Abstract

A comprehensive analysis of tropical cyclone (TC) intensity change in a warming climate is undertaken with high-resolution (6- and 2-km grid spacing) idealized simulations using the Weather Research and Forecasting (WRF) model. With the goal of isolating the influence of thermodynamic aspects of climate change on maximum hurricane intensity, an idealized TC is placed within a quiescent, horizontally uniform tropical environment computed from averaged reanalysis data for the tropical Atlantic Ocean. The analyzed tropical environment is used for control simulations. Changes between the periods 1990–99 and 2090–99 are computed using output from 13 GCMs from the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4), for the A1B, A2, and B1 emissions scenarios. These changes are then added to the reanalysis-derived initial and boundary conditions used in the control simulations. Some processes known to impact TC intensity, such as environmental vertical wind shear and sea surface wake cooling, are not considered in this study. Future TC intensity increased for 75 of 78 future simulations using 6-km grid length, with a 9% (~8 hPa) average increase in central surface-pressure deficit. For the 2-km simulations, the average increase was 14% (~14 hPa). The depth of the TC secondary circulation increases in future simulations, consistent with an increase in the height of the freezing level and tropopause. Inner-core precipitation increases of 10%–30% are found for future simulations, with large sensitivity to the emission scenario. The increase in precipitation is consistent with a stronger potential vorticity tower, a warmer eye, and lower central pressure. Enhanced upper-tropospheric warming in the GCM environment is shown to be an important mitigating influence on TC intensity change but is also shown to exhibit large uncertainty in GCM projections.

1. Introduction

In the absence of external factors detrimental to tropical cyclone (TC) intensification (e.g., vertical wind shear, dry air entrainment, oceanic upwelling, land interactions, etc.) the maximum intensity of a TC generally increases as the sea surface temperature (SST) increases (e.g., DeMaria and Kaplan 1994; Emanuel 1986, 1988, 1995; Holland 1997). This relationship between TC intensity and SST suggests that a future increase in SST could potentially support increased intensity of TCs, provided that other climate processes do not offset this effect. This possibility was raised by Emanuel (1987), who used a theoretical model of TC intensity to propose that TCs in a greenhouse gas–warmed climate would have higher potential intensities than in the present-day climate. Since this original inquiry, a large number of additional studies have pursued this topic, utilizing a variety of approaches.

Projections of future global climate change are often made with the aid of coupled atmosphere–ocean general circulation models (GCMs). Because of the large computational expense associated with long-term integrations, GCMs are required to use horizontal resolution that is unable to realistically simulate important storm-scale physical processes that determine TC intensity (e.g., storm-scale variations in turbulent fluxes, secondary circulation, and eye–eyewall processes), in part stemming from the need to parameterize subgrid-scale convection. For coarse GCM simulations, cumulus parameterization (CP) implicitly accounts for subgrid-scale precipitation, parameterizing a portion of the TCs secondary circulation, degrading its realism and impacting TC intensity as well (e.g., Davis and Bosart 2002). Because of these limitations, tropical disturbances in these models are often referred to as being “TC like” in that the modeled storms are larger in size and weaker than those observed. Recently, GCM projections with grid spacings of less than 30 km have provided evidence for increasing TC intensity in a warmed climate (e.g., Oouchi et al. 2006; Bengtsson et al. 2007); however, still higher resolution is required to reproduce observed TC intensities. Bengtsson et al. (2007) specifically found that model projections with increased resolution predicted a larger increase in the frequency of intense TCs in future climates, highlighting the need for high-resolution models, with explicit convection, in studies of future TC intensity change.

Using dynamical downscaling approaches, numerical models utilizing observed SSTs and atmospheric reanalyses have been able to reproduce year-to-year variability in Atlantic hurricane counts for 1980–2006 (Knutson et al. 2007). Shorter integration lengths and a smaller domain allowed runs with substantially higher resolution (18-km grid spacing) than full GCM projections. A similar approach was later adopted by Knutson et al. (2008) who used late twenty-first-century changes simulated by an ensemble of the World Climate Research Program (WCRP) Coupled Model Intercomparison Project 3 (CMIP3) models, under Intergovernmental Panel on Climate Change (IPCC) emissions scenario A1B. Although finding a reduction in the total number of tropical storms and hurricanes, these simulations indicate an increase in the frequency and intensity of the strongest hurricanes, and an increase in near-hurricane rainfall rates. Bender et al. (2010) further extended the modeling approach by downscaling each individual model storm from the 18-km simulations. Utilizing ensemble-mean projected climate changes, an increase in the number of category 4 and 5 hurricanes in the future climate was simulated, although results were highly sensitive to projected shear changes.

Another approach to downscaling (hereafter referred to as idealized downscaling) involves model simulations of an individual idealized TC embedded in a simplified large-scale environment with no other disturbances and no land (e.g., Knutson and Tuleya 1999, 2004; Knutson et al. 2001; Shen et al. 2000). Large-scale conditions, including SST, atmospheric moisture, temperature, and wind, are based on observational analyses or model data, and future environmental conditions are specified by GCM projections. Knutson and Tuleya (1999) demonstrated the viability of this methodology by showing that simulated TC intensification in the future climate was similar using either high-resolution regional model simulations or the downscaling technique. They found a simulated increase in surface wind speed of 3–7 m s−1 (5%–11%), a decrease in central pressure of 7–24 hPa, and an increase in near-storm precipitation of 28% in future storms. The impact of ocean coupling on simulated TC intensity change was found to be minimal by Knutson et al. (2001), who identified statistically significant increases in maximum surface wind speed of 5%–6% along with a 20% increase in rainfall within 100 km of the TC center. Knutson and Tuleya (2004) tested the sensitivity of the results to the choice of GCM environmental conditions and the CP scheme used in high-resolution TC simulations. Nearly all combinations of large-scale conditions and CP schemes produced an increase in both simulated storm intensity and near-storm precipitation rates in future TCs. Averaged over all simulations, a 14% increase in central pressure fall, a 6% increase in maximum surface wind speed, and an 18% increase in averaged precipitation rate within 100 km of the TC center was found in high-CO2 experiments relative to the control.

Shen et al. (2000) studied the effect of thermodynamic environmental changes on hurricane intensity. They utilized large-scale values of temperature and moisture based on observations, and examined sensitivity to changes in the SST and/or vertical temperature profile. Increasing the SST while maintaining a fixed vertical temperature profile always led to more intense TCs as anticipated. Alterations to the environmental temperature profile that led to less (more) stable lapse rates tended to produce higher (lower) hurricane intensity; the most intense hurricanes were found with warmer SSTs and less stable tropospheric lapse rates. As stated by the authors, in the tropics SST and tropospheric stability are related, with GCM output indicating increases in both SST and tropospheric stability in future climate projections. Increases in SST and tropospheric stabilization consistent with the GCM output led to simulated intensity increases of 7–8 hPa. In the absence of the tropospheric stabilization, however, the same SST increase led to an intensity increase of roughly double this amount.

The goal of this study is to build on the results of previous idealized downscaling studies and further examine the impact of climate change on maximum TC intensity and structure. Specifically, we examine the impact of thermodynamic changes in the tropical environment (changes in SST, atmospheric temperature, and moisture) on TC structure and intensity. The impact of changes in wind shear will not be addressed; future shear change remains uncertain, with GCMs indicating regional increases in monthly-mean wind shear although the increase is not significant in the main development region (MDR; e.g., Vecchi and Soden 2007; Talgo 2009). Regardless of future shear changes, the most intense TCs tend to form in regions and at times of low wind shear; therefore, for the study of the most intense TCs the omission of future shear changes is warranted.

