations that performed better in reproducing the area-averaged TC rain rates tended to perform worse in reproducing the peak rain rate magnitude, which deserves further investigation as more HighResMIP simulations become available. The above results emphasized the benefit of examining TC rainfall structures with more than one metric, as TC rainfall metrics could exhibit opposite trend with resolution.
It is important to note that our findings are somewhat different from those of Vannière et al. (2020), who performed a rainfall analysis using the same HighResMIP simulations and found that the amount of rainfall per TC does not appear to vary significantly with model resolution. This is likely due to the differences in the composite method. Vannière et al. (2020) created the composites of the lifetime mean precipitation of the 200 strongest (by minimum sea level pressure) Northern Hemisphere TCs, while our composites are for individual TC snapshots having the same intensity (by maximum 10-m wind speed). By creating the composites from the strongest 200 TCs, the average TC intensity in the higher-resolution simulation composites is likely greater than that in the lower-resolution counterparts, thus compensating for the lower rainfall in the higher-resolution simulation composites.
Last, our results showed that there exists notable intermodel spread in TC rainfall among models with comparable horizontal resolution. While many studies have investigated the effect of model configurations, including the cumulus parameterization, on the intensity and frequency of simulated TCs (e.g., Vitart et al. 2001; Zhao et al. 2012; Stan 2012; Kim et al. 2012; Duvel et al. 2017), not much is known about the extent to which TC rainfall is affected by individual model parameterizations. A systematic investigation of the origin of the model-to-model spread in TC rainfall is warranted to reduce model uncertainty associated with TC rainfall in the current and future climates.
Acknowledgments.
This work is supported by the DOE RGMA Grant DE‐SC0016223 and NOAA MAPP Grants NA15OAR4310087, NA15OAR4310095, NA18OAR4310270, NA18OAR4310276, and NA18OAR4310277. DK was also supported by the Brain Pool program funded by the Ministry of Science and ICT through the National Research Foundation of Korea (NRF-2021H1D3A2A01039352) and the KMA R&D program grant KMI2021-01210. MJR acknowledges funding from the PRIMAVERA project, funded by the European Union’s Horizon 2020 programme under Grant Agreement 641727. The CMIP6 HighResMIP simulations are available from the Earth System Grid Federation CMIP6 portal. This work is a contribution to the process-oriented diagnostics efforts of the NOAA MAPP Model Diagnostics Task Force (Maloney et al. 2019).
REFERENCES
Adcroft, A., and Coauthors, 2019: The GFDL global ocean and sea ice model OM4.0: Model description and simulation features. J. Adv. Model. Earth Syst., 11, 3167–3211, https://doi.org/10.1029/2019MS001726.
Bengtsson, L., K. I. Hodges, M. Esch, N. Keenlyside, L. Kornblueh, J.-J. Luo, and T. Yamagata, 2007: How may tropical cyclones change in a warmer climate? Tellus, 59A, 539–561, https://doi.org/10.1111/j.1600-0870.2007.00251.x.
Blake, E. S., 2018: The 2017 Atlantic hurricane season: Catastrophic losses and costs. Weatherwise, 71, 28–37, https://doi.org/10.1080/00431672.2018.1448147.
Bretherton, C. S., M. E. Peters, and L. E. Back, 2004: Relationships between water vapor path and precipitation over the tropical oceans. J. Climate, 17, 1517–1528, https://doi.org/10.1175/1520-0442(2004)017<1517:RBWVPA>2.0.CO;2.
Camargo, S. J., and A. A. Wing, 2016: Tropical cyclones in climate models. Wiley Interdiscip. Rev.: Climate Change, 7, 211–237, https://doi.org/10.1002/wcc.373.
Camargo, S. J., and Coauthors, 2020: Characteristics of model tropical cyclone climatology and the large-scale environment. J. Climate, 33, 4463–4487, https://doi.org/10.1175/JCLI-D-19-0500.1.
Chang, L. T.-C., K. K. W. Cheung, and J. McAneney, 2013: Case study of TRMM satellite rainfall estimation for landfalling tropical cyclones: Issues and challenges. Trop. Cyclone Res. Rev., 2, 109–123, https://doi.org/10.6057/2013TCRR02.04.
Chen, S. S., J. A. Knaff, and F. D. Marks, 2006: Effects of vertical wind shear and storm motion on tropical cyclone rainfall asymmetries deduced from TRMM. Mon. Wea. Rev., 134, 3190–3208, https://doi.org/10.1175/MWR3245.1.
