The Future of Supercells in the United States

The frequency (y axis) of 3.75 km grid cells of hourly maximum 2–5 km AGL UH that meet certain thresholds (x axis) for the three epochs (colored lines) for (a) annual, (b) winter (DJF), (c) spring (MAM), (d) summer (JJA), and (e) fall (SON) for the central CONUS (cf. Fig. 3a dashed rectangle for domain).

Initially, we evaluate the frequency of the full spectrum of hourly maximum 2–5 km AGL UH thresholds for the central and eastern CONUS ( Fig. 2 ). Generally, a separation in total grid cell counts by UH magnitude for the three epochs does not occur until a threshold of ∼150 m 2 s −2 is exceeded, with the overall count ceiling much greater for particularly intense UH (≥200 m 2 s −2 ) grid cells in the projections for the end of the twenty-first century. While there is much separation in intense thresholded grid cell counts between the HIST and FUTR epochs, there is little difference in annual counts between the FUTR epochs. This lack of difference between FUTR epochs is largely due to a seasonal discrepancy in grid cell counts, particularly in the winter, spring, and summer. Intense threshold counts in the FUTR4.5 epoch exceed the FUTR8.5 epoch in the spring, whereas the FUTR8.5 exceeds the FUTR4.5 counts in the winter and summer. Despite seasonal differences, it is projected that particularly intense storm rotation will be more prevalent in a future climate under both intermediate and pessimistic greenhouse gas concentration trajectories. The strong correlation between UH and SCS perils ( Clark et al. 2013 ; Sobash et al. 2016a ; Gagne et al. 2017 ) suggests that this would lead to more hazards in the future, including significant tornado and hail events.

Spatial and temporal climatology.

Spatially, the WRF-BCC does an admirable job of representing the probable climatology of supercells (Fig. 3a; Gensini and Ashley 2011; Smith et al. 2012; Taszarek et al. 2020; Davenport 2021), though an extensive climatology of CONUS supercells based on observed data has not been generated to date. As discussed in the methods, we are likely undercounting supercells that tend to be shallow or transient as inferred by environmental proxies and/or SCS peril climatologies (Sherburn and Parker 2014; Davis and Parker 2014; Childs et al. 2018; Ashley et al. 2019), which may lead to climatological unrepresentativeness—depending on how one defines a supercell—in parts of the Southeast, Ohio valley, and mid-Atlantic.

Fig. 3. Mean annual supercell track counts on an 80 km grid for the three simulation epochs: (a) HIST, (b) FUTR4.5, and (d) FUTR8.5. (c),(e) The mean annual supercell count track differences, or deltas, between FUTR4.5 and HIST and between FUTR8.5 and HIST, respectively. (a) illustrates domains assessed, which include east CONUS (solid outline east of Continental Divide); central CONUS signified (dashed rectangle); and subregions centered in the northern plains, southern plains, Midwest, and mid-South (circles). Stippling denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR8.5. Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide View Full Size Fig. 3. Mean annual supercell track counts on an 80 km grid for the three simulation epochs: (a) HIST, (b) FUTR4.5, and (d) FUTR8.5. (c),(e) The mean annual supercell count track differences, or deltas, between FUTR4.5 and HIST and between FUTR8.5 and HIST, respectively. (a) illustrates domains assessed, which include east CONUS (solid outline east of Continental Divide); central CONUS signified (dashed rectangle); and subregions centered in the northern plains, southern plains, Midwest, and mid-South (circles). Stippling denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR8.5. Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide Fig. 3. Mean annual supercell track counts on an 80 km grid for the three simulation epochs: (a) HIST, (b) FUTR4.5, and (d) FUTR8.5. (c),(e) The mean annual supercell count track differences, or deltas, between FUTR4.5 and HIST and between FUTR8.5 and HIST, respectively. (a) illustrates domains assessed, which include east CONUS (solid outline east of Continental Divide); central CONUS signified (dashed rectangle); and subregions centered in the northern plains, southern plains, Midwest, and mid-South (circles). Stippling denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR8.5. Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide

Deltas, or changes, between the HIST and FUTR epochs follow a similar pattern with increased supercell tracks found east of 95°W and, generally, decreases in the Great Plains (Rossum and Lavin 2000), or west of Interstate 35 (Figs. 3b–e). The greatest positive track change is found between HIST and FUTR8.5, with increased events maximizing in north Texas and the Ark-La-Tex and Ozark Plateau regions with broad increases found elsewhere in the mid-South and western Great Lakes. The net positive pattern is similar in the HIST and FUTR4.5 delta but is more muted in the eastern Arkansas River valley, with delta maxima displaced toward the mid-South and central Gulf Coast. The FUTR increases in the eastern Arkansas and lower Mississippi River valleys are driven by rising seasonal numbers, particularly in the spring and, to a lesser extent, the winter (Fig. ES1 in the online supplemental material; https://doi.org/10.1175/BAMS-D-22-0027.2). Future decreases in both delta assessments are concentrated in portions of the Great Plains, from south Texas to South Dakota, with a notable reduction found from the High Plains of Colorado through the middle Missouri valley (Fig. 3). Contrary to the South, these changes are caused by reduced supercell counts across wide expanses of the central CONUS during the summer (Fig. ES1).