Previous idealized downscaling studies are extended here in several ways. Many of these studies utilize grid spacing of at least , and more often ⅙°, usually requiring the use of CP. Here, high-resolution (6- and 2-km grid length) TC simulations eliminate sensitivity associated with CP, and allow full TC intensity to be simulated. A more detailed investigation of the changes in the structure and processes in the simulated TCs is possible with these high-resolution simulations, allowing a direct examination of TC structural changes and identification of the physical mechanisms responsible for the changes. The sensitivity of the results to model surface and planetary boundary layer parameterization is also investigated, as the parameterization scheme choice has been shown to impact simulated TC intensity and structure (e.g., Braun and Tao 2000; Davis and Bosart 2002; Hill and Lackmann 2009a; Nolan et al. 2009a,b). An ensemble of GCM projections from three different emissions scenarios allows examination of the sensitivity to environmental changes, along with identification of sources of uncertainty from the GCM output that are important to the TC-intensity problem.

Section two contains a more detailed description of the methods used. Section 3 provides an overview of changes in the large-scale environment projected by the GCMs, while section 4 presents results concerning TC intensity changes. Section 5 focuses on changes in TC structure, and section 6 provides further analysis and a concluding discussion.

2. Experimental design and methods

a. Large-scale environments

Whereas some previous studies utilized GCM output with fixed CO2 values for current climate conditions (i.e., the control simulation), environmental current climate values in this study were computed using September 1990–99 2.5° National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data (Kalnay et al. 1996), averaged over the region extending from 8.5°–15°N, 40°–60°W in the western portion of the tropical Atlantic tropical cyclone main development region (MDR). This region was chosen as it is representative of the conditions that accompany TC activity in the Atlantic basin. Spatial and temporal averaging dampens diurnal and transient variations; the resulting horizontally uniform fields are not highly sensitive to the details of this process (not shown). The horizontally uniform domain SST was calculated from averages of daily high-resolution blended SST analyses derived from Advanced Very High Resolution Radiometer (AVHRR) infrared satellite SST data, available on a 0.25° grid (Reynolds et al. 2007).

Future climate conditions were determined using GCM simulations conducted for the IPCC Fourth Assessment Report (AR4) with temperature, moisture, and SST data available for the climate of the twentieth-century experiment and for twenty-first century projections using the A1B, B1, and A2 emissions scenarios. These data were obtained from the WCRP CMIP3 multimodel dataset [see Meehl et al. (2007) for more information]. The emissions scenarios, described in detail in the IPCC special report on emissions scenarios (Nakicenovic and Swart 2000), provide a range of estimates of greenhouse gas emissions, and vary as a result of different projections of population growth, societal change, and technological advance. These data were available for 13 different GCMs, listed in Table 1. The projected change in each aforementioned variable was calculated as the difference between the 2090–99 mean and the 1990–99 mean, averaged spatially for the region specified previously. The 1990–99 average values were computed using the twentieth-century simulations for each GCM. Future climate conditions were calculated by summing the GCM projected changes in each variable with the reanalysis-derived current climate average.

Table 1.

CMIP3 I.D., modeling group and country, and atmospheric grid spacing for GCMs that contained necessary data fields for the twentieth-century simulation and future simulations with the A1B, A2, and B1 emissions scenarios.

CMIP3 I.D., modeling group and country, and atmospheric grid spacing for GCMs that contained necessary data fields for the twentieth-century simulation and future simulations with the A1B, A2, and B1 emissions scenarios.
CMIP3 I.D., modeling group and country, and atmospheric grid spacing for GCMs that contained necessary data fields for the twentieth-century simulation and future simulations with the A1B, A2, and B1 emissions scenarios.

The benefit to this approach is that while each GCM may have a mean climate state that deviates from observations, the use of the projected change eliminates this individual model sensitivity, leaving only the change due to increasing greenhouse gas concentrations and other time-varying forcing agents. While the realism of GCM-derived change projections can be questioned because of difficulties reproducing ENSO, the ITCZ, and other important circulation patterns, the use of the GCM ensemble allows quantification of this variability. As described in more detail in subsequent sections, the Weather Research and Forecasting (WRF) model is used to perform high-resolution TC simulations using projected changes from individual GCMs or ensemble-mean changes for each emissions scenario, averaged over the 13 individual GCM runs for each emissions scenario.

b. TC initialization

An axisymmetric warm-core vortex, initially possessing maximum tangential winds of ~30 m s−1 and a minimum sea level pressure (MSLP) of ~981 hPa, was inserted within each of the horizontally uniform environments described in section 2a. Multiple model experiments indicated that the maximum intensity after a sufficiently long integration period was not sensitive to the strength of the initial vortex (not shown). The initialization procedure used here is similar to that described in Hill and Lackmann (2009a,b); the reader is referred to that study for details of the process, in which a TC is placed in a quiescent, horizontally uniform base-state environment. Outside of the influence of the initial vortex, the geopotential height, temperature, and moisture fields are horizontally uniform, yielding a calm environment favorable for TC intensification. No TC secondary circulation is initialized and relative humidity throughout the vortex is not adjusted from the large-scale values, and therefore a delay occurs during the model integration as the TC secondary circulation, cloud, and precipitation fields evolve.

c. TC model simulations

The WRF model, version 2.2 (Skamarock et al. 2007), was used to simulate idealized TCs in the different large-scale environments. The WRF model has been used extensively in TC forecasts and simulations (e.g., Davis et al. 2008; Hill and Lackmann 2009a,b), and has demonstrated the ability to simulate strong TCs when run at sufficient resolution (e.g., Fierro et al. 2009; Gentry and Lackmann 2010). With specified boundary conditions, the WRF model did not exhibit climate drift over the short-range simulations conducted; test simulations indicate that when the model simulation exceeded two weeks, this became a more significant issue (not shown). The fidelity with which this model is able to reproduce and predict TCs in the current environment, documented in the aforementioned studies, makes it a suitable choice for the study of TC changes in future environments. The model was run on a geophysical domain, without land (the TC is initially centered within the domain at 13°N), but with full model physics. An f plane valid at 13°N was utilized in order to minimize TC movement during the simulations and thereby reduce the impact of the lateral boundaries; additional experiments confirm that the use of an f plane has minimal impact TC maximum intensity or structure (not shown). The SST remained constant during the model integrations. A total of 78 model simulations utilized a domain with 6-km grid spacing (2400 × 2400 km2 in dimension), while 6 additional experiments were performed with a nested high-resolution inner domain (2-km grid spacing, 1016 × 1016 km2).

Simulations with only a 6-km grid were performed using projected changes from individual GCMs, and for each emissions scenario, with 2 combinations of model physics (described below) yielding a total of 78 future simulations. Additional experiments with a 2-km nest within a 6-km parent domain were performed using the current climate conditions or future values computed using the ensemble-mean projected changes from each of the 3 selected emissions scenarios, in order to better resolve structural differences between the simulations and investigate the sensitivity of projected intensity changes to model resolution. An additional experiment (referred to as no stable), designed to investigate the role of tropospheric stabilization on future TC intensity, was also conducted. In this simulation, A1B ensemble-mean projected changes in SST and moisture were utilized, while tropospheric temperature increases were fixed at the projected increase at the lowest GCM pressure level (1000 hPa).

In the vertical, 47 unevenly spaced layers were used, with a higher concentration in the boundary layer and a model top of 50 hPa. Single-domain model simulations were integrated for 5 days with output saved every 3 h, while nested-domain simulations were integrated for 10 days with output saved every hour. These simulation lengths were found to allow a quasi-steady maximum TC intensity to be attained, and the hourly output was beneficial in the construction of time-averaged quantities, used to analyze structural changes. Model simulations featured explicit convection (no CP), either the Mellor–Yamada–Janjic (MYJ) or Yonsei University (YSU) surface layer/PBL parameterization scheme, the WRF Single-Moment 6-Class Microphysics scheme (WSM6; Hong and Lim 2006), the rapid radiative transfer model (RRTM) longwave radiation scheme, and the Goddard shortwave radiation scheme. Additional model experiments have been performed to further assess the sensitivity of the results to the choice of surface layer–PBL and microphysics parameterization schemes, including six combinations of microphysics and surface layer–PBL parameterization schemes (Hill et al. 2008). Although simulated TC intensity is sensitive to model physics, these previous experiments (not shown) indicate that the two combinations of model physics used here best represent the mean change in TC intensity in future simulations relative to the control.1 Therefore, although it would be desirable to repeat the experiments performed here with additional combinations of model physics, experiments indicate that simulated changes in future TC intensity and structure shown here are not highly sensitive to model physics choice.