Chen, Y., E. E. Ebert, K. J. E. Walsh, and N. E. Davidson, 2013: Evaluation of TRMM 3B42 precipitation estimates of tropical cyclone rainfall using PACRAIN data. J. Geophys. Res. Atmos., 118, 2184–2196, https://doi.org/10.1002/jgrd.50250.
Cherchi, A., and Coauthors, 2019: Global mean climate and main patterns of variability in the CMCC-CM2 coupled model. J. Adv. Model. Earth Syst., 11, 185–209, https://doi.org/10.1029/2018MS001369.
Corbosiero, K. L., and J. Molinari, 2003: The relationship between storm motion, vertical wind shear, and convective asymmetries in tropical cyclones. J. Atmos. Sci., 60, 366–376, https://doi.org/10.1175/1520-0469(2003)060<0366:TRBSMV>2.0.CO;2.
Davis, C. A., 2018: Resolving tropical cyclone intensity in models. Geophys. Res. Lett., 45, 2082–2087, https://doi.org/10.1002/2017GL076966.
DeMaria, M., and J. Kaplan, 1994: A Statistical Hurricane Intensity Prediction Scheme (SHIPS) for the Atlantic basin. Wea. Forecasting, 9, 209–220, https://doi.org/10.1175/1520-0434(1994)009<0209:ASHIPS>2.0.CO;2.
DeMaria, M., M. Mainelli, L. K. Shay, J. A. Knaff, and J. Kaplan, 2005: Further improvements to the Statistical Hurricane Intensity Prediction Scheme (SHIPS). Wea. Forecasting, 20, 531–543, https://doi.org/10.1175/WAF862.1.
Duvel, J.-P., S. J. Camargo, and A. H. Sobel, 2017: Role of the convection scheme in modeling initiation and intensification of tropical depressions over the North Atlantic. Mon. Wea. Rev., 145, 1495–1509, https://doi.org/10.1175/MWR-D-16-0201.1.
EC-Earth, 2018: EC-Earth-Consortium EC-Earth3P-HR model output prepared for CMIP6 HighResMIP. Earth System Grid Federation, accessed September 2019, https://doi.org/10.22033/ESGF/CMIP6.2323.
EC-Earth, 2019: EC-Earth-Consortium EC-Earth3P model output prepared for CMIP6 HighResMIP. Earth System Grid Federation, accessed September 2019, https://doi.org/10.22033/ESGF/CMIP6.2322.
Emanuel, K. A., 1986: An air–sea interaction theory for tropical cyclones. Part I: Steady state maintenance. J. Atmos. Sci., 43, 585–605, https://doi.org/10.1175/1520-0469(1986)043<0585:AASITF>2.0.CO;2.
Emanuel, K. A., 2018: 100 years of progress in tropical cyclone research. Meteor. Monogr., No. 59, Amer. Meteor. Soc., 15.1–15.68. https://doi.org/10.1175/AMSMONOGRAPHS-D-18-0016.1.
Eyring, V., S. Bony, G. A. Meehl, C. A. Senior, B. Stevens, R. J. Stouffer, and K. E. Taylor, 2016: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev., 9, 1937–1958, https://doi.org/10.5194/gmd-9-1937-2016.
Frank, W. M., and G. S. Young, 2007: The interannual variability of tropical cyclones. Mon. Wea. Rev., 135, 3587–3598, https://doi.org/10.1175/MWR3435.1.
Gao, K., L. Harris, J.-H. Chen, S.-J. Lin, and A. Hazelton, 2019: Improving AGCM hurricane structure with two‐way nesting. J. Adv. Model. Earth Syst., 11, 278–292, https://doi.org/10.1029/2018MS001359.
Haarsma, R. J., and Coauthors, 2016: High Resolution Model Intercomparison Project (HighResMIP v1.0) for CMIP6. Geosci. Model Dev., 9, 4185–4208, https://doi.org/10.5194/gmd-9-4185-2016.
Haarsma, R. J., and Coauthors, 2020: HighResMIP versions of EC-Earth: EC-Earth3P and EC-Earth3P-HR—Description, model computational performance and basic validation. Geosci. Model Dev., 13, 3507–3527, https://doi.org/10.5194/gmd-13-3507-2020.
Habib, E., A. Henschke, and R. F. Adler, 2009: Evaluation of TMPA satellite-based research and real-time rainfall estimates during six tropical-related heavy rainfall events over Louisiana, USA. Atmos. Res., 94, 373–388, https://doi.org/10.1016/j.atmosres.2009.06.015.
