AUGUST 2016 A G E E E T A L . 1681

Spatial Redistribution of U.S. Activity between 1954 and 2013

ERNEST AGEE AND JENNIFER LARSON Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette,

SAMUEL CHILDS Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

ALEXANDRA MARMO Department of Earth, Atmospheric, and Planetary Sciences, Purdue University, West Lafayette, Indiana

(Manuscript received 25 November 2015, in final form 18 May 2016)

ABSTRACT

Climate change over the past several decades prompted this preliminary investigation into the possible effects of global warming on the climatological behavior of U.S. tornadoes for the domain bounded by 308– 508Nand808–1058W. On the basis of a warming trend over the past 30 years, the modern tornado record can be divided into a cold ‘‘Period I’’ from 1954 to 1983 and a subsequent 30-year warm ‘‘Period II’’ from 1984 to 2013. Tornado counts and days for (E)F1–(E)F5, significant, and the most violent tornadoes across a 2.5832.58 gridded domain indicate a general decrease in tornado activity from Period I to Period II concentrated in Texas/Oklahoma and increases concentrated in Tennessee/Alabama. These changes show a new geographical distributionoftornadoactivityforPeriodIIwhencomparedwithPeriodI. Statistical analysis that is based on field significance testing and the bootstrapping method provides proof for the observed decrease in annual tornado activity in the traditional ‘‘Tornado Alley’’ and the emergence of a new maximum center of tornado activity. Seasonal analyses of both counts and days for tornadoes and significant tornadoes show similar results in the spring, summer, and winter seasons, with a substantial decrease in the central plains during summer. The autumn season displays substantial increases in both tornado counts and significant-tornado counts in the region stretching from Mississippi into Indiana. Similar results are found from the seasonal analysis of both tornado days and significant-tornado days. This temporal change of spatial patterns in tornado activity for successive cold and warm periods may be suggestive of climate change effects yet warrants the climatological study of meteorological parameters responsible for tornado formation.

1. Introduction Strader (2016). Brooks et al. (2014a),aswellasAgee and Childs (2014), have noted the various un- The temporal change of spatial patterns in the U.S. certainties in the tornado record that can potentially ‘‘’’ is an increasingly important impede a determination of any climate change effects research area because of the potential effects of global on tornado occurrences. Climate-model simulations warming on key meteorological fields of information (e.g., Trapp et al. 2007; Diffenbaugh et al. 2013)point that affect severe-thunderstorm and tornado devel- to the possible effects due to increasing CAPE, yet opment. Additional value in having such updated in- these model predictions also show decreasing shear in formation is also evident in the study by Ashley and the lower levels of the troposphere. These conflicting meteorological predictions in a warming climate in- Denotes Open Access content. troduce further uncertainty in detecting changes in severe-thunderstorm behavior that exceed the natural internal variability. Corresponding author address: Ernest Agee, Dept. of Earth, Atmospheric, and Planetary Sciences, Purdue University, 550 Dynamic downscaling of reanalysis data, as well as Stadium Mall Drive, West Lafayette, IN 47907. climate-model simulation, offers the opportunity to E-mail: [email protected] examine regional patterns of meteorological change