3. GCM projections

Before presenting GCM projected changes, it is important to discuss how anthropogenic forcings bring about changes in the tropical atmosphere. Changes in tropospheric lapse rate in the tropics are dependent on a number of physical processes, most of which are parameterized in climate models. Increases in CO2 concentration lead to an increase in downwelling infrared radiation to the surface, which subsequently causes a rise in surface temperature (or over the ocean; i.e., SST). Changes in the SST over the tropics in turn lead to tropospheric lapse rate changes, primarily through the vertical transfer of heat by convective processes (e.g., Rennick 1977; Stone and Carlson 1979; Sobel et al. 2002; Santer et al. 2005). As the tropical surface temperature increases, moist convection acts to stabilize the tropospheric lapse rate, resulting in strongest warming aloft due to latent heating. Increased tropospheric temperatures, combined with nearly constant relative humidity (e.g., Held and Soden 2006) leads to increases in water vapor, which provides an additional increase in downwelling infrared radiation and subsequent increases in surface temperature, the so-called water vapor feedback.

Stratospheric processes (see Ramaswamy et al. 2001 for a comprehensive summary) also impact temperature changes in the upper troposphere. Forster et al. (2007) found that cooling in the stratosphere (in the 70–30-hPa layer) due to ozone depletion leads to reduced downwelling longwave radiation, and the associated cooling extends downward into the upper troposphere (e.g., the 150–70-hPa layer). Cordero and Forster (2006) found that those GCMs from the IPCC AR4 that included stratospheric ozone depletion (11 of 19 models investigated) produced upper-tropospheric temperature trends that were significantly closer to observations than the GCMs that omit ozone variations, further suggesting the influence of stratospheric processes on temperatures in the upper troposphere. Furthermore, Son et al. (2009a,b) and Dall’Amico et al. (2010a,b) demonstrate the importance of ozone recovery in GCM representation of several aspects of climate change, including the tropical upper troposphere.

To summarize, based on the known response of the tropical atmosphere to CO2 forcing, it is anticipated that GCM-projected temperature changes in the tropics would tend to increase with height up to the tropopause. Model-to-model variability in projected temperature increases are expected to increase with height, as temperature changes in the upper troposphere are sensitive to a wide variety of model characteristics, including CP scheme, vertical resolution, and ozone treatment. Thorough analysis of the representation of these physical processes in the different GCMs may allow for a more informed decision concerning which GCMs to trust in projections of climate change, which in turn could increase confidence in projections of TC intensity change. Analysis of the individual models in light of the varying means in which they treat the critical processes according to Cordero and Forster (2006), along with comparison of the GCM output to chemical climate model (CCM) data obtained from E.C. Cordero (2009, personal communication), indicates that the IPCC GCM ensemble represents a reasonable estimate of the uncertainty in the tropical tropospheric temperature change. Rather than selecting GCMs based on some measure of accuracy, model resolution, or physics, we elect to utilize the ensemble suite of GCM output to assess the uncertainty in the projected changes of TC intensity and structure.

Figure 1 and Tables 2 and 3 provide a summary of the control climate conditions and ensemble-mean GCM-projected changes in the Atlantic MDR from each emissions scenario as a function of pressure. Each scenario features a similar profile of temperature change, with warming throughout the troposphere, maximized in the upper troposphere (near the 250-hPa level) at more than double that near the surface (1000 hPa). This increasing warming with height is similar to that found in previous studies (e.g., Knutson and Tuleya 1999, 2004), and, as previously described, is a consequence of moist convective adjustment. Stratospheric cooling (of up to 8 K at the 10-hPa level) is projected in each emissions scenario, although additional model experiments (not shown) and Shen et al. (2000) indicate that stratospheric cooling has a relatively small influence on TC intensity relative to tropospheric changes. The GCMs generally indicate little change in tropospheric relative humidity in this location, which, combined with increasing temperatures, lead to an increase in atmospheric water vapor. GCM simulations from each emissions scenario indicate an increase in atmospheric moisture throughout the troposphere, with the largest increases near the surface (Fig. 1b). The maximum near-surface moistening in the A2 scenario (3.7 g kg−1) is more than double that in the B1 scenario (1.8 g kg−1) due to the larger warming in A2 and the nonlinear relationship between saturation vapor pressure and temperature.

Fig. 1.

GCM ensemble-mean projected change in spatially and temporally averaged (a) temperature (K) and (b) mixing ratio (g kg−1) for the following emissions scenarios: B1 (blue), A1B (red), and A2 (green). Change is calculated as the difference between the 2090–99 mean and the 1990–99 mean, spatially averaged over the region extending over 8.5°–15°N and 40°–60°W.

Fig. 1.

GCM ensemble-mean projected change in spatially and temporally averaged (a) temperature (K) and (b) mixing ratio (g kg−1) for the following emissions scenarios: B1 (blue), A1B (red), and A2 (green). Change is calculated as the difference between the 2090–99 mean and the 1990–99 mean, spatially averaged over the region extending over 8.5°–15°N and 40°–60°W.

Table 2.

Temperature (K) values for the current climate and 13-member ensemble-mean future projected changes from each emissions scenario.

Temperature (K) values for the current climate and 13-member ensemble-mean future projected changes from each emissions scenario.
Temperature (K) values for the current climate and 13-member ensemble-mean future projected changes from each emissions scenario.
Table 3.

Mixing ratio (g kg−1) values for the current climate and 13-member ensemble-mean future projected changes from each emissions scenario.

Mixing ratio (g kg−1) values for the current climate and 13-member ensemble-mean future projected changes from each emissions scenario.
Mixing ratio (g kg−1) values for the current climate and 13-member ensemble-mean future projected changes from each emissions scenario.

Focusing on individual GCM projections of temperature change from each emissions scenario, it is evident that there is considerable model-to-model variability (Fig. 2). The GCMs as a whole demonstrate a high correlation between the maximum tropospheric warming and the increase in SST (Fig. 3, correlation coefficient of ~0.9), indicating that as SST increases the average lapse rate through the depth of the troposphere decreases, and net lapse-rate stabilization results. 2 As indicated by a close examination of Fig. 2, however, the vertical level at which individual GCMs project the largest warming varies between 250 and 150 hPa. This discrepancy is evident in the standard deviation of projected warming (Fig. 2d), which exhibits an increase with height through the troposphere, reaching a maximum at the 150-hPa level. Overall, GCM projections using the A2 (B1) emissions scenario generally indicate the largest (smallest) increase in SST and in tropospheric water vapor and stability, with A1B generally falling in between. Within each emissions scenario there is considerable model-to-model variability, most notably for projected changes in upper-tropospheric temperature.

Fig. 2.

Individual GCM-projected change in spatially and temporally averaged atmosphere temperature (K) for the following emissions scenarios: (a) A1B, (b) A2, and (c) B1. Ensemble-mean, minimum, and maximum changes at each pressure level are indicated in the data table. Change is calculated as the difference between the 2090–99 mean and the 1990–99 mean, spatially averaged over the region extending over 8.5°–15°N and 40°–60°W. (d) The standard deviation among all 39 GCM projections with pressure.

Fig. 2.

Individual GCM-projected change in spatially and temporally averaged atmosphere temperature (K) for the following emissions scenarios: (a) A1B, (b) A2, and (c) B1. Ensemble-mean, minimum, and maximum changes at each pressure level are indicated in the data table. Change is calculated as the difference between the 2090–99 mean and the 1990–99 mean, spatially averaged over the region extending over 8.5°–15°N and 40°–60°W. (d) The standard deviation among all 39 GCM projections with pressure.

