1. Introduction
Assessments from the Intergovernmental Panel on Climate Change (IPCC) show that the frequency of tropical cyclone (TC) genesis in the North Atlantic Ocean basin will be reduced under future climate conditions (Bender et al. 2010; Villarini and Vecchi 2013; Walsh et al. 2016). Despite fewer TCs, those that do develop will likely be more intense, owing to warmer sea surface temperatures (SSTs) (Villarini et al. 2010; Ramsay and Sobel 2011; Villarini and Vecchi 2012; Sobel et al. 2016; Walsh et al. 2016; and references therein). Much effort has been put forth to understand the effects of climate change on the tracks and intensities of TCs, as well as on the coastal impacts such as rainfall, surge, and winds (e.g., Emanuel et al. 2008; Mendelsohn et al. 2012; Emanuel 2013; Scoccimarro et al. 2013; Villarini and Vecchi 2013; Emanuel 2017).
However, little research has focused on the effects of climate change on the inland impacts of landfalling TCs, namely, inland flooding and tornadoes. This is an important shortcoming, given that (i) the number of tropical cyclone tornadoes (TCTs) generated per TC event tends to depend on the TC intensity (Weiss 1985; Novlan and Gray 1974; Gentry 1983; McCaul 1991; Verbout et al. 2007; Moore and Dixon 2011; Edwards 2012; Rhodes and Senkbeil 2014), implying a possible future increase in TCT generation and (ii) TC rain rates are likely to increase in the future, with rain rates scaling with a 7% per degree Celsius increase of water vapor in future climates (Knutson et al. 2020). In fact, flooding is historically one of the largest contributors to TC fatalities (Rappaport 2014), and recent TC events, (e.g., Hurricane Harvey 2017), were shown to be wetter than they otherwise would have been without the recent climate change (Emanuel 2017). Thus, the work herein is largely motivated by the possibility of these enhanced inland TC threats will limit safe evacuation options for coastal evacuees (and inland communities) who might inadvertently place themselves (or are already located) in the paths of damaging tornadoes or severe flooding. Specifically, it is hypothesized that anthropogenic climate change will enhance the environment conducive for intense rainfall and tornadoes with TCs, thus resulting in more inland flooding and tornado damage from landfalling TCs.
Because of the relatively small scale of TCTs and their parent thunderstorms, one challenge in addressing this hypothesis is the relatively coarse resolution of existing global climate models (GCMs), and the computational expense of dynamical downscaling at high resolutions over large regions necessary to resolve such features. Recent studies have used high-resolution dynamical downscaling to investigate severe-convective-scale processes (e.g., Hoogewind et al. 2017; Trapp et al. 2011; see also Prein et al. 2015). A number of studies have additionally used the pseudo–global warming (PGW) approach (e.g., Sato et al. 2007; Lynn et al. 2009; Lackmann 2015; Trapp and Hoogewind 2016; Prein et al. 2017a; Patricola and Wehner 2018).
The PGW method evolved out of the surrogate warming method introduced by Schär et al. (1996) and Frei et al. (1998). The PGW method involves the simulation of a historical weather or climate event under its original time-dependent 3D meteorological forcing [the control (CTRL)], which is then compared with a simulation in which the meteorological forcing has been modified by a “climate-change delta.” Lackmann (2015) applied the PGW method to simulate Hurricane Sandy (2012) with climate-change deltas derived from GCMs in phase 3 of the Coupled Model Intercomparison Project (CMIP3). He found that in a past climate Hurricane Sandy was weaker and tracked farther south than observed whereas in a future climate (under the A2 climate-change scenario) Sandy had a statistically significant increase in intensity and a more northward track. To address extreme tornadic events in future climates, Trapp and Hoogewind (2016) applied the PGW method using data from three GCMs in phase 5 of CMIP (CMIP5) under representative concentration pathway 8.5 (RCP8.5). They found that some storms had significantly more intense updrafts and stronger low-level vertical vorticity in the future environment while others failed to initiate because of increased convective inhibition and reduced forcing. It was confirmed that the future thermodynamic environment had a large impact on tornadic storms.
Herein, the basic method used in these key studies will be applied to Hurricane Ivan (2014) and is described in section 2. The results are presented in two parts, following the framework put forth by Sillmann et al. (2017), to advance understanding of extreme events in future climates. The first part in section 3 explores the effect of future climates on TC intensity and structure, and the second part in section 4 explores the effect of future climates on TC rainfall and TCT-surrogate (TCT-S) generation. Concluding remarks are given in section 5.
2. Methods
a. Case study Hurricane Ivan 2004
Hurricane Ivan (2004) serves as the historical case in this study. In brief, Ivan made landfall as a category-3 (Simpson 1974) hurricane in Orange Beach, Alabama, at 0650 UTC 16 September 2004. Ivan also produced up to 17 in. (43 cm) of rainfall, resulting in widespread inland and coastal flooding that caused 5 of the 25 total deaths in the contiguous United States (Stewart 2011), and a record-breaking 118 tornadoes (Baker et al. 2009; Edwards 2010), responsible for 7 of the 25 total deaths. Additional details of the environment favorable for supercell and tornado development in Hurricane Ivan can be found in Stewart (2011), Baker et al. (2009) and references therein. The present study seeks to understand whether such extreme TCT-generation as well as inland rainfall would be enhanced in a future climate.
b. WRF Model configuration
Hurricane Ivan’s landfall and overland progression are simulated using the Advanced Research core of the Weather Research and Forecasting (WRF) Model (Skamarock et al. 2008) over the interval from 0000 UTC 14 September to 0000 UTC 19 September 2004. The computational domain consists of an outer domain (D01) with 9-km grid spacing and a nested inner domain (D02) with 3-km grid spacing; both domains are simulated over the entire 5-day interval (see Fig. 1). Note that tornadoes are not explicitly resolved on a 3-km horizontal grid and thus that the forthcoming analysis does not consider tornado strength and is limited to TCT generation only. Initial, lateral, and lower boundary conditions for the simulation are from the 6-hourly National Centers for Environmental Prediction Final (NCEP FNL) Operational Global Analysis, provided on a 1° by 1° grid, at the surface and at 26 mandatory levels from 1000 to 10 hPa (NOAA/NCEP 2000).