DOI: 10.1175/JAMC-D-15-0342.1

Ó 2016 American Meteorological Society Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1682 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55 that affect hazardous convective weather (HCW). Trapp et al. (2011) with downscaling of reanalysis data and Gensini and Mote (2015) through high-resolution dynamic downscaling of the CCSM3 (acronym defi- nitions can be found at http://www.ametsoc.org/ PubsAcronymList) show regions favored for in- creased HCW. Downscaling performed by Gensini and Mote (2015) to a 4-km grid spacing, using the Weather Research and Forecasting (WRF) Model, makes a comparison of severe-weather events east of the Continental Divide for the decade 1980–90 versus 2080–90. Their results show that the greatest increase of HCW is for the middle and lower Mississippi val- ley, Tennessee valley, and lower Ohio valley regions with decreased events to the north and the west of FIG. 1. Plot of CONUS annual mean surface air temperature from these areas. Results presented later support such 1954 to 2013. The U.S. cold period is defined as 1954–83. The warm findings. period is defined as 1984–2013. Least squares–fit linear trend lines are From an observation perspective, there is growing shown for each period; the trend lines show that temperature de- interest in finding evidence of changes in tornado cli- creases by 0.358Fforthecoldperiodandincreases by 1.198Fforthe matology associated with the possible effects of global warm period. [The data are courtesy of the National Climatic Data Center (now the National Centers for Environmental Information); warming in the U.S. tornado region for the past several http://www.ncdc.noaa.gov/cag.] decades. Elsner et al. (2015) present empirical evidence of changes in tornado climatology that are possibly related to climate change. Dixon et al. (2011) help to et al. 2006), and these events are not included in this identify the emerging evidence of a ‘‘,’’ study. Note also that shifts in rural population into which represents an eastward extension of the tradi- expanding suburban areas may affect annual counts, tional ‘‘Tornado Alley’’ in the central plains. These especially for weaker tornadoes (but likely not for efforts point to the need and opportunity to examine counts of strong and violent tornadoes). statistically the possible temporal and spatial changes Next, Student’s t tests on 2.5832.58 grid boxes in the in the tornado climatology (particularly since 1954, aforementioned domain (comparable to the resolution in which is the accepted starting point of the modern the NCEP–NCAR reanalysis data) showed significant tornado record). differences in the annual mean tornado counts for the A preliminary investigation of the modern tornado four grid boxes with the most extreme change, which record encouraged this study by finding substantial encompassed Oklahoma (maximum decrease) and Ten- differences in annual tornado counts for key tornado nessee (maximum increase). Although these preliminary states such as Oklahoma (in the traditional Tornado results were suggestive of two distinct populations, the t Alley) and Tennessee (in Dixie Alley). This pre- test is not the most effective statistical test to establish this liminary effort defined two successive 30-year spatial shift. Therefore, more rigorous statistical analysis, periods—a cold ‘‘Period I’’ (1954–83) and a warm such as 1) field significance testing and 2) the comparison ‘‘Period II’’ (1984–2013)—on the basis of surface air of individual grid boxes using the bootstrapping method temperature for the region 808–1058Wand308–508N. of resampling, is required to establish acceptable spatial State counts were compiled for each period (but not and temporal shifts in tornado activity. In essence, this is presented here). For example, in Oklahoma the tor- the nature of the analysis and the results presented below nado counts for tornadoes that are rated (E)F1–(E)F5 in this study. on the (enhanced) have decreased from A fundamental premise underlying this study has 1096 in Period I to 713 in Period II for a loss of 383 been to analyze two equal-length tornado records for (or a 35% decrease). Tennessee, however, increased the NCEP–NCAR gridded domain (808–1058W, 308– from 275 in Period I to 457 in Period II for a gain of 182 508N) corresponding to two successive 30-year pe- (or a 66% increase). riods, 1954–83 and 1984–2013. It was noted a priori Increasing population has historically accounted that changes in surface air temperature for the two for a steady increase in annual tornado counts until periods selected also represented two successive recent years. Much of this increase is attributed to multidecadal periods characterized by cold surface air large increases in (E)F0 tornado counts (Verbout temperature followed by a warmer period for the