Fig. 3.

GCM-projected change (calculated as the difference between the 2090–99 mean and the 1990–99 mean) in SST (K) vs the maximum warming aloft (K) for each GCM from each emissions scenario.

Fig. 3.

GCM-projected change (calculated as the difference between the 2090–99 mean and the 1990–99 mean) in SST (K) vs the maximum warming aloft (K) for each GCM from each emissions scenario.

4. TC intensity changes

a. Simulations with 6-km grid spacing

Maximum TC intensity was assessed by examining the minimum central pressure (MCP) of the simulated TC at each 3-hourly output time. For each combination of model physics, future climate simulations were compared to the control simulation that used the same model physics combination. The calculation of future intensity change was found to not be sensitive to temporal averaging of MCP values (not shown). Hereafter an “increase” in TC intensity means that the TC in the simulation with GCM-projected changes is more intense (a lower MCP) than that in the control simulation. Changes in maximum surface wind speed are generally smaller than those in central pressure, owing to the nonlinear increase in the frictional force with wind speed and also the manner in which turbulent momentum fluxes are computed in the WRF model (Hill and Lackmann 2009a). Simulations with the MYJ surface–planetary boundary layer scheme tend to be more intense than those with YSU, consistent with Hill and Lackmann (2009a) and likely due to larger values of surface latent heat flux, although the future intensity increase relative to the respective control runs is similar (Fig. 4). For reference, the MCP in the YSU (MYJ) control simulation is 940 (928) hPa. Combined results for all 78 model runs indicate that although 75 of 78 (~96%) future climate simulations indicate an increase in TC intensity, considerable spread exists (Fig. 4c). Two simulations indicate intensity increases of greater than 18%, and three simulations (~4%) exhibit future weakening. Runs based on the environmental changes provided by one GCM (the IPSL model) yield two of the three future simulations with reduced TC intensity; this GCM also featured the largest tropospheric stabilization (not shown). The largest increase in TC intensity is found when altering the large-scale conditions based on the A1B projection of the Bjerknes Centre for Climate Research (BCCR) GCM, with a 22% increase in central pressure deficit. The collective results from the 6-km simulations indicate that the most likely increase in future TC intensity is 9%–12%.

Fig. 4.

Frequency diagrams illustrating the percentage change in central pressure deficit in future climate simulations (consisting of all emissions scenarios) with 6-km grid spacing, for (a) MYJ WSM6 simulations, (b) YSU WSM6 simulations, and (c) all 6-km simulations.

Fig. 4.

Frequency diagrams illustrating the percentage change in central pressure deficit in future climate simulations (consisting of all emissions scenarios) with 6-km grid spacing, for (a) MYJ WSM6 simulations, (b) YSU WSM6 simulations, and (c) all 6-km simulations.

Averaged over the individual GCMs, the B1 emissions scenario produces the smallest increase in TC intensity (a 6-hPa decrease in MSLP, or a 8% increase in pressure deficit relative to the ambient environment), while A1B and A2 produce similar increases in TC intensity (8 hPa, 10%). These intensity increases are actually closer than would be anticipated just based on average SST increase (1.4° and 2.7°C in the B1 and A2 scenarios, respectively) due to larger lapse rate stabilization in A2 projections (both the ensemble mean, and in each individual model projection). These average percentage increases in pressure deficit are slightly less than the average increase of 14% found by Knutson and Tuleya (2004), although there is a considerable range in the individual simulations, and results with 2-km grid spacing (shown subsequently) indicate greater future intensification.

b. Simulations with 2-km grid spacing

Ideally, simulations with 2-km grid spacing would be conducted using the projected changes from each individual GCM, as was done for simulations with 6-km grid spacing. Computing limitations, however, make this number of 2-km simulations unfeasible, and instead 6-km simulations with 2-km nested domains have been conducted that utilize the ensemble-mean projected changes from each emissions scenario. Output from the 2-km future simulations is compared with a 2-km control simulation in order to provide a consistent evaluation. Additional simulations with 6-km grid spacing and ensemble-mean projected atmospheric and SST changes produce similar results to those described previously, after averaging over TC simulations with individual GCM projected changes; this demonstrates that a comparison between simulated changes in TC intensity in 2-km simulations driven from ensemble-mean initial and boundary conditions with the mean values from the 6-km simulations is meaningful.

Compared with the 6-km simulations, the 2-km simulations are more intense (Table 4), consistent with previous studies suggesting an increase in simulated TC intensity with increasing resolution (e.g., Hill and Lackmann 2009a; Gentry and Lackmann 2010). Averaged over each model physics combination, increases in central pressure deficit of 11%, 12%, and 19% are found in the B1, A1B, and A2 simulations, respectively. Stabilization of the troposphere reduces the future intensification by approximately half; specifically, simulations with no stabilization reached a MCP of 893 hPa.

Table 4.

Summary of maximum intensity and future climate intensity changes in model simulations.

Summary of maximum intensity and future climate intensity changes in model simulations.
Summary of maximum intensity and future climate intensity changes in model simulations.

5. Structure changes

The 2-km simulations are best suited for analysis of structural changes and diagnosis of the physical mechanisms associated with the intensity increases that were previously described in section 4.

a. Rainfall

The typical hourly precipitation pattern for the simulated TC in the current environment is provided in Fig. 5. Although a detailed comparison with observations is not provided, the rain rate in the simulated TC eyewall for the current environment is qualitatively comparable to that found in observational estimates (e.g., Jiang et al. 2006), and model simulations (e.g., Houze et al. 2007) of strong hurricanes. Higher eyewall precipitation rates are generally found in simulations using the MYJ parameterization scheme, consistent with larger moisture fluxes at high wind speed (e.g., Hill and Lackmann 2009a), and also in the future climate simulations.

Fig. 5.

A comparison of hourly rainfall accumulation (h−1) in control simulations with 2-km grid spacing ending at simulation hour 120 for the following surface–boundary layer parameterization and large-scale environment: (left) YSU and (right) MYJ.

Fig. 5.

A comparison of hourly rainfall accumulation (h−1) in control simulations with 2-km grid spacing ending at simulation hour 120 for the following surface–boundary layer parameterization and large-scale environment: (left) YSU and (right) MYJ.

To quantitatively analyze rainfall rates during the entire simulation period, the frequency of model grid cells within 500 km of TC center possessing certain rainfall rates (x axis) as a function of simulation hour (y axis) is displayed in Fig. 6. There are 196 281 grid cell centers within 500 km of the TC center, and therefore the 0.01% contour corresponds roughly to 20 grid cells. During the first 48 h of the simulations the rainfall rates increase, and upon reaching maturity the rainfall distribution is quasi-steady, changing little between hours 48 and 240 (Fig. 6a). To assess the change in future TC rainfall, Figs. 6b displays the change in the number of grid cells (relative to the control) with certain hourly rainfall rates. In all future simulations there is an increase in the number of grid cells with heavier rainfall rates, consistent with increased atmospheric water vapor and moisture convergence. Averaged over the last 5 days of the simulations within various distances from TC center, there is a considerable increase in precipitation near the TC center in the future simulations (Fig. 7). Utilizing the 100-km averaging distance as in Knutson and Tuleya (2004), increases in average rainfall of 19%, 13%, and 11% are found in the A1B, A2, and B1 simulations, respectively. The values from simulations with A1B projected changes are similar to Knutson and Tuleya (2004), 18%, although changes in the eye size influence area averages over such a small radius. Averaged over a 200-km distance, the percentage increases in precipitation are larger, reaching up to 23% in the A2 simulation. Analysis of 300-km area-averaged quantities (not shown) reveals that the increase in rainfall is mainly due to increased atmospheric water vapor, with changes in the vertical velocity accounting for the remaining difference.

Fig. 6.