WRF Model domain: outer domain (D01) with 9-km grid spacing centered about the CONUS, and inner nest (D02) with 3-km grid spacing centered about the eastern region of the United States.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
The WRF Model is configured for the CTRL and primary PGW simulations as follows. The Kain–Fritsch (Kain 2004) scheme is used to parameterize cumulus convection on D01 only. On all domains, the updated Rapid Radiative Transfer Model (RRTMG; Iacono et al. 2008) scheme parameterizes longwave and shortwave radiation; the Eta surface layer scheme and Noah land surface model (Chen and Dudhia 2001) are used for surface layer interactions and surface physics, respectively; the Mellor–Yamada–Janjić (MYJ; Janjić 1994) scheme parameterizes planetary boundary layer (PBL) (hereinafter, PBL2) processes; and the Thompson scheme (Thompson et al. 2008) (hereinafter, MP8) is used to parameterize the cloud and precipitation physics. The Thompson microphysics is well tested across many studies that use WRF to simulate regional climate and tropical cyclones, in particular (e.g., Raktham et al. 2015; Gutmann et al. 2018). Following the suggestion of Lackmann (2015), 6-hourly sea surface temperature updates are used to improve the model representation of TC intensity, as was the Garratt formulation of “isftcflx” (option 2; Lackmann 2015). This formulation provides more realistic values of heat and moisture by correcting surface bulk drag and enthalpy coefficients (see Green and Zhang 2013; Parker et al. 2018, and references therein on details for isftcflx). To test the sensitivity of TCT-S generation to changes in physical parameterizations, additional experiments are conducted using the WRF single moment 6-class microphysics scheme (WSM6; Hong and Lim 2006) (hereinafter, MP experiment with MP6) and the Yonsei University (YSU; Hong et al. 2006) PBL scheme (hereinafter, PBL experiment with PBL1) (Table 1).
Summary of WRF Model experiments for the CTRL and PGW simulations.
c. Global climate models
Output used in this study comes from three coupled atmosphere–ocean GCM simulations in CMIP5: Community Climate System Model, version 4 (NCAR); Model for Interdisciplinary Research on Climate, version 5 (MIROC); and Geophysical Fluid Dynamics Laboratory Climate Model, version 3 (GFDL). These models were chosen because of the range in their representations of historical TC genesis in the North Atlantic Ocean. For example, Tory et al. (2013) found that NCAR exhibited a low bias in TC genesis in the North Atlantic upon comparing historical simulations with observational data. In contrast, MIROC and GFDL were found to have a high and moderate genesis bias, respectively. With regard to TC genesis in the future climate, NCAR showed low TC activity whereas MIROC and GFDL showed no substantial changes in future TC activity in the North Atlantic for the period 2070–2100 (Tory et al. 2013).
d. Pseudo–global warming application
Using these three GCMs, climate-change deltas are calculated from 10-yr monthly means of the following (2D and 3D) thermodynamic variables: temperature, relative humidity, SST, and pressure, and kinematic variables: u and υ components of the wind. The 10-yr means for September are determined for both a historical period (1980–90) and a future period (2080–90). The future time period assumes greenhouse gas emissions described by RCP8.5, which represents the worst-case or “business as usual” scenario wherein radiative forcing will steadily rise to 8.5 W m−2. These deltas are interpolated to the NCEP FNL horizontal grid and vertical levels and then added to the initial and time-dependent boundary conditions from the NCEP FNL. Table 1 lists all experiments conducted.
Examples of deltas in SST/skin temperature, the u component of the 500-hPa wind, and 500-hPa geopotential heights are shown in Fig. 2. The GFDL GCM had the largest change in skin surface temperature, with at least a 7.5-K delta over 80% of the contiguous United States (CONUS). The MIROC and NCAR GCMs exhibited similar behaviors, except with warming of about 5 K. All three GCMs show reduced zonal wind values in the lower half of CONUS (Fig. 2), indicative of weaker eastward steering flow, and reduced vertical shear in future climates. The GFDL GCM shows the largest decrease in zonal winds (~4 m s−1) in the eastern half of CONUS. Positive deltas in 500-hPa geopotential height are consistent with the positive surface temperature deltas. The GFDL GCM produced the largest height increases of approximately 150 m over 90% of CONUS, consistent with the strongest surface warming. The height deltas imposed on the initial state suggest changes to the initial strength and locations of the midlevel trough and ridge that differ among the GCMs. These changes also help to interpret the midlevel zonal wind changes.
For (left) NCAR, (center) GFDL, and (right) MIROC, the September mean climate-change deltas in (a)–(c) SST and skin temperature (K), (d)–(f) the u component of the 500-hPa wind (m s−1), and (g)–(i) the 500-hPa geopotential height (m) together with the absolute values of 500-hPa height (dashed white contours) from the initial conditions of CTRL. These contributed to the initial and boundary conditions for the PGW simulations.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
e. TCT surrogates
As previously mentioned, tornadoes are not explicitly resolved on a 3-km horizontal grid, and thus their identification in the simulations requires diagnosis of a proxy, or surrogate. TCT-S are identified on the D02 (3-km) domain using a gridpoint-based exceedance algorithm defined by simulated radar reflectivity (REF) and maximum updraft helicity (UHmax); UH is widely used in severe weather applications to identify supercells and associated hazards (Kain et al. 2008) (see Carroll-Smith et al. 2019). A UHmax threshold of 83 m2 s−2 is used [which represents a 99.95 percentile value, as determined by Carroll-Smith et al. (2019)], as is a REF threshold of 30 dBZ. These thresholds produced the best spatial and quantitative representation of TCT-Ss in the model simulations as compared to tornado reports (see section 3b in Carroll-Smith et al. 2019). The UH is calculated over a 2–5-km layer (see Kain et al. 2008). This choice of layer is based on tests showing that UH calculations over a 0–3-km layer overproduced TCT-Ss as compared with tornado reports, whereas calculations over a 1–4-km layer were essentially the same as those over a 2–5-km layer. The 30-dBZ REF threshold was chosen to ensure that only convective storms were considered. In addition, Carroll-Smith et al. (2019) conducted simulations at 1-km grid spacings and used these to demonstrate that 3-km grid spacing is sufficient to represent the bulk properties of the potentially tornadic cells in TC environments.