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1683

FIG. 2. Tornado counts (left) (E)F1–(E)F5, (center) (E)F2–(E)F5, and (right) (E)F3–(E)F5 for the NCEP–NCAR gridded domain for (a)–(c) Period I (1954–83), (d)–(f) Period II (1984–2013), and (g)–(i) Period II minus Period I. continental (CONUS). This trend in Period I (1954–83) and warm Period II (1984–2013). temperature may have climate change implications Figure 1 shows the two trend lines of the average for tornado behavior, but this possibility has not been annual surface air temperature for CONUS that investigated in this particular study. correspond to the cold [0.358F (0.198C) decrease] and warm [1.198F (0.668C) increase] periods. The trend lines of the average annual surface air temperature 2. Selection of data, time periods, and analysis for the domain of this study are comparable to the To identify temporal changes in spatial patterns of trend lines for CONUS (in Fig. 1), but CONUS tornado activity it is appropriate to define a domain provides a slightly larger region over which surface that encompasses the U.S. Tornado Alley, as well as air temperature trends can be examined because it appropriate time periods with homogenous data re- encompasses the domain of this study as well as the cords. The domain selected (308–508N, 808–1058W) surrounding area in which storm systems that affect covers the principal region of U.S. tornado activity the chosen domain may form. and is divided into a 2.5832.58 gridbox array, cor- a. Tornado counts and tornado days: Period I versus responding to the NCEP–NCAR reanalysis data re- Period II cord. The modern tornado record for climatological studies begins with 1954, and in this study the period As can be noted in the previous studies that were chosen is from 1954 to 2013. The objective was to use referenced earlier, the recognized tornado record the longest possible data record for each period and for climatological studies consists of the (E)F1–(E)F5 that these two periods have equal length, which tornado events. Tornado counts for each grid box happened to correspond to successive cold and warm are determined on the basis of the following: 1) tor- periods. Surface air temperature for the tornado data nado starts in the grid box and ends elsewhere, 2) tornado record helps to define the above-mentioned cold starts elsewhere and ends in the grid box, 3) tornado starts

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1684 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

FIG.3.AsinFig. 2, but for tornado days. and ends in the grid box, or 4) tornado starts elsewhere Figs. 2b, 2e, and 2h). It is proposed that the new ‘‘heart of and ends elsewhere but the straight-line path crosses Tornado Alley’’ as based on annual totals (and not on any though the grid box. It is noted at this point that the particular season) is now located in central Tennessee/ gridbox numbers for all figures that follow are not pre- northern Alabama and not in eastern Oklahoma. The sented but rather are simply referenced and represented findings in Figs. 2a, 2d, and 2g (and the additional results by the contour plots. Figures 2a–i present the respective to follow) are also very supportive of the shift in the (E)F1–(E)F5, (E)F2–(E)F5, and (E)F3–(E)F5 tornado traditional Tornado Alley, as well as the targeted region counts for Period I (Figs. 2a–c), Period II (Figs. 2d–f), for the Verification of the Origins of Rotation in Torna- and Period II minus Period I (Figs. 2g–i). Figures 2a, 2d, does Experiment-Southeast (VORTEX-SE) field pro- and 2g show the (E)F1–(E)F5 tornado counts across gram scheduled for 2016. Next, as shown in Figs. 2b, 2e, the domain. The grid box (2.5832.58)withthemaximum and 2h, for the significant tornadoes (E)F2–(E)F5 similar (E)F1–(E)F5 count in Period I, as shown in Fig. 2a,wasin results are found, and the new maximum number (185) is southeastern Oklahoma and northeastern Texas (with located in northern Alabama while the greatest de- 477 events). For Period II the (E)F1–(E)F5 count for this crease (2159) is located in southeastern Oklahoma/ grid box decreased to 260 events, as shown in Fig. 2d, for a northeastern Texas. The maximum significant-tornado reduction of 217 (or a decrease of 45%). For Period II, the grid box for Period I was in eastern Oklahoma (271 grid box with the maximum tornado count is now located events), which decreased to 123 events in Period II in northern Alabama (also 477 events). For Period I the (for a reduction of 55%). Although the maximum count for this grid box was 323 events, which represents gridbox count for Period II was located in northern an increase of 48% from Period I to Period II. These Alabama and was comparable to Period I, the adjacent contour plots that are based on data for each grid box northern grid box in central Tennessee had a maximum show strong evidence of a possible major shift in the increase in counts of nearly 100% (from 84 to 166). geographical location of the most tornado activity (as Also, it is evident in Fig. 2h that the significant torna- well as for the most significant tornadoes, shown in does in northeastern Texas and eastern Oklahoma