(a) Contoured frequency by time diagram of hourly rainfall rate (in. h−1) within 500 km of TC center for the control simulation. (b) Change in number of grid cells with certain hourly rainfall rates, relative to the control simulation, for A1B. Only rainfall rates exceeding 2 in. h−1 are shown in order to highlight changes in heavy rainfall rate occurrence.

Fig. 6.

(a) Contoured frequency by time diagram of hourly rainfall rate (in. h−1) within 500 km of TC center for the control simulation. (b) Change in number of grid cells with certain hourly rainfall rates, relative to the control simulation, for A1B. Only rainfall rates exceeding 2 in. h−1 are shown in order to highlight changes in heavy rainfall rate occurrence.

Fig. 7.

Percentage increase (relative to the control simulation) in area-averaged rainfall rate within various distances of the TC center averaged over the entire simulation period in TC simulations with 2-km grid spacing.

Fig. 7.

Percentage increase (relative to the control simulation) in area-averaged rainfall rate within various distances of the TC center averaged over the entire simulation period in TC simulations with 2-km grid spacing.

b. Secondary circulation strength

Changes in the thermal structure of the tropospheric environment may impact the TC secondary circulation, although to our knowledge previous work has not investigated this possibility in detail. Dynamically, one aspect of the thermal structure that impacts TC structure is the height of the freezing level (Fig. 8). Relative to the control, the outer core freezing level increases by ~700, ~800, and ~400 m in the A1B, A2, and B1 simulations, respectively. The upward shift of the freezing level in future simulations is larger in the TC core than in the ambient environment due to enhanced latent heat release, with increases in the height of the freezing level at 30-km radius of ~1100 m in the A1B and A2 simulations, and ~600 m in the B1 simulation. A contoured frequency by altitude diagram (CFAD)3 of vertical velocity between hours 216 and 240 (Fig. 9) agrees fairly well with the previous observational study of Black et al. (1996), and recent high-resolution modeling studies of observed TCs (e.g., Rogers et al. 2007; Fierro et al. 2009; Gentry and Lackmann 2010). Focusing first on upward motion, maximum updrafts tend to occur below the freezing level, at roughly 4.5-km altitude in the control simulation, and between 5.5 and 6 km in future simulations. The increase in height of maximum updrafts in future simulations is roughly equivalent to the increase in height of the freezing level in the eyewall. Maximum updraft speeds are generally comparable in all simulations, reaching ~10 m s−1. Above the level of maximum updrafts, stronger upward motion is found in the future simulations relative to the control up to an altitude of 18 km.

Fig. 8.

Azimuthally averaged freezing level, time-averaged between simulation hours 216 and 240. Simulation indicated in legend.

Fig. 8.

Azimuthally averaged freezing level, time-averaged between simulation hours 216 and 240. Simulation indicated in legend.

Fig. 9.

CFAD of vertical velocity (m s−1) averaged between simulation hours 216 and 240 for (a) control simulation and (b) A1B. In diagrams for the future TC simulations, the 0.01% contour from the control simulation is shown for comparison.

Fig. 9.

CFAD of vertical velocity (m s−1) averaged between simulation hours 216 and 240 for (a) control simulation and (b) A1B. In diagrams for the future TC simulations, the 0.01% contour from the control simulation is shown for comparison.

Quantitative examination of radial inflow and outflow (not shown) demonstrates a slight increase in the strength of the secondary circulation in the future simulations, along with an upward shift in the outflow altitude in the future simulations.

c. Inflow and outflow temperatures

Changes in temperature of inflow and outflow could potentially impact the maximum intensity of future TCs, as there is a well-established link between TC maximum potential intensity and temperatures difference between the inflow and outflow layers (e.g., Emanuel 1986). Cross sections of azimuthally averaged temperature and outflow velocity (Fig. 10) demonstrate that while outflow in future simulations occurs at higher altitudes, it also warmer, as the increased SSTs and subsequent tropospheric stabilization have led to warmer temperatures in the upper troposphere and an increase in the height of the tropopause. To better quantify changes in the TC thermodynamics, inward and outward mass flux was binned by temperature (bin width of 4°C), and normalized by the total flux. Focusing first on outflow (Figs. 11a,b), it is evident that in the control simulation a larger fraction of the outflowing mass occurs at colder temperatures, as tropospheric stabilization in future simulations leads to a larger fraction of the outflowing mass occurring at warmer temperatures. Specifically, the percentage of outflowing mass that is colder than −59°C is ~95% in the control simulations and ~65% in the A2 simulation, illustrating that the projected climate changes lead to warmer TC outflow (Fig. 11b). Inflow and temperature cross sections (not shown) indicate that inflow is collocated with warmer temperatures in the future simulations than in the control simulation, as anticipated because of the SST increase.

Fig. 10.

Cross section of outflow (contoured, m s−1) and temperature (shaded, °C) averaged between simulation hours 216 and 240 for (a) control simulation and (b) A1B.

Fig. 10.

Cross section of outflow (contoured, m s−1) and temperature (shaded, °C) averaged between simulation hours 216 and 240 for (a) control simulation and (b) A1B.

Fig. 11.

(a) Normalized outward mass flux (%) averaged over simulation hours 216–240 as a function of temperature (°C) and (b) cumulative normalized outward mass flux. (c) As in (a), but for inward mass flux. Values are restricted to inflow in the 0–2-km layer and outflow in the 10–20-km layer.

Fig. 11.

(a) Normalized outward mass flux (%) averaged over simulation hours 216–240 as a function of temperature (°C) and (b) cumulative normalized outward mass flux. (c) As in (a), but for inward mass flux. Values are restricted to inflow in the 0–2-km layer and outflow in the 10–20-km layer.

To further quantify thermodynamic changes in the simulated TCs, Table 5 provides the mass-weighted inflow and outflow temperatures along with the thermodynamic efficiency, averaged over simulation hours 216–240. The future increase in mass-weighted inflow temperature is roughly comparable to the increase in SST. Overall, the thermodynamic efficiency is similar in the control and A1B simulations, but is reduced in the A2 and B1 future simulations. This reduction in thermodynamic efficiency is attributable to a warmer mass-weighted outflow temperature, and seemingly contradictory to the increased intensity. A larger number of model simulations would be desirable in order to further test the robustness of changes in thermodynamic efficiency; however, changes in thermodynamic efficiency do not alone influence changes in TC intensity. Despite the lower thermodynamic efficiency in the A2 simulation, the larger amount of tropospheric water vapor and heavier precipitation in this simulation allows for stronger latent heat release and a stronger precipitation mass sink effect (Lackmann and Yablonsky 2004). These effects are consistent with increased diabatic potential vorticity generation, and subsequently a more intense TC. The potential vorticity will be discussed further in section 5e.

Table 5.

Mass-weighted average inflow temperature, outflow temperature, and thermodynamic efficiency, averaged over simulation hours 216–240.

Mass-weighted average inflow temperature, outflow temperature, and thermodynamic efficiency, averaged over simulation hours 216–240.
Mass-weighted average inflow temperature, outflow temperature, and thermodynamic efficiency, averaged over simulation hours 216–240.

d. Eye temperature anomalies

Hydrostatic pressure in the TC center is related to the layer-average temperature in the overlying air column; thus, we now examine time–height diagrams of the eye-temperature anomaly, where the anomaly is computed as the difference in temperature between the TC core and the original large-scale environment (Fig. 12). The TC core here is defined as those grid cells that are centered within 2 km of the grid point with the lowest sea level pressure. The eye structure follows a similar evolution in all simulations; the temperature anomaly structure from the control simulation is shown in Fig. 12a. The initial weak warm core that extends between 3 and 12 km gradually reaches a mature quasi–steady state after simulation hour 84. The largest positive temperature anomalies in the eye are generally found at an altitude of 10–14 km, where temperatures exceed that of the ambient environment by greater than 14°C. To highlight climate change influences, Figs. 12b displays difference fields between the future simulations and the control. It is evident that in each future simulation the eye temperature anomaly is warmer in the 14–18-km layer, indicating that the warm eye extends deeper than in the control simulation, consistent with the increased height of updrafts and outflow.