3. Part 1: TC track, intensity, structure, and evolution
In the following, simulated TCs are tracked using the location of the lowest mean sea level pressure every 6 h within a 200-km box. Track lengths are then calculated using the great circle distance between each point and then taking the sum of all the line segments to get the total length. The start of the track is defined by the location of the TC at the first simulated forecast hour, and the track end is defined as the TC location of the final hour. Simulated TC intensity is determined using the lowest minimum pressure within 200 km of the storm center at each simulated forecast hour.
We focus first on the TCs in the primary simulations (1–4 in Table 1, CTRL, PGW: GFDL, MIROC, NCAR). The TC in the CTRL simulation made landfall at approximately 30.5°N, 88°W (near Dauphin Island) as a category-2 hurricane. Note from Carroll-Smith et al. (2019) that the simulated TC track and intensity in the CTRL simulation best matched the observed track and intensity over land where the TCT-S and rainfall analyses are conducted. The CTRL-simulated TC structure also matched well with the observations, capturing the overall shape and TC structure, including various convective modes (cluster, discrete cells, and linear systems) observed in the TC rainbands [see section 3a in Carroll-Smith et al. (2019) for additional details on the CTRL]. While we have not tested its statistical significance, the TCs in the PGW simulations all tracked west of the CTRL, with the TC in the GFDL simulation tracking the farthest west (Fig. 3), and the MIROC TC tracking closest to the CTRL. The TC in the GFDL simulation had the highest categorical intensity (based on the runtime maximum wind speed) as a midrange category-3 hurricane, followed by the TCs in the NCAR and MIROC simulations, which were high-end category-2 hurricanes. However, the MIROC TC was the most intense at the end of the simulation (see Fig. 3). The length of the TC tracks in the GFDL and NCAR simulations were noticeably shorter than that of the CTRL, whereas the TC track in the MIROC simulation was much longer than that of the CTRL; we will show later that the track length has implications on rainfall accumulation as well as TCT-S production.
CTRL (solid green), MIROC (yellow), GFDL (red), and NCAR (blue) TC (a) tracks and (b) intensity defined by minimum sea level pressure at 6-hourly time steps beginning at the 24th forecast hour.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
A subjective comparison of the REF across the simulations (Fig. 4) shows similarities in TC structure between the CTRL and PGW runs, however, the GFDL TC is slightly less symmetrical. Indeed, relative to the other simulations, the GFDL TC has the largest stratiform rain region at this time period. Figure 4 also reveals different modes of tornado producing convective cells within the rainbands, as described by Edwards et al. (2012). Discrete cells are found in the outer rainbands of the CTRL TC, discrete cells within a line are found in the NCAR TC, discrete cells in clusters and in linear convective systems are found in the GFDL TC, and finally, discrete cells within clusters are found in the MIROC TC. The presence of REF greater than 35 dBZ associated with these convective modes in the distant rainbands, and near the TC center, is also indicative of heavy rainfall with rainfall rates greater than 6 mm h−1.
Simulated radar reflectivity for (a) CTRL, (b) NCAR, (c) GFDL, and (d) MIROC runs, at 2300 UTC 15 Sep 2004, which is the time of the first TCT-S in the CTRL.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
The track, intensity, and structure (Figs. 5 and 6) of the simulated TCs in the two sensitivity experiments with different microphysics (5–12 in Table 1) are generally similar to those in the primary simulations, with the exception of the stratiform rain region, which is not as pronounced in the MP6 and PBL1 experiments (Fig. 6). Additional analyses, discussed in the section to follow, reveal that the TCT-Ss and rainfall are more sensitive to the PGW than to these changes in physical parameterization, thus giving us confidence in the robustness of the PGW response. To further confirm that changes in the PGW simulations are primarily due to the applied climate deltas rather than due to internal model variability, a test for model drift in the larger-scale environment was performed by comparing the CTRL to an analysis of the NCEP FNL 500-hPa geopotential heights (Fig. 7a). The results showed little drift of the simulated large-scale model environment from the reanalysis at the end of the 5-day simulation.
As in Fig. 3, but with the addition of the sensitivity experiments: primary runs (solid lines), PBL2_MP6 (dotted lines), and PBL1_MP8 (dash–dotted lines).
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
As in Fig. 4, but for (top) NCAR, (middle) GFDL, and (bottom) MIROC for the (a)–(c) primary PGW runs and (d)–(f) PGW PBL1 and (g)–(i) PGW MP6 experiments valid at 0000 UTC 16 Sep 2004.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
The 500-hPa height (dam) and absolute vertical vorticity (shaded; 10−5 s−1) for (a) CTRL, (b) NCAR, (c) GFDL, and (d) MIROC at 0000 UTC 19 Sep 2004, with the 500-hPa height from the NCEP FNL reanalysis data superimposed on the CTRL in red contours.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
Spectral nudging has been used in other studies to constrain the large scales within the model domain to resemble large scales in the driving data. Nudging can be beneficial but requires careful testing to avoid the potential for suppressing aspects of TC development. Molina et al. (2020) successfully applied nudging to constrain the large-scale environment for a study into sensitivities of a wintertime tornado outbreak. Moon et al. (2018) found that with careful tuning, nudging improved their TC track and intensity simulations. Wang et al. (2013) also found that nudging improved track simulation over the ocean through improved steering flow guidance but was detrimental to the response of the TC to complex terrain. They suggest nudging overamplifies the control of the large scale at low levels. Of course, this effect may be ameliorated by only nudging mid- and upper levels. On climate time scales nudging can improve TC frequency simulation (e.g., Choi and Lee 2016) and precipitation extremes (Otte et al. 2012), and appears not to suppress the simulation of historical TCs when applied over very large scales and above the boundary later (Gutmann et al. 2018). However, given the small drift in the large scales in our CTRL simulation, we chose not to apply spectral nudging. A discussion of the implications of not applying nudging is provided in section 5.