Like UH grid cell counts, a dichotomy is found in the HIST and FUTR deltas, with robust and, in some cases, significant, supercell count and areal footprint increases found in the late-twenty-first-century epochs for February, March, and April and declines in June, July, and September (Fig. 4; Table 1; Table ES1 and Figs. ES2–ES5). Thus, supercells are projected to be more prevalent in the early severe weather season, while declining in frequency in the late season. Considering the seasonality of SCSs and their environments, this suggests that the early-season supercells will affect regions of the southern tier of the CONUS, which is an area that is particularly vulnerable to supercell perils (Ashley 2007; Ashley et al. 2008; Ashley and Strader 2016; Strader et al. 2017a,b, 2021). Meanwhile, late season events, which typify the High Plains and northern plains, will be less frequent, likely a consequence of the increasing CIN that is expected across that geography (Hoogewind et al. 2017; Rasmussen et al. 2017; cf. “Environmental ingredients” section).

Fig. 4. (a) Annual cumulative frequency of supercell counts for the three epochs for the eastern CONUS (see Fig. 3a for domain), as well as (b) seasonal supercell counts and (c) monthly supercell counts for the three epochs illustrated using box-and-whisker diagrams. (d)–(f) As in (a)–(c), but for cumulative areal footprint of supercells. In (a) and (d), means are denoted by thicker lines with the 25th- and 75th-percentile bounds provided in epoch-respective color shading. For (b) and (c) and (e) and (f), means are denoted by black dots, medians by the black lines, the boxes represent the interquartile range, the whiskers illustrate the 5th and 95th percentiles, and the clear circles denote outliers. A triangle (square) denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR4.5 (FUTR8.5). Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide View Full Size Fig. 4. (a) Annual cumulative frequency of supercell counts for the three epochs for the eastern CONUS (see Fig. 3a for domain), as well as (b) seasonal supercell counts and (c) monthly supercell counts for the three epochs illustrated using box-and-whisker diagrams. (d)–(f) As in (a)–(c), but for cumulative areal footprint of supercells. In (a) and (d), means are denoted by thicker lines with the 25th- and 75th-percentile bounds provided in epoch-respective color shading. For (b) and (c) and (e) and (f), means are denoted by black dots, medians by the black lines, the boxes represent the interquartile range, the whiskers illustrate the 5th and 95th percentiles, and the clear circles denote outliers. A triangle (square) denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR4.5 (FUTR8.5). Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide Fig. 4. (a) Annual cumulative frequency of supercell counts for the three epochs for the eastern CONUS (see Fig. 3a for domain), as well as (b) seasonal supercell counts and (c) monthly supercell counts for the three epochs illustrated using box-and-whisker diagrams. (d)–(f) As in (a)–(c), but for cumulative areal footprint of supercells. In (a) and (d), means are denoted by thicker lines with the 25th- and 75th-percentile bounds provided in epoch-respective color shading. For (b) and (c) and (e) and (f), means are denoted by black dots, medians by the black lines, the boxes represent the interquartile range, the whiskers illustrate the 5th and 95th percentiles, and the clear circles denote outliers. A triangle (square) denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR4.5 (FUTR8.5). Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide

Table 1. Measures of central tendency for eastern CONUS annual supercell metrics illustrated in Fig. 5. The two right columns include positive percentage changes in central tendencies for FUTR vs HIST epochs. The only significant (p < 0.05; Mann–Whitney U test) difference is in the 200 m2 s−2 threshold at RCP8.5. View Table