Fig. 12.

(a) Eye temperature anomalies (K; relative to the initial ambient environment) as a function of time in the control simulation. (b) Eye temperature anomalies (K) relative to the control simulation temperature anomalies for A1B.

Fig. 12.

(a) Eye temperature anomalies (K; relative to the initial ambient environment) as a function of time in the control simulation. (b) Eye temperature anomalies (K) relative to the control simulation temperature anomalies for A1B.

e. Potential vorticity

With increased precipitation and latent heat release, it is anticipated that the future simulations would produce a stronger diabatic PV tower. Although there are several notable differences in the potential vorticity structure in the different simulations, the focus here is on differences in the interior cyclonic PV tower. This cyclonic PV is generated by latent heat release, with the rate of diabatic PV generation being related to the projection of the heating gradient onto the absolute vorticity vector (e.g., Raymond 1992; Stoelinga 1996). The interior cyclonic PV tower is stronger in future simulations than in the control experiment (Fig. 13), with the strength of the lower-tropospheric portion of the PV tower strongest in the A2 and A1B simulations. This aligns with the TC intensity results, and suggests that the mechanism for stronger TCs in the future is associated with increased diabatic heating and PV production. As discussed by Lackmann and Yablonsky (2004), the precipitation mass sink effect also contributes to PV production. A PV budget would be required to determine the relative importance of increased latent heating and the mass sink effect to the strengthened PV tower in the future simulations. Volume-averaged PV correlates strongly with area-averaged rainfall (Fig. 14a), which in turn is highly correlated with area-averaged sea level pressure (Fig. 14b). These correlations are also found in simulations with 6-km grid spacing, and these points are presented in Fig. 14 as well.

Fig. 13.

Cross section of time-averaged (simulation hours 216–240) azimuthally averaged potential vorticity (shaded; PVU), potential temperature (black contours, K), and the 0°C isotherm (blue contour) for (a) control simulation and (b) A1B. Cross sections are west–east and extend 3° of longitude on either side of the TC center.

Fig. 13.

Cross section of time-averaged (simulation hours 216–240) azimuthally averaged potential vorticity (shaded; PVU), potential temperature (black contours, K), and the 0°C isotherm (blue contour) for (a) control simulation and (b) A1B. Cross sections are west–east and extend 3° of longitude on either side of the TC center.

Fig. 14.

Scatterplot of change (relative to control) in (a) volume-averaged PV (inside a radius of 200 km and below an altitude of 6 km) with area-averaged rainfall inside a radius of 200 km. (b) As in (a), but for volume-averaged PV with area-averaged SLP.

Fig. 14.

Scatterplot of change (relative to control) in (a) volume-averaged PV (inside a radius of 200 km and below an altitude of 6 km) with area-averaged rainfall inside a radius of 200 km. (b) As in (a), but for volume-averaged PV with area-averaged SLP.

6. Discussion and conclusions

In this study, the thermodynamic impact of anthropogenic climate change on maximum TC intensity is investigated. The methodology used here is a variation on previous techniques: utilizing analyzed data to represent the average environment in which current TCs form; GCM output to assess twenty-first-century changes in SST, air temperature, and moisture; and a high-resolution mesoscale model (WRF) to simulate idealized TCs. Previous work was extended in this study by utilizing a larger number of GCMs forced with three different greenhouse gas emissions scenarios to estimate climate change, which allowed for a detailed analysis of uncertainty. The TC simulations featured higher resolution than in previous idealized downscaling studies, and the explicit convection simulations allowed for a more realistic representation of TC structure and analysis of TC structure changes in a future climate. The high-resolution model output was used to investigate structural changes, and to explore the mechanism of future intensity changes.

Before assessing results, it is important to emphasize the applicability of this study to observed TCs and to also identify some assumptions that could potentially impact the results. The idealized initial conditions, devoid of atmospheric features that inhibit TC intensification, may only exist for brief periods and in isolated locations during a hurricane season. Therefore, this approach is essentially designed to examine changes in the maximum intensity of TCs during time periods which are favorable for strong TC development, and does not address changes in the frequency or geographical distribution of the TC climatology. The impact of future changes in vertical wind shear on TC intensity is not considered here. Although GCMs do indicate trends in average shear over parts of the tropical Atlantic b, it is likely that periods of weak shear would still take place, and thus future shear changes may not influence the intensity of the strongest TCs that would develop during these favorable periods. Furthermore, Vecchi and Soden (2007) analyze trends in Atlantic shear in the IPCC GCMs, and do not find a significant change over the western portion of the Atlantic development region. The impact of oceanic upwelling is not included in this study, although Knutson et al. (2001) suggest that the net impact of ocean coupling on simulated CO2-induced TC intensification is relatively minor.

Based on an ensemble of state-of-the-art GCMs, the average large-scale changes in the environment over a subsection of the Atlantic MDR during the twenty-first century can be summarized as an increase in SST of ~1.0°C to ~3.5°C, near-surface atmospheric warming of slightly more than the SST increase, and tropospheric warming that increases with height with maximum warming of approximately double the SST increase typically found between 250–150 hPa. In each emissions scenario, there is large model-to-model variability, especially in the projections of upper-tropospheric temperature change. This uncertainty is attributable to processes parameterized in GCMs, including convection and treatment of ozone, and other model attributes, including vertical resolution.

High-resolution TC simulations were performed using the WRF model with initial and boundary conditions derived from spatially and temporally averaged reanalysis data, and including an idealized incipient vortex. This control simulation was compared to otherwise identical runs, but with GCM-derived SST, temperature, and moisture changes added to the control initial and boundary conditions. Future climate change predictions were based on individual GCMs for a set of 78 simulations with 6-km grid length, and ensemble-mean values from each emissions scenario for a set of 6 simulations run with a 2-km grid length.

Simulation results with 6-km grid spacing indicate an increase in TC intensity in 75 of the 78 future climate simulations relative to the control (Fig. 4). The bulk of the 6-km simulations exhibited an increased pressure deficit in the 4%–8%, 8%–12%, or 12%–16% bin, relative to the control simulation. Averaged over all simulations with 6-km grid spacing, future TCs had an increase in central pressure deficit of 9%. Future TC intensity increases were found to be sensitive to emissions scenario with average increases in central pressure deficit of 10%, 11%, and 5% found in future simulations with A1B, A2, and B1 emissions scenarios, respectively. The central pressure deficit increases in A1B and A2 simulations of 10% and 11% are similar but slightly less than the 14% found by Knutson and Tuleya (2004). Increases in future TC central pressure deficit of 11%, 19%, and 12% were found in simulations with 2-km grid spacing using the A1B, A2, and B1 emissions scenarios ensemble-mean projected changes, or 13% averaged over all 2-km simulations.

Overall, results presented here are consistent with previous studies and indicate a likely increase in the intensity of the strongest future TCs. The change in TC intensity found in the future simulations is linked to both projected changes in the atmosphere and ocean; tropospheric lapse-rate stabilization, present in all the GCM projections, plays a role in offsetting the larger increase in intensity that would occur solely based on the projected SST change. Specifically, simulations with 2-km grid spacing indicate that stabilization reduces future intensification by approximately 50%, consistent with Shen et al. (2000). Future weakening in a small number of future simulations indicates that an increase in tropospheric stability can completely negate the intensity increase that would occur due to a modest increase in SST, highlighting the importance of the balance between SST increase and tropospheric stabilization. Furthermore, the large standard deviation in the strength of upper-tropospheric warming between the GCM runs reduces confidence in this important aspect of the projected changes in TC intensity.