4. Part 2: Effect of future climates on TC rainfall and TCT-S generation
a. Changes in TC rainfall
Not only were the PGW TCs more intense than the CTRL TC, but they also produced more accumulated rainfall (including domain averaged; Table 2) over a larger area than did the CTRL TC (Fig. 8). This result is consistent across all PGW experiments, but possibly for different reasons, as discussed herein. The area of large accumulations (i.e., rainfall exceeding 381 mm, or 15 in.) generated by the GFDL TC exceeded that of the other PGW TCs by more than 300% (Table 2). These results can partially be attributed to the GFDL TC having the slowest forward speed (Table 2), which is found to be correlated to extreme localized rainfall totals in TCs (Konrad et al. 2002). Table 2 shows that the MIROC TC produced the largest area of accumulated rainfall, consistent with this TC having a longer track, and the NCAR TC had the highest rainfall rate. While the maximum PGW rainfall rates increased by roughly 37%–53%, the average rainfall volume (Prein et al. 2017b) over a 500-km radius, increased at a much larger rate of 66%–101% higher than the CTRL TC (Table 2). A similar result for mesoscale convective systems (MCS) is shown in a previous PGW study by Prein et al. (2017a). They found maximum precipitation rates in MCSs increased by 15%–40% while rainfall volume increased by 80%, due to an increase in regions affected by MCSs.
TC track and rainfall metrics for the four simulations.
As in Fig. 4, but for simulated accumulated rainfall (mm) over land points only for (a)–(c) CTRL, (d)–(f) NCAR, (g)–(i) MIROC, and (j)–(l) GFDL for the (top) primary, (middle) PBL1, and (bottom) MP6 experiments.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
An assessment of the distribution of rainfall rates shows an increase in the probability of hourly rain rates greater than 10 mm in the PGW TCs (Fig. 9), with an increase in the extreme hourly rainfall rates (Table 2). The expansion in heavy rainfall has also been shown in other PGW studies of TCs (e.g., Gutmann et al. 2018; Parker et al. 2018, and references therein). This finding is also supported through a landfall-relative time series analysis of maximum (Fig. 10a) and average (Fig. 10b) hourly rainfall rates. Maximum rainfall rates are higher in the PGW experiments postlandfall, whereas the average rainfall rates for NCAR and GFDL simulations are higher than the CTRL until about 42 h after landfall. The MIROC TC maintains the highest average rain rate past this time, likely due to it having a longer track.
Probability of rainfall rates greater than 10 mm h−1, over land points only, for CTRL (green), MIROC (yellow), GFDL (red), and NCAR (blue) TCs.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
Time series of (a) maximum hourly and (b) average hourly rainfall rate over land, relative to time of landfall (t = 0).
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
b. Changes in TCTs
In focusing first on the primary runs, Fig. 11 suggests that the MIROC simulation produces more TCT-Ss than the other two PGW experiments and the CTRL. These observations are confirmed with a quantitative assessment of TCT-S generation (Table 3). The MIROC, GFDL, and NCAR TCs produced 1224, 547, and 441 TCT-Ss, respectively, as compared with the CTRL at 928. Changes in TCT-S generation could be partially attributed to the track length, where a longer track can provide more opportunities to produce TCT-Ss. Consider MIROC for example, which consistently had longer track lengths, and generated the most TCT-Ss (Fig. 11) and was also the most proficient at producing TCT-Ss (0.48 TCTs per kilometer of track; Table 3).
TCT-S (red dots) for (a)–(c) CTRL, (d)–(f) NCAR, (g)–(i) GFDL, and (j)–(l) MIROC for the (top) primary, (middle) PBL1, and (bottom) MP6 experiments. TC tracks (black line) are plotted for the primary experiments only.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
Peak values of updraft helicity UH, vertical velocity w, and vertical vorticity ζ, along with total TCT-S, for the CTRL and PGW experiments.
The TCT-S generation in all simulations varied with the TC evolution: in general, the number of TCT-S decreased with inland movement of the TC (Fig. 12). The time series in Fig. 12 also reveals peak times for TCT-S production for each PGW case based on the time of landfall (time 0 is equal to landfall). For example, the GFDL produced more TCT-S than the CTRL in the hours leading to and just after landfall. This contrasts with the GFDL producing fewer TCT-S overall. MIROC produced more TCT-S than the CTRL 12–24 h after landfall, and NCAR had a brief peak (and its highest) above the CTRL during hours 20–24 h after landfall. Interestingly, 71%, 88%, 51%, and 45% of the TCT-S generated in the CTRL, GFDL, MIROC, and NCAR simulations, respectively, occurred during the landfalling period (from 1800 UTC 15 September to 1800 UTC 16 September).
Time series of TCT-S production between 1800 UTC 15 Sep and 1800 UTC 18 Sep, relative to time of landfall.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
Figure 13 shows an assessment of peak values of UH, vertical velocity w, and vertical vorticity ζ over the 0–3-km layer. In the primary experiments, all three PGW simulations had stronger peak updrafts than did CTRL; however, only NCAR had stronger 0–3-km vertical vorticity. NCAR and GFDL had stronger UH peaks than CTRL, implying stronger rotating updrafts. Despite these stronger rotating updrafts, the GFDL and NCAR simulations produced fewer TCT-S and had shorter tracks than the MIROC simulation.