Hack, J. J., and W. H. Schubert, 1986: Nonlinear response of atmospheric vortices to heating by organized cumulus convection. J. Atmos. Sci., 43, 1559–1573, https://doi.org/10.1175/1520-0469(1986)043<1559:NROAVT>2.0.CO;2.
Harris, L. M., S.-J. Lin, and C. Tu, 2016: High-resolution climate simulations using GFDL HiRAM with a stretched global grid. J. Climate, 29, 4293–4314, https://doi.org/10.1175/JCLI-D-15-0389.1.
Held, I. M., and Coauthors, 2019: Structure and performance of GFDL’s CM4.0 climate model. J. Adv. Model. Earth Syst., 11, 3691–3727, https://doi.org/10.1029/2019MS001829.
Hodges, K., A. Cobb, and P. L. Vidale, 2017: How well are tropical cyclones represented in reanalysis datasets? J. Climate, 30, 5243–5264, https://doi.org/10.1175/JCLI-D-16-0557.1.
Holloway, C. E., and J. D. Neelin, 2010: Temporal relations of column water vapor and tropical precipitation. J. Atmos. Sci., 67, 1091–1105, https://doi.org/10.1175/2009JAS3284.1.
Horn, M., and Coauthors, 2014: Tracking scheme dependence of simulated tropical cyclone response to idealized climate simulations. J. Climate, 27, 9197–9213, https://doi.org/10.1175/JCLI-D-14-00200.1.
Huang, H., C. M. Patricola, and W. D. Collins, 2021: The influence of ocean coupling on simulated and projected tropical cyclone precipitation in the HighResMIP–PRIMAVERA simulations. Geophys. Res. Lett., 48, e2021GL094801, https://doi.org/10.1029/2021GL094801.
Jones, E., A. A. Wing, and R. Parfitt, 2021: A global perspective of tropical cyclone precipitation in reanalyses. J. Climate, 34, 8461–8480, https://doi.org/10.1175/JCLI-D-20-0892.1.
Joyce, R. J., J. E. Janowiak, P. A. Arkin, and P. Xie, 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487–503, https://doi.org/10.1175/1525-7541(2004)005<0487:CAMTPG>2.0.CO;2.
Judt, F., and Coauthors, 2021: Tropical cyclones in global storm-resolving models. J. Meteor. Soc. Japan, 99, 579–602, https://doi.org/10.2151/jmsj.2021-029.
Kim, D., A. H. Sobel, A. D. Del Genio, Y. Chen, S. J. Camargo, M.-S. Yao, M. Kelley, and L. Nazarenko, 2012: The tropical subseasonal variability simulated in the NASA GISS general circulation model. J. Climate, 25, 4641–4659, https://doi.org/10.1175/JCLI-D-11-00447.1.
Kim, D., and Coauthors, 2018: Process-oriented diagnosis of tropical cyclones in high-resolution GCMs. J. Climate, 31, 1685–1702, https://doi.org/10.1175/JCLI-D-17-0269.1.
Kim, H.-S., G. A. Vecchi, T. R. Knutson, W. G. Anderson, T. L. Delworth, A. Rosati, F. Zeng, and M. Zhao, 2014: Tropical cyclone simulation and response to CO2 doubling in the GFDL CM2.5 high-resolution coupled climate model. J. Climate, 27, 8034–8054, https://doi.org/10.1175/JCLI-D-13-00475.1.
Kimball, S. K., and M. S. Mulekar, 2004: A 15-year climatology of North Atlantic tropical cyclones. Part I: Size parameters. J. Climate, 17, 3555–3575, https://doi.org/10.1175/1520-0442(2004)017<3555:AYCONA>2.0.CO;2.
Knapp, K. R., M. C. Kruk, D. H. Levinson, H. J. Diamond, and C. J. Neumann, 2010: The International Best Track Archive for Climate Stewardship (IBTrACS). Bull. Amer. Meteor. Soc., 91, 363–376, https://doi.org/10.1175/2009BAMS2755.1.
Knutson, T. R., and R. E. Tuleya, 2004: Impact of CO2-induced warming on simulated hurricane intensity and precipitation: Sensitivity to the choice of climate model and convective parameterization. J. Climate, 17, 3477–3495, https://doi.org/10.1175/1520-0442(2004)017<3477:IOCWOS>2.0.CO;2.
Knutson, T. R., and Coauthors, 2010: Tropical cyclones and climate change. Nat. Geosci., 3, 157–163, https://doi.org/10.1038/ngeo779.