Changes in TC structural characteristics were also investigated using high-resolution WRF simulations with 2-km grid spacing. The average TC rainfall was found to increase in the future climate simulations, with increases in average rainfall within 100 km of the TC center of approximately 19%, 13%, and 11% in simulations with A1B, A2, and B1 projected changes, respectively. These increases are comparable to the 18% found by Knutson and Tuleya (2004), and demonstrate that the increase in rainfall is tied to projected increases in atmospheric water vapor. Overall, the increase in rainfall in future TC simulations is consistent with the greater intensity, despite the tropospheric stabilization seen in the GCM projections. This is consistent with a stronger PV tower and greater diabatic PV production and perhaps also to an increased precipitation mass sink effect (Lackmann and Yablonsky 2004).

Analysis of the TC secondary circulation was performed in order to assess how CO2-induced changes would impact the strength and location of updrafts, downdrafts, and radial inflow and outflow. Updrafts are of similar maximum strength in all simulations, but maximum updrafts are higher in altitude and updrafts extend higher in future simulations than in the control. These increases are likely associated with the increase in the height of the freezing level and tropopause, and also increased buoyancy due to higher water vapor content in future climate simulations. Outflow, while occurring at higher altitudes in the future simulations, is warmer than in the control simulation, partially offsetting the increased thermodynamic efficiency that would occur solely due to the increase in inflow temperature. Calculated thermodynamic efficiency, relative to the control simulation, is not larger in future simulations, contradictory to the simulated intensity increase. The increase in precipitation (related to water vapor increases) in the future climate is consistent with a stronger cyclonic PV tower, warmer eye, and thus a more intense TC despite the lack of increase in thermodynamic efficiency.

Acknowledgments

This research was supported by DOE Grant ER64448 and NSF Grant ATM-0334427, both awarded to North Carolina State University. The WRF model was made available through NCAR, which is sponsored by the NSF. Model simulations were performed at the Renaissance Computing Institute (RENCI), which is supported by UNC Chapel Hill, NCSU, Duke University, and the state of North Carolina. We thank Dr. Eugene Cordero and his graduate student Sium Tesfai of San Jose State University for discussing model-to-model sensitivity of upper-tropospheric and lower-stratospheric temperature changes, and for examining chemical GCM output in the tropical region. Dr. Walt Robinson of NC State University is acknowledged for providing valuable comments on an earlier version of the manuscript. We also thank the Program for Climate Model Diagnosis and Intercomparison (PCMDI) for collecting and archiving the CMIP3 model output, and the WCRP’s Working Group on Coupled Modeling (WGCM) for organizing the model data analysis activity. The WCRP CMIP3 multimodel dataset is supported by the Office of Science, U.S. Department of Energy.