Fractional change relative to the CTRL for peak UH, w, and ζ values for primary runs.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
Complementing the assessment of the peak values is an analysis of the change in the full distribution of UH, w, and ζ over the full spatial and temporal domain (Table 4). Like the behavior in TCT-S generation, only the MIROC simulation increased the number of grid points exceeding the 83 m2 s−2 UH threshold, specifically by ~23%. All three PGW simulations increased the distribution of w values > 5 m s−1 by ~16%–40% and decreased the distribution of ζ values > 0.005 s−1 by ~8%–19%. These observations support the hypothesis that the number of TCT-S are correlated in part to TC track length.
Percent change of the area under the curve for the storm-relative variables UH, w, and ζ, calculated using the distribution of the variable over the entire spatial and temporal domain.
c. Changes in environment
To understand the environments that supported changes in TC rainfall and TCT-S generation during the landfall period, the convective available potential energy (CAPE), convective inhibition (CIN), 0–3-km bulk wind shear (03BS), and 0–1-km storm-relative environmental helicity (01SRH) were analyzed at 1900 UTC 15 September and then averaged over a small region in southern Alabama and the Florida Panhandle (Fig. 14). This time was chosen to quantify the environmental variables without the influence of TC rainbands and convective storms. Given that the TCT-Ss depend on the location of the parent TC track, and that the PGW TCs all had westward-displaced tracks relative to that of the CTRL TC, the region chosen was the best fit to compare favorable environments for all simulations.
Region where CAPE, CIN, 03BS, and 01SRH are averaged to assess favorable TCT-S environment. Red dots represent TCT-S for the CTRL simulation.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
Results in Table 5 show that the PGW environments had larger values of domain-averaged 01SRH, CAPE, and CIN than did the CTRL. MIROC had the largest domain-averaged 01SRH and 03BS values, which would appear to be consistent with the large number of TCT-Ss in that simulation. GFDL had the largest domain-averaged CAPE, which agrees with the fact that it also produced the most intense rotating updraft during this period and had the highest domain average hourly rain rate. Increased CAPE relates to theoretically stronger updraft speeds, which in turn are positively correlated with rain rates. NCAR had the largest domain-averaged CIN, which likely contributed to its fewer TCT-Ss because of fewer initiated updrafts. Painting a larger picture of the overall environment, Fig. 15 shows the distribution of CAPE, CIN, 01SRH, and 03BS over the full 3-km domain. While the subdomain maximum showed relatively small differences between the future and historical environment, this distribution shows a shift to higher extreme values in all four variables.
Average 01SRH, 03BS, CAPE, and CIN values over the region specified in Fig. 14. Maximum values for each variable are shown with boldface type.
As in Fig. 9, but showing the 3-km-domain-wide distribution of CAPE, CIN, 03BS, and 01SRH at 1900 UTC 15 Sep 2004.
Citation: Journal of Applied Meteorology and Climatology 60, 1; 10.1175/JAMC-D-20-0090.1
The synoptic-scale flow that contributed to these environmental conditions and, perhaps more importantly, to the TC evolution, is shown in Fig. 7. From the 500-hPa geopotential heights at 0000 UTC 19 September 2004 one can infer that the simulated TC in each case was steered west and then northeast after landfall (Fig. 7). An upper-level ridge building west out of the Atlantic appeared to impact both the equatorward extent of a migratory short-wave trough and the poleward extent of the simulated TC tracks (Fig. 7). This ridge was strongest and deepest in the GFDL simulation, and likely had the effect of cutting the TC off from the upper-level trough, and thus limiting its poleward extent to northern Mississippi. As noted earlier in this section, the TCT-S production is partially related to the TC track length.
The interaction of the 500-hPa geopotential heights and the TC has been shown to have some impact on TCT generation itself as the parent TC moves inland (Moore and Dixon 2015). They found that more TCTs were produced if the TC interacted with or was embedded within the westerlies. TCs interacting with the westerlies can result in an environment comprised of veering winds with height, which is shown to be favorable for supercell development (Davies-Jones 1984), the dominant convective mode for TCTs (Edwards et al. 2012). Indeed, all but the GFDL TC is embedded within the 500-hPa trough at the end of the simulation and is noticeably weaker than the other simulated TCs (Fig. 7). The displacement of the 500-hPa trough farther north could be attributed to the higher heights associated with the GFDL simulation, resulting in part from the warmer temperature deltas associated with this GCM. Moore and Dixon (2015) also point out that it is the combination of the 500-hPa trough interaction, in addition to other meso- and storm-scale ingredients (e.g., dry air intrusions, baroclinic boundaries, CAPE, SRH, and BS), that determine TCT potential. This is confirmed in this research, given that NCAR produced the fewest TCT-Ss, despite being embedded within the 500-hPa trough. Relatively larger CIN, especially at landfall, in addition to the NCAR TC weakening after landfall, potentially contributed to the overall lower total number of TCT-Ss produced.
5. Summary and conclusions
This paper documented possible changes in TC rainfall and TCT generation within Hurricane Ivan (2004) simulated in a future climate. A control simulation conducted with the WRF Model was compared with simulations using a “pseudo–global warming” approach. The PGW simulations involved future climate conditions over the late (2080–90) twenty-first century period under RCP8.5, as extracted from three CMIP5 GCMs (NCAR, MIROC, and GFDL). Changes in TC intensity, TC rainfall and TCT production for the PGW-modified Ivan were documented and analyzed.