Additional annual metrics—supercell hours, total accumulation of UH values for all supercells, and cumulative spatial extent of UH within supercells—follow similar trends as those found in annual supercell track counts (Fig. 5; Table 1; Table ES1 and Figs. ES2–ES5), with overlap in the interquartile range, but with general increases found for future epochs, though with mixed results when testing for statistical significance. The annual UH track accumulation, which is effectively the sum of all UH pixel values in all qualifying supercells swaths, has less overlap in the interquartile range compared to track counts and hours, with a notable increase—though, not statistically significant—in central tendencies. This jump in future total track accumulation suggests the potential for not only more supercells, but longer-lived, or more intense ones, when they do occur. The most striking differences between HIST and FUTR epochs were found in the annual UH cumulative areal extents for strong (≥100 m2 s−2) and intense (≥200 m2 s−2) thresholds. Track accumulation extents at the strong threshold have a mean (median) increase of 9.3% (28.7%) from HIST to FUTR4.5 and 19.2% (26.8%) from HIST to FUTR8.5, whereas intense threshold accumulation extents have a mean (median) increase of 25.8% (11.9%) from HIST to FUTR4.5 and 60.2% (92.1%) from HIST to FUTR8.5. These projected FUTR threshold areal extent increases are concerning as these extreme mesocyclones are more likely to produce perils that have societal impacts.

Fig. 5. Box-and-whisker plots illustrating annual eastern CONUS (a) supercell counts, (b) sum of supercell track hours, or count of hourly supercell slices, (c) UH accumulation, or the sum of all UH values in all qualifying supercell slices, and (d)–(f) areal track accumulation extents for 75 m2 s−2, 100 m2 s−2 (intense), and 200 m2 s−2 (extreme) thresholds, respectively. Box-and-whisker distributions and central tendencies as in Fig. 4. A triangle (square) denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR4.5 (FUTR8.5). Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide View Full Size Fig. 5. Box-and-whisker plots illustrating annual eastern CONUS (a) supercell counts, (b) sum of supercell track hours, or count of hourly supercell slices, (c) UH accumulation, or the sum of all UH values in all qualifying supercell slices, and (d)–(f) areal track accumulation extents for 75 m2 s−2, 100 m2 s−2 (intense), and 200 m2 s−2 (extreme) thresholds, respectively. Box-and-whisker distributions and central tendencies as in Fig. 4. A triangle (square) denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR4.5 (FUTR8.5). Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide Fig. 5. Box-and-whisker plots illustrating annual eastern CONUS (a) supercell counts, (b) sum of supercell track hours, or count of hourly supercell slices, (c) UH accumulation, or the sum of all UH values in all qualifying supercell slices, and (d)–(f) areal track accumulation extents for 75 m2 s−2, 100 m2 s−2 (intense), and 200 m2 s−2 (extreme) thresholds, respectively. Box-and-whisker distributions and central tendencies as in Fig. 4. A triangle (square) denotes a significant (p < 0.05; Mann–Whitney U test) difference between HIST and FUTR4.5 (FUTR8.5). Citation: Bulletin of the American Meteorological Society 104, 1; 10.1175/BAMS-D-22-0027.1 Download Figure

Download figure as PowerPoint slide

Despite net positive percent changes in annual mean and median values for all supercell metrics from HIST to FUTR epochs (Table 1), seasonal trends during the typical peak of SCS season are varied (Table ES1). Much like supercell counts, supercell hours, spatial extents, and cumulative UH metrics all have dramatic increases in the winter and early spring for both FUTR epochs, while stable to decreasing values for summer. This, again, suggests a more robust earlier SCS season in the future due to increased supercell populations and/or magnitudes, with reduced presence of supercells and their perils during June–September. The most dramatic percent increases in supercells metrics from HIST to FUTR are found during the winter and early spring, with, in some cases, a near doubling of monthly metrics, particularly for the FUTR8.5 delta. Conversely, many of the June–September supercell metrics decline, on average, from 10% to 30% in the FUTR. Despite the notable increases found in the cool and spring transition seasons, we acknowledge that our data and methods do not fully account for the potential changes that may occur since events during this period have a higher likelihood to be supercells, sometimes embedded in QLCS structures, that tend to be smaller, shallower, transient, and/or weaker due to a weakness in environmental ingredients such as moisture and instability (Smith et al. 2012; Sherburn and Parker 2014; Ashley et al. 2019). Thus, the net increase in the number of supercells, their magnitudes, and their perils may be even more substantial during future early SCS seasons. May, which is typically the CONUS peak climatology for supercell-related perils of hail and tornadoes, is the only month during the climatologically active part of the season without substantial changes in supercell metrics. This may be due to the environmental “goldilocks” character of this month, when ingredients of SCSs have, and will likely continue to, comingle frequently and at just the proper amounts, while months on either side of May deal with dichotomous changes in their respective overlap of important ingredients (Hua and Anderson-Frey 2022).

A regional perspective on the differences in supercell counts between epochs reinforces the narrative of “more early, less later,” with the eastern CONUS, northern and southern plains, and Midwest all having a notable increase in counts for their respective early severe weather seasons, with decreases in the latter half of the season (Fig. 6). The mid-South domain has notable FUTR positive increases restricted to February, March, and April, with negligible change in all other months.