REFERENCES

REFERENCES
Bender
,
M. A.
,
T. R.
Knutson
,
R. E.
Tuleya
,
J. J.
Sirutis
,
G. A.
Vecchi
,
S. T.
Garner
, and
I. M.
Held
,
2010
:
Modeled impact of anthropogenic warming on the frequency of intense Atlantic hurricanes
.
Science
,
327
,
454
458
.
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
.
Black
,
M. L.
,
R. W.
Burpee
, and
F. D.
Marks
,
1996
:
Vertical motion characteristics of tropical cyclones determined with airborne Doppler radial velocities
.
J. Atmos. Sci.
,
53
,
1887
1909
.
Braun
,
S. A.
, and
W.-K.
Tao
,
2000
:
Sensitivity of high-resolution simulation of Hurricane Bob (1991) to planetary boundary layer parameterizations
.
Mon. Wea. Rev.
,
128
,
3941
3961
.
Cordero
,
E. C.
, and
P. M.
de F. Forster
,
2006
:
Stratospheric variability and trends in models used for the IPCC AR4
.
Atmos. Chem. Phys.
,
6
,
5369
5380
.
Dall’Amico
,
M.
,
L. J.
Gray
,
K. H.
Rosenlof
,
A. A.
Scaife
,
K. P.
Shine
, and
P. A.
Stott
,
2010a
:
Stratospheric temperature trends: Impact of ozone variability and the QBO
.
Climate Dyn.
,
34
,
381
398
.
Dall’Amico
,
M.
,
P. A.
Stott
,
A. A.
Scaife
,
L. J.
Gray
,
K. H.
Rosenlof
, and
A. Y.
Karpechko
,
2010b
:
Stratospheric temperature trends: Impact of ozone variability and the QBO
.
Climate Dyn.
,
34
,
399
417
.
Davis
,
C.
, and
L. F.
Bosart
,
2002
:
Numerical simulations of the genesis of Hurricane Diana (1984). Part II: Sensitivity of track and intensity prediction
.
Mon. Wea. Rev.
,
130
,
1100
1124
.
Davis
,
C.
, and
Coauthors
,
2008
:
Prediction of landfalling hurricanes with the advanced hurricane WRF model
.
Mon. Wea. Rev.
,
136
,
1990
2005
.
DeMaria
,
M.
, and
J.
Kaplan
,
1994
:
Sea surface temperature and the maximum intensity of Atlantic tropical cyclones
.
J. Climate
,
7
,
1324
1334
.
Emanuel
,
K. A.
,
1986
:
An air-sea interaction theory for tropical cyclones. Part I: Steady-state maintenance
.
J. Atmos. Sci.
,
43
,
585
604
.
Emanuel
,
K. A.
,
1987
:
The dependence of hurricane intensity on climate
.
Nature
,
326
,
483
485
.
Emanuel
,
K. A.
,
1988
:
The maximum intensity of hurricanes
.
J. Atmos. Sci.
,
45
,
1143
1155
.
Emanuel
,
K. A.
,
1995
:
Sensitivity of tropical cyclones to surface exchange coefficients and a revised steady-state model incorporating eye dynamics
.
J. Atmos. Sci.
,
52
,
3969
3976
.
Fierro
,
A. O.
,
R. F.
Rogers
,
F. D.
Marks
, and
D. S.
Nolan
,
2009
:
The impact of horizontal grid spacing on the microphysical and kinematic structures of strong tropical cyclones simulated with the WRF-ARW model
.
Mon. Wea. Rev.
,
137
,
3717
3741
.
Forster
,
P. M.
,
G.
Bodeker
,
R.
Schofield
,
S.
Solomon
, and
D.
Thompson
,
2007
:
Effects of ozone cooling in the tropical lower stratosphere and upper troposphere
.
Geophys. Res. Lett.
,
34
,
L23813
,
doi:10.1029/2007GL031994
.
Gentry
,
M. S.
, and
G.
Lackmann
,
2010
:
Sensitivity of simulated tropical cyclone structure and intensity to horizontal resolution
.
Mon. Wea. Rev.
,
138
,
688
704
.
Held
,
I. M.
, and
B. J.
Soden
,
2006
:
Robust responses of the hydrological cycle to global warming
.
J. Climate
,
19
,
5686
5699
.
Hill
,
K. A.
, and
G. M.
Lackmann
,
2009a
:
Analysis of idealized tropical cyclone simulations using the Weather Research and Forecasting model: Sensitivity to turbulence parameterization and grid spacing
.
Mon. Wea. Rev.
,
137
,
745
765
.
Hill
,
K. A.
, and
G. M.
Lackmann
,
2009b
:
Influence of environmental humidity on tropical cyclone size
.
Mon. Wea. Rev.
,
137
,
3294
3315
.
Hill
,
K. A.
,
G. M.
Lackmann
, and
A.
Aiyyer
,
2008
:
Model simulated changes in TC intensity due to global warming
.
Preprints, 28th Conf. on Hurricanes and Tropical Meteorology, Orlando, FL, Amer. Meteor. Soc., 7B.2. [Available online at http://ams.confex.com/ams/pdfpapers/138181.pdf.]
.
Holland
,
G. J.
,
1997
:
The maximum potential intensity of tropical cyclones
.
J. Atmos. Sci.
,
54
,
2519
2541
.
Hong
,
Y.-Y.
, and
J.-O. J.
Lim
,
2006
:
The WRF single-moment 6-class microphysics scheme (WSM6)
.
J. Korean Meteor. Soc.
,
42
,
129
151
.
Houze
,
R. A.
,
S. S.
Chen
,
B. F.
Smull
,
W.-C.
Lee
, and
M. M.
Bell
,
2007
:
Hurricane intensity and eyewall replacement
.
Science
,
315
,
1235
1239
,
doi:10.1126/science.1135650
.
Jiang
,
H.
,
P. G.
Black
,
E. J.
Zipser
,
F. D.
Marks
, and
E. W.
Uhlhorn
,
2006
:
Validation of rain-rate estimation in hurricanes from the stepped frequency microwave radiometer: Algorithm correction and error analysis
.
J. Atmos. Sci.
,
63
,
252
267
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Knutson
,
T. R.
, and
R. E.
Tuleya
,
1999
:
Increased hurricane intensities with CO2-induced warming as simulated using the GFDL hurricane prediction system
.
Climate Dyn.
,
15
,
503
519
.
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
.
Knutson
,
T. R.
,
R. E.
Tuleya
,
W.
Shen
, and
I.
Ginis
,
2001
:
Impact of CO2-induced warming on hurricane intensities as simulated in a hurricane model with ocean coupling
.
J. Climate
,
14
,
2458
2468
.
Knutson
,
T. R.
,
J. J.
Sirutis
,
S. T.
Garner
,
I. M.
Held
, and
R. E.
Tuleya
,
2007
:
Simulation of the recent multidecadal increase of Atlantic hurricane activity using an 18-km-grid regional model
.
Bull. Amer. Meteor. Soc.
,
88
,
1549
1565
.
Knutson
,
T. R.
,
J. J.
Sirutis
,
S. T.
Garner
,
G. A.
Vecchi
, and
I. M.
Held
,
2008
:
Simulated reduction in Atlantic hurricane frequency under twenty-first century warming conditions
.
Nat. Geosci.
,
1
,
359
364
.
Lackmann
,
G. M.
, and
R. M.
Yablonsky
,
2004
:
The importance of the precipitation mass sink in tropical cyclones and other heavily precipitating systems
.
J. Atmos. Sci.
,
61
,
1674
1692
.
Meehl
,
G. A.
,
C.
Covey
,
T.
Delworth
,
M.
Latif
,
B.
McAvaney
,
J. F. B.
Mitchell
,
R. J.
Stouffer
, and
K. E.
Taylor
,
2007
:
The WCRP CMIP3 multimodel dataset: A new era in climate change research
.
Bull. Amer. Meteor. Soc.
,
88
,
1383
1394
.
Nakicenovic
,
N.
, and
R.
Swart
, Eds.,
2000
:
Emission scenarios
.
Special Report on Emissions Scenarios, Cambridge University Press, 239–292
.
Nolan
,
D. S.
,
J. A.
Zhang
, and
D. P.
Stern
,
2009a
:
Evaluation of planetary boundary layer parameterizations in tropical cyclones by comparison of in situ observations and high-resolution simulations of Hurricane Isabel (2003). Part I: Initialization, maximum winds, and outer-core boundary layer
.
Mon. Wea. Rev.
,
137
,
3651
3674
.
Nolan
,
D. S.
,
J. A.
Zhang
, and
D. P.
Stern
,
2009b
:
Evaluation of planetary boundary layer parameterizations in tropical cyclones by comparison of in situ observations and high-resolution simulations of Hurricane Isabel (2003). Part II: Inner-core boundary layer and eyewall structure
.
Mon. Wea. Rev.
,
137
,
3675
3698
.
Oouchi
,
K.
,
J.
Yoshimura
,
H.
Yoshimura
,
R.
Mizuta
,
S.
Kusunoki
, and
A.
Noda
,
2006
:
Tropical cyclone climatology in a global-warming climate as simulated in a 20 km-mesh global atmospheric model: Frequency and wind intensity analyses
.
J. Meteor. Soc. Japan
,
84
,
259
276
.
Ramaswamy
,
V.
, and
Coauthors
,
2001
:
Stratospheric temperature trends: Observations and model simulations
.
Rev. Geophys.
,
39
,
72
122
.
Raymond
,
D. J.
,
1992
:
Nonlinear balance and potential-vorticity thinking at larger Rossby number
.
Quart. J. Roy. Meteor. Soc.
,
118
,
987
1015
.
Rennick
,
M. A.
,
1977
:
The parameterization of tropospheric lapse rates in terms of surface temperature
.
J. Atmos. Sci.
,
34
,
854
862
.
Reynolds
,
R. W.
,
T. M.
Smith
,
C.
Liu
,
D. B.
Chelton
,
K. S.
Casey
, and
M. G.
Schlax
,
2007
:
Daily high-resolution-blended analyses for sea surface temperature
.
J. Climate
,
20
,
5473
5496
.
Rogers
,
R. F.
,
M. L.
Black
,
S. S.
Chen
, and
R. A.
Black
,
2007
:
An evaluation of microphysics fields from mesoscale model simulations of tropical cyclones. Part I: Comparisons with observations
.
J. Atmos. Sci.
,
64
,
1811
1834
.
Santer
,
B. D.
, and
Coauthors
,
2005
:
Amplification of surface temperature trends and variability in the tropical atmosphere
.
Science
,
309
,
1551
1556
.
Shen
,
W.
,
R. E.
Tuleya
, and
I.
Ginis
,
2000
:
A sensitivity study of the thermodynamic environment on GFDL model hurricane intensity: Implications for global warming
.
J. Climate
,
13
,
109
121
.
Skamarock
,
W. C.
,
J. B.
Klemp
,
J.
Dudhia
,
D. O.
Gill
,
D. M.
Barker
,
W.
Wang
, and
J. G.
Powers
,
2007
:
A description of the advanced research WRF Version 2
.
NCAR Tech. Note NCAR/TN-468+STR, 88 pp
.
Sobel
,
A. H.
,
I. M.
Held
, and
C. S.
Bretherton
,
2002
:
The ENSO signal in tropical tropospheric temperature
.
J. Climate
,
15
,
2702
2706
.
Son
,
S.-W.
,
N. F.
Tandon
,
L. M.
Polvani
, and
D. W.
Waugh
,
2009a
:
Ozone hole and Southern Hemisphere climate change
.
Geophys. Res. Lett.
,
36
,
L15705
,
doi:10.1029/2009GL038671
.
Son
,
S.-W.
, and
Coauthors
,
2009b
:
The impact of stratospheric ozone recovery on tropopause height trends
.
J. Climate
,
22
,
429
445
.
Stoelinga
,
M. T.
,
1996
:
A potential vorticity–based study of the role of diabatic heating and friction in a numerically simulated baroclinic cyclone
.
Mon. Wea. Rev.
,
124
,
849
874
.
Stone
,
P. H.
, and
J. H.
Carlson
,
1979
:
Atmospheric lapse rate regimes and their parameterization
.
J. Atmos. Sci.
,
36
,
415
423
.
Talgo
,
K. D.
,
2009
:
Tropical Atlantic vertical wind shear variability in a future climate. M.S. thesis, Dept. of Marine, Earth, and Atmospheric Sciences, North Carolina State University, 97 pp
.
Vecchi
,
G. A.
, and
B. J.
Soden
,
2007
:
Increased tropical Atlantic wind shear in model projections of global warming
.
Geophys. Res. Lett.
,
34
,
L08702
,
doi:10.1029/2006GL028905
.

Footnotes

1

Experiments demonstrate that some combinations of model parameterization do not work well together (e.g., the combination of the MYJ PBL scheme with the Lin microphysics scheme yields unrealistically strong TCs). The choices used here are most representative of the overall average of TC intensity.

2

The increase in lower-tropospheric water vapor actually results in a future increase in convective available potential energy (CAPE); we use the term “lapse-rate stabilization” to refer to the stabilization due to maximized warming in the upper troposphere.

3

In Fig. 9 and several subsequent figures, only the A1B results are shown for brevity. For the displayed quantities, the changes are consistently largest for the A2 scenario, and smallest for B1, with A1B in the middle.