Relative to the CTRL, all three PGW simulations showed an increase in peak TC intensity and generated substantially more accumulated rainfall over the course of Ivan’s progression over land. With regard to TCT production, MIROC produced more TCT-Ss than the CTRL, while the GFDL and NCAR models produced fewer. Given the spread in the TCT-S response to PGW, our testing of the hypothesis that anthropogenic climate change will intensify landfalling TCs and result in an increase in TCT generation is inconclusive. Existing research shows a general trend of increasing TCTs with TC intensity, but with substantial scatter such that some weak storms produce large numbers of TCTs, and strong storms can produce few TCTs. Results from this study supports that scatter, given that all three PGW storms were more intense, yet the number of TCTs varied. Instead, other drivers of TCT generation have been suggested by this study—contributions by a storm’s track length and the interaction between a TC and environmental winds associated with midlevel troughs. With regard to rainfall, there is more confidence in the resulting relationship in rainfall production, which aligns with results from other studies with increased rain rates (37%–53%), and an even higher increase in rainfall volume (66%–101%).
Increased CAPE, 03BS, and 01SRH values in future environments contributed to more TCT-S generation during the landfall period, particularly for the MIROC and GFDL simulations. The NCAR PGW model had significantly lower 03BS than the other two PGW models and the CTRL, and consequentially less TCT-S production at landfall. In addition, the length of the track also had some correlation to the number of TCT-Ss generated in the PGW experiments compared to the CTRL. The MIROC had the longest track and thus was more productive in TCT and rainfall production, per unit track. This is compared to the GFDL and NCAR PGW TCs, which had a shorter track and fewer TCT-Ss than the CTRL. Analysis of the 500-hPa geopotential height showed the impact of a westward moving upper-level ridge, limiting the equatorward extent of the short-wave trough and the poleward extent of the simulated TC tracks. This ridge was especially pronounced in the GFDL simulations. Implications of these results suggest that TCs like Ivan (2004) in future climates could be potentially more intense, producing heavier rainfall, and could result in more intense rotating storms (such as seen in NCAR and GFDL), or more numerous TCT-Ss (i.e., MIROC).
Future work will explore additional cases of TCs producing a wide range of TCTs, to analyze the robustness of the results shown in this paper. While this study did not conclusively show that TCT generation would increase in future climates, this work does show that TCT generation is sensitive to changes to the large-scale deltas added to the existing environment. Future research should focus on better understanding the relationship between the TC environment and tornado production, including work understanding the relationship between TC intensity and TCT generation. This work should inspire more idealized studies to understand the mechanisms in which TCTs would be generated in future climate environments.
These climate-change results should be interpreted within the context of limitations of the modeling approach. First, the performance of parameterization schemes may change in a changing climate because of possible climate-dependent biases. If model bias changes then a case could be made to modify the parameterization schemes. However, separating the climate-change signal from a possible change in model bias is a vexing problem (Maraun 2016) and was beyond the scope of this study. But our results should be interpreted within this context of a possible convolution of climate change with changing model bias. Second, although we found little evidence for large-scale drift in our control simulation, other model setups and TC cases, including the PGW simulations presented here, may indeed experience drift. This would indicate contamination of the climate-change signal by model internal variability. Further work is needed in this area to carefully isolate the change contribution due to the future climate deltas from variations in model physics performance and the way the PGW simulations evolve through time.
Acknowledgments
The authors thank the National Science Foundation Graduate Research Fellowship dev-00053298 and Blue Waters Super Computing facilities for providing the resources to complete this research. In addition, author Trapp was supported in part by National Science Foundation AGS 1923042. The National Center for Atmospheric Research is sponsored by the National Science Foundation.
REFERENCES
Baker, A. K., M. D. Parker, and M. D. Eastin, 2009: Environmental ingredients for supercells and tornadoes within Hurricane Ivan. Wea. Forecasting, 24, 223–244, https://doi.org/10.1175/2008WAF2222146.1.
Bender, M. A., T. R. Knudson, R. T. 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, https://doi.org/10.1126/science.1180568.
Carroll-Smith, D. L., L. C. Dawson, and R. J. Trapp, 2019: High-resolution real-data WRF modeling and verification of tropical cyclone tornadoes associated with Hurricane Ivan (2004). Electron. J. Severe Storms Meteor., 14 (2), https://www.ejssm.org/ojs/index.php/ejssm/article/viewArticle/168.
Chen, F., and J. Dudhia, 2001: Coupling an advanced land surface-hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129, 569–585, https://doi.org/10.1175/1520-0493(2001)129<0569:CAALSH>2.0.CO;2.
Choi, S. J., and D. K. Lee, 2016: Impact of spectral nudging on the downscaling of tropical cyclones in regional climate simulations. Adv. Atmos. Sci., 33, 730–742, https://doi.org/10.1007/s00376-016-5061-y.
Davies-Jones, R., 1984: Streamwise vorticity: The origin of updraft rotation in supercell storms. J. Atmos. Sci., 41, 2991–3006, https://doi.org/10.1175/1520-0469(1984)041<2991:SVTOOU>2.0.CO;2.
Edwards, R., 2010: Tropical cyclone tornado records for the modernized NWS era. 25th Conf. on Severe Local Storms, Denver, CO, Amer. Meteor. Soc., P3.1, https://ams.confex.com/ams/25SLS/techprogram/paper_175269.htm.
Edwards, R., 2012: Tropical cyclone tornadoes: A review of knowledge in research and prediction. Electron. J. Severe Storms Meteor., 7 (6), https://ejssm.org/ojs/index.php/ejssm/article/viewArticle/97.
Edwards, R., A. R. Dean, R. L. Thompson, and B. T. Smith, 2012: Convective modes for significant severe thunderstorms in the contiguous United States. Part III: Tropical cyclone tornadoes. Wea. Forecasting, 27, 1507–1519, https://doi.org/10.1175/WAF-D-11-00117.1.
Emanuel, K., 2013: Downscaling CMIP5 climate models shows increased tropical cyclone activity over the 21st century. Proc. Natl. Acad. Sci. USA, 110, 12 219–12 224, https://doi.org/10.1073/pnas.1301293110.
Emanuel, K., 2017: Assessing the present and future probability of Hurricane Harvey’s rainfall. Proc. Natl. Acad. Sci. USA, 114, 12 681–12 684, https://doi.org/10.1073/pnas.1716222114.