Knutson, T. R., and Coauthors, 2013: Dynamical downscaling projections of twenty-first-century Atlantic hurricane activity: CMIP3 and CMIP5 model-based scenarios. J. Climate, 26, 6591–6617, https://doi.org/10.1175/JCLI-D-12-00539.1.
Knutson, T. R., J. J. Sirutis, M. Zhao, R. E. Tuleya, M. Bender, G. A. Vecchi, G. Villarini, and D. Chavas, 2015: Global projections of intense tropical cyclone activity for the late twenty-first century from dynamical downscaling of CMIP5/RCP4.5 scenarios. J. Climate, 28, 7203–7224, https://doi.org/10.1175/JCLI-D-15-0129.1.
Knutson, T. R., and Coauthors, 2020: Tropical cyclones and climate change assessment: Part II: Projected response to anthropogenic warming. Bull. Amer. Meteor. Soc., 101, E303–E322, https://doi.org/10.1175/BAMS-D-18-0194.1.
Kreussler, P., and Coauthors, 2021: Tropical cyclone integrated kinetic energy in an ensemble of HighResMIP simulations. Geophys. Res. Lett., 48, e2020GL090963, https://doi.org/10.1029/2020GL090963.
Liu, M., G. A. Vecchi, J. A. Smith, and H. Murakami, 2018: Projection of landfalling–tropical cyclone rainfall in the eastern United States under anthropogenic warming. J. Climate, 31, 7269–7286, https://doi.org/10.1175/JCLI-D-17-0747.1.
Lonfat, M., F. D. Marks Jr., and S. S. Chen, 2004: Precipitation distribution in tropical cyclones using the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager: A global perspective. Mon. Wea. Rev., 132, 1645–1660, https://doi.org/10.1175/1520-0493(2004)132<1645:PDITCU>2.0.CO;2.
Maloney, E. D., and Coauthors, 2019: Process-oriented evaluation of climate and weather forecasting models. Bull. Amer. Meteor. Soc., 100, 1665–1686, https://doi.org/10.1175/BAMS-D-18-0042.1.
Manabe, S., J. L. Holloway Jr., and H. M. Stone, 1970: Tropical circulation in a time-integration of a global model of the atmosphere. J. Atmos. Sci., 27, 580–613, https://doi.org/10.1175/1520-0469(1970)027<0580:TCIATI>2.0.CO;2.
Manganello, J. V., and Coauthors, 2012: Tropical cyclone climatology in a 10-km global atmospheric GCM: Toward weather-resolving climate modeling. J. Climate, 25, 3867–3893, https://doi.org/10.1175/JCLI-D-11-00346.1.
Moon, Y., and Coauthors, 2020a: Azimuthally averaged wind and thermodynamic structures of tropical cyclones in global climate models and their sensitivity to horizontal resolution. J. Climate, 33, 1575–1595, https://doi.org/10.1175/JCLI-D-19-0172.1.
Moon, Y., D. Kim, S. J. Camargo, A. A. Wing, K. A. Reed, M. F. Wehner, and M. Zhao, 2020b: A new method to construct a horizontal resolution‐dependent wind speed adjustment factor for tropical cyclones in global climate model simulations. Geophys. Res. Lett., 47, e2020GL087528, https://doi.org/10.1029/2020GL087528.
Murakami, H., and Coauthors, 2015: Simulation and prediction of category 4 and 5 hurricanes in the high-resolution GFDL HiFLOR coupled climate model. J. Climate, 28, 9058–9079, https://doi.org/10.1175/JCLI-D-15-0216.1.
Nolan, D. S., Y. Moon, and D. P. Stern, 2007: Tropical cyclone intensification from asymmetric convection: Energetics and efficiency. J. Atmos. Sci., 64, 3377–3405, https://doi.org/10.1175/JAS3988.1.
Patricola, C. M., and M. F. Wehner, 2018: Anthropogenic influences on major tropical cyclone events. Nature, 563, 339–346, https://doi.org/10.1038/s41586-018-0673-2.
Peters, O., and J. D. Neelin, 2006: Critical phenomena in atmospheric precipitation. Nat. Phys., 2, 393–396, https://doi.org/10.1038/nphys314.
Price, J. F., 1981: Upper ocean response to a hurricane. J. Phys. Oceanogr., 11, 153–175, https://doi.org/10.1175/1520-0485(1981)011<0153:UORTAH>2.0.CO;2.
Reed, K. A., and C. Jablonowski, 2011: Assessing the uncertainty in tropical cyclone simulations in NCAR’s Community Atmosphere Model. J. Adv. Model. Earth Syst., 3, M08002, https://doi.org/10.1029/2011MS000076.