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1685

FIG. 4. Winter (DJF) tornado counts for (left) (E)F1–(E)F5 and (right) (E)F2–(E)F5 for (a),(b) Period I (1954–83) (c),(d) Period II (1984– 2013), and (e),(f) Period II minus Period I. decreased by counts of 159 and 148, respectively heart of Tornado Alley from 1954 to 1983, which has (equivalent to reductions of 62% and 55%). Further- also experienced a substantial decline in tornado ac- more, these results are similar to the collective count of tivity. Statistical support for this statement is the strong-to-violent (E)F3–(E)F5 tornadoes, as shown presented later. in Figs. 2c, 2f, and 2i. On the basis of the results shown Although these figures and percentages of change in Figs. 2a–i, from central Tennessee to northern Ala- are noteworthy, a new paradigm for U.S. tornado ac- bama is presented as the modern-day center for annual tivity can be further defended by additional analysis. tornado activity, replacing Oklahoma, the previous The question can be raised, for example, as to whether

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1686 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

FIG.5.AsinFig. 4, but for spring (MAM). major tornado outbreaks affecting the ‘‘Dixie’’ states, same format of data presentation as Figs. 2a–i.These such as 3–4 April 1974 and 27–28 April 2011, can bias results, especially for the significant-tornado days, the results. By examining tornado days [defined as a support the same general conclusion as deduced from day with at least one (E)F11 tornado], major out- tornado counts; that is, central Tennessee/northern breaks are counted as one or two days rather than a Alabama is the candidate new heart of Tornado Al- large quantity of tornadoes, thus eliminating the bias. ley. Figure 3a shows the most tornado days (246) from The results of this tornado-day analysis further sup- south-central Oklahoma to north-central Texas in port the tentative conclusion presented in this study. Period I, but the new maximum in Period II (shown in Figures 3a–i are presented for tornado days with the Fig. 3d) is 208 in southern Mississippi. These regions

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1687

FIG.6.AsinFig. 4, but for summer (JJA).

experienced decreases of 48% and 5%, respectively, almost the entire domain, with the maximum decrease while central Tennessee increased by 29 tornado occurring in the traditional Tornado Alley. Figures 3c, days. Figures 3b, 3e, and 3h for the significant-tornado 3f, and 3i show the counts of tornado days for (E)F3– days show a maximum gridbox count of 151 in Period (E)F5, with two pronounced maxima: one in the tra- I, which decreases to 56 in Period II, with the greatest ditional Tornado Alley and the second one in the increase seen in central Tennessee of 25 significant- Dixie Alley. It is also noted that Period II shows an tornado days. Also, from Period I to Period II there overall weaker tornado-day signal than does Period I, is a general decline in significant-tornado days for but the new maximum is now located in northern

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1688 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

FIG.7.AsinFig. 4, but for autumn (SON).

Alabama. Central Tennessee shows the greatest in- for Period I, Period II, and their difference for both the crease of 100% (from 14 to 28 days) for the most- (E)F1–(E)F5 tornadoes, as shown in Figs. 4a, 4c, and violent tornadoes. 4e, and the significant tornadoes, as shown in Figs. 4b, 4d, and 4f. Period II has seen a substantial increase in b. Seasonal changes (counts and days): Period I (E)F1–(E)F5 tornado counts from Tennessee to the versus Period II lower Mississippi valley. Similar results are shown in The changes noted above can be further examined Figs. 4b, 4d, and 4f for the significant tornadoes; there for seasonality, beginning with the winter season (DJF) are decreases near the Gulf Coast for Period II but an

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1689

FIG. 8. Winter (DJF) tornado days for (left) (E)F1–(E)F5 and (right) (E)F2–(E)F5 for (a),(b) Period I (1954–83), (c),(d) Period II (1984– 2013), and (e),(f) Period II minus Period I. increase in central and western Tennessee from Period southwestern Oklahoma had a maximum decrease of I to Period II. The spring season (MAM), shown in 172. Figures 5b, 5d, and 5f show similar results for the Figs. 5a and 5b, identifies the traditional center of ex- significant tornadoes, and values have continued to pected springtime tornado activity in Oklahoma for decline from Period I to Period II in the traditional Period I. For Period II, however, two distinct maxima Tornado Alley (with a maximum gridbox decrease of are apparent, as shown in Figs. 5c and 5d:oneincentral 105). Central Tennessee shows an increase in signifi- Oklahoma and one in northern Alabama. Central cant tornadoes of 55% (going from 65 in Period I to