Emanuel, K., R. Sundararajan, and J. Williams, 2008: Hurricanes and global warming: Results from downscaling IPCC AR4 simulations. Bull. Amer. Meteor. Soc., 89, 347–367, https://doi.org/10.1175/BAMS-89-3-347.
Frei, C., C. Schär, D. Lüthi, and H. C. Davies, 1998: Heavy precipitation processes in a warmer climate. Geophys. Res. Lett., 25, 1431–1434, https://doi.org/10.1029/98GL51099.
Gentry, R. C., 1983: Genesis of tornadoes associated with hurricanes. Mon. Wea. Rev., 111, 1793–1805, https://doi.org/10.1175/1520-0493(1983)111<1793:GOTAWH>2.0.CO;2.
Green, B. W., and F. Zhang, 2013: Impacts of air–sea flux parameterizations on the intensity and structure of tropical cyclones. Mon. Wea. Rev., 141, 2308–2324, https://doi.org/10.1175/MWR-D-12-00274.1.
Gutmann, E. D., R. M. Rasmussen, L. Changhai, and K. Ikeda, 2018: Changes in hurricanes from a 13-yr convection-permitting pseudo–global warming simulation. J. Climate, 31, 3643–3657, https://doi.org/10.1175/JCLI-D-17-0391.1.
Hong, S. Y., and J. O. J. Lim, 2006: The WRF single-moment 6-class microphysics scheme (WSM6). J. Korean Meteor. Soc., 42, 129–151.
Hong, S. Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
Hoogewind, K. A., M. E. Baldwin, and R. J. Trapp, 2017: The impact of climate change on hazardous convective weather in the United States: Insight from high-resolution dynamical downscaling. J. Climate, 30, 10 081–10 100, https://doi.org/10.1175/JCLI-D-16-0885.1.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Janjić, Z. I., 1994: The step-mountain Eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945, https://doi.org/10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.
Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–181, https://doi.org/10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO;2.
Kain, J. S., and Coauthors, 2008: Some practical consideration regarding horizontal resolution in the first generation of operational convection-allowing NWP. Wea. Forecasting, 23, 931–952, https://doi.org/10.1175/WAF2007106.1.
Knutson, T., 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.
Konrad, C. E., M. F. Meaux, and D. A. Meaux, 2002: Relationships between tropical cyclone attributes and precipitation totals: Considerations of scale. Int. J. Climatol., 22, 237–247, https://doi.org/10.1002/joc.721.
Lackmann, G. M., 2015: Hurricane Sandy before 1900 and after 2100. Bull. Amer. Meteor. Soc., 96, 547–560, https://doi.org/10.1175/BAMS-D-14-00123.1.
Lynn, B., R. Healy, and L. Druyan, 2009: Investigation of Hurricane Katrina characteristics for future, warmer climates. Climate Res., 39, 75–86, https://doi.org/10.3354/cr00801.
Maraun, D., 2016: Bias correcting climate change simulations—A critical review. Curr. Climate Change Rep., 2, 211–220, https://doi.org/10.1007/s40641-016-0050-x.
McCaul, E. W., 1991: Buoyancy and shear characteristics of hurricane-tornado environments. Mon. Wea. Rev., 119, 1954–1978, https://doi.org/10.1175/1520-0493(1991)119<1954:bascoh>2.0.co;2.
Mendelsohn, R., K. Emanuel, S. Chonabayashi, and L. Bakkensen, 2012: The impact of climate change on global tropical cyclone damage. Nat. Climate Change, 2, 205–209, https://doi.org/10.1038/nclimate1357.
Molina, M. J., J. T. Allen, and A. F. Prein, 2020: Moisture attribution and sensitivity analysis of a winter tornado outbreak. Wea. Forecasting, 35, 1263–1288, https://doi.org/10.1175/WAF-D-19-0240.1.
Moon, J., D. H. Cha, M. Lee, and J. Kim, 2018: Impact of spectral nudging on real-time tropical cyclone forecast. J. Geophys. Res. Atmos., 123, 12 647–12 660, https://doi.org/10.1029/2018JD028550.
Moore, T. W., and R. W. Dixon, 2011: Climatology of tornadoes associated with Gulf Coast-landfalling hurricanes. Geogr. Rev., 101, 371–395, https://doi.org/10.1111/j.1931-0846.2011.00102.x.
Moore, T. W., and R. W. Dixon, 2015: Patterns in 500 hPa geopotential height associated with temporal clusters of tropical cyclone tornadoes. Meteor. Appl., 22, 314–322, https://doi.org/10.1002/met.1456.
NOAA/NCEP, 2000: NCEP FNL Operational Model Global Tropospheric Analyses, continuing from July 1999 (updated daily). National Center for Atmospheric Research Computational and Information Systems Laboratory Research Data Archive, accessed 6 December 2015, https://doi.org/10.5065/D6M043C6.
Novlan, D., and W. Gray, 1974: Hurricane-spawned tornadoes. Mon. Wea. Rev., 102, 476–488, https://doi.org/10.1175/1520-0493(1974)102<0476:HST>2.0.CO;2.
Otte, T. L., C. G. Nolte, M. J. Otte, and J. H. Bowden, 2012: Does nudging squelch the extremes in regional climate modeling? J. Climate, 25, 7046–7066, https://doi.org/10.1175/JCLI-D-12-00048.1.
Parker, C. L., C. L. Bruyère, P. A. Mooney, and A. H. Lynch, 2018: The response of land-falling tropical cyclone characteristics to projected climate change in northeast Australia. Climate Dyn., 51, 3467–3485, https://doi.org/10.1007/s00382-018-4091-9.
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.
Prein, A. F., and Coauthors, 2015: A review on regional convection-permitting climate modeling: Demonstrations, prospects, and challenges. Rev. Geophys., 53, 323–361, https://doi.org/10.1002/2014RG000475.
Prein, A. F., R. M. Rasmussen, K. Ikeda, C. Liu, M. P. Clark, and G. J. Holland, 2017a: The future intensification of hourly precipitation extremes. Nat. Climate Change, 7, 48–52, https://doi.org/10.1038/nclimate3168.