Roberts, M., 2017a: MOHC HadGEM3-GC31-LM model output prepared for CMIP6 HighResMIP, version 20170906. Earth System Grid Federation, accessed September 2019, https://doi.org/10.22033/ESGF/CMIP6.1321.
Roberts, M., 2017b: MOHC HadGEM3-GC31-MM model output prepared for CMIP6 HighResMIP, version 20180818. Earth System Grid Federation, accessed September 2019, https://doi.org/10.22033/ESGF/CMIP6.1902.
Roberts, M., 2017c: MOHC HadGEM3-GC31-HM model output prepared for CMIP6 HighResMIP, version 20170831. Earth System Grid Federation, accessed September 2019, https://doi.org/10.22033/ESGF/CMIP6.446.
Roberts, M., 2019a: CMIP6 HighResMIP: Tropical storm tracks as calculated by the TRACK algorithm. Centre for Environmental Data Analysis, accessed September 2019, http://catalogue.ceda.ac.uk/uuid/0b42715a7a804290afa9b7e31f5d7753.
Roberts, M., 2019b: CMIP6 HighResMIP: Tropical storm tracks as calculated by the TempestExtremes algorithm. Centre for Environmental Data Analysis, accessed September 2019, http://catalogue.ceda.ac.uk/uuid/438268b75fed4f27988dc02f8a1d756d.
Roberts, M., and Coauthors, 2015: Tropical cyclones in the UPSCALE ensemble of high-resolution global climate models. J. Climate, 28, 574–596, https://doi.org/10.1175/JCLI-D-14-00131.1.
Roberts, M., and Coauthors, 2018: The benefits of global high resolution for climate simulation: Process understanding and the enabling of stakeholder decisions at the regional scale. Bull. Amer. Meteor. Soc., 99, 2341–2359, https://doi.org/10.1175/BAMS-D-15-00320.1.
Roberts, M., and Coauthors, 2019: Description of the resolution hierarchy of the global coupled HadGEM3-GC3.1 model as used in CMIP6 HighResMIP experiments. Geosci. Model Dev., 12, 4999–5028, https://doi.org/10.5194/gmd-12-4999-2019.
Roberts, M., and Coauthors, 2020a: Impact of model resolution on tropical cyclone simulation using the HighResMIP–PRIMAVERA multimodel ensemble. J. Climate, 33, 2557–2583, https://doi.org/10.1175/JCLI-D-19-0639.1.
Roberts, M., and Coauthors, 2020b: Projected future changes in tropical cyclones using the CMIP6 HighResMIP multimodel ensemble. Geophys. Res. Lett., 47, e2020GL088662, https://doi.org/10.1029/2020GL088662.
Schubert, W. H., and J. J. Hack, 1982: Inertial stability and tropical cyclone development. J. Atmos. Sci., 39, 1687–1697, https://doi.org/10.1175/1520-0469(1982)039<1687:ISATCD>2.0.CO;2.
Scoccimarro, E., P. G. Fogli, K. A. Reed, S. Gualdi, S. Masina, and A. Navarra, 2017a: Tropical cyclone interaction with the ocean: The role of high-frequency (subdaily) coupled processes. J. Climate, 30, 145–162, https://doi.org/10.1175/JCLI-D-16-0292.1.
Scoccimarro, E., A. Bellucci, and D. Peano, 2017b: CMCC CMCC-CM2-HR4 model output prepared for CMIP6 HighResMIP. Earth System Grid Federation, accessed August 2019, https://doi.org/10.22033/ESGF/CMIP6.1359.
Scoccimarro, E., A. Bellucci, and D. Peano, 2017c: CMCC CMCC-CM2-VHR4 model output prepared for CMIP6 HighResMIP. Earth System Grid Federation, accessed August 2019, https://doi.org/10.22033/ESGF/CMIP6.1367.
Scoccimarro, E., S. Gualdi, A. Bellucci, D. Peano, A. Cherchi, G. A. Vecchi, and A. Navarra, 2020: The typhoon-induced drying of the Maritime Continent. Proc. Natl. Acad. Sci. USA, 117, 3983–3988, https://doi.org/10.1073/pnas.1915364117.
Shaevitz, D. A., and Coauthors, 2014: Characteristics of tropical cyclones in high-resolution models in the present climate. J. Adv. Model. Earth Syst., 6, 1154–1172, https://doi.org/10.1002/2014MS000372.