Prein, A. F., C. Liu, K. Ikeda, S. B. Trier, R. M. Rasmussen, G. J. Holland, and M. P. Clark, 2017b: Increased rainfall volume from future convective storms in the US. Nat. Climate Change, 7, 880–884, https://doi.org/10.1038/s41558-017-0007-7.
Raktham, C., C. Bruyère, J. Kreasuwun, J. Done, C. Thongbai, and W. Promnopas, 2015: Simulation sensitivities of the major weather regimes of the Southeast Asia region. Climate Dyn., 44, 1403–1417, https://doi.org/10.1007/s00382-014-2156-y.
Ramsay, H. A., and A. H. Sobel, 2011: Effects of relative and absolute sea surface temperature on tropical cyclone potential intensity using a single-column model. J. Climate, 24, 183–193, https://doi.org/10.1175/2010JCLI3690.1.
Rappaport, E. N., 2014: Fatalities in the United States from Atlantic tropical cyclones: New data and interpretation. Bull. Amer. Meteor. Soc., 95, 341–346, https://doi.org/10.1175/BAMS-D-12-00074.1.
Rhodes, C. L., and J. C. Senkbeil, 2014: Factors contributing to tornadogenesis in landfalling Gulf of Mexico tropical cyclones. Meteor. Appl., 21, 940–947, https://doi.org/10.1002/met.1437.
Sato, T., F. Kimura, and A. Kitoh, 2007: Projection of global warming onto regional precipitation over Mongolia using a regional climate model. J. Hydrol., 333, 144–154, https://doi.org/10.1016/j.jhydrol.2006.07.023.
Schär, C., C. Frei, D. Lüthi, and H. C. Davies, 1996: Surrogate climate-change scenarios for regional climate models. Geophys. Res. Lett., 23, 669–672, https://doi.org/10.1029/96GL00265.
Scoccimarro, E., S. Gualdi, A. Bellucci, M. Zampieri, and A. Navarra, 2013: Heavy precipitation events in a warmer climate: Results from CMIP5 models. J. Climate, 26, 7902–7911, https://doi.org/10.1175/JCLI-D-12-00850.1.
Sillmann, J., and Coauthors, 2017: Understanding, modeling and predicting weather and climate extremes: Challenges and opportunities. Wea. Climate Extremes, 18, 65–74, https://doi.org/10.1016/j.wace.2017.10.003.
Simpson, R. H., 1974: The hurricane disaster-potential scale. Weatherwise, 27, 169–186, https://doi.org/10.1080/00431672.1974.9931702.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.
Sobel, A. H., S. J. Camargo, T. M. Hall, C.-Y. Lee, M. K. Tippett, and A. A. Wing, 2016: Human influence on tropical cyclone intensity. Science, 353, 242–246, https://doi.org/10.1126/science.aaf6574.
Stewart, S. R., 2011: Tropical cyclone report: Hurricane Ivan (AL092004), 2–24 September 2004. NHC Tech. Rep., 44 pp., https://www.nhc.noaa.gov/data/tcr/AL092004_Ivan.pdf.
Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 5095–5115, https://doi.org/10.1175/2008MWR2387.1.
Tory, K. J., J. L. McBride, H. Ye, and R. A. Dare, 2013: Projected changes in late-twenty-first century tropical cyclone frequency in 13 coupled climate models from the Coupled Model Intercomparison Project Phase 5. J. Climate, 26, 9946–9959, https://doi.org/10.1175/JCLI-D-13-00010.1.
Trapp, R. J., and K. Hoogewind, 2016: The realization of extreme tornadic storm events under future anthropogenic climate change. J. Climate, 29, 5251–5265, https://doi.org/10.1175/JCLI-D-15-0623.1.
Trapp, R. J., E. Robinson, M. Baldwin, N. Diffenbaugh, and B. J. Schwedler, 2011: Regional climate of hazardous convective weather through high-resolution dynamical downscaling. Climate Dyn., 37, 677–688, https://doi.org/10.1007/s00382-010-0826-y.
Verbout, S. M., D. M. Schultz, L. M. Leslie, H. E. Brooks, D. J. Karoly, and K. L. Elmore, 2007: Tornado outbreaks associated with landfalling hurricanes in the North Atlantic basin: 1954–2004. Meteor. Atmos. Phys., 97, 255–271, https://doi.org/10.1007/s00703-006-0256-x.
Villarini, G., and G. A. Vecchi, 2012: Twenty-first-century projections of North Atlantic tropical storms from CMIP5 models. Nat. Climate Change, 2, 604–607, https://doi.org/10.1038/nclimate1530.
Villarini, G., and G. A. Vecchi, 2013: Projected increases in North Atlantic tropical cyclone intensity from CMIP5 models. J. Climate, 26, 3231–3240, https://doi.org/10.1175/JCLI-D-12-00441.1.
Villarini, G., G. A. Vecchi, and J. A. Smith, 2010: Modeling the dependence of tropical storm counts in the North Atlantic basin on climate indices. Mon. Wea. Rev., 138, 2681–2705, https://doi.org/10.1175/2010MWR3315.1.
Walsh, K. J., and Coauthors, 2016: Tropical cyclones and climate change. Wiley Interdiscip. Rev.: Climate Change, 7, 65–89, https://doi.org/10.1002/WCC.371.
Wang, H., Y. Wang, and H. Xu, 2013: Improving simulation of a tropical cyclone using dynamical initialization and large-scale spectral nudging: A case study of Typhoon Megi (2010). Acta Meteor. Sin., 27, 455–475, https://doi.org/10.1007/s13351-013-0418-y.
Weiss, S. J., 1985: On the operational forecasting of tornadoes associated with tropical cyclones. Preprints, 14th Conf. on Severe Local Storms, Indianapolis, IN, Amer. Meteor. Soc., 293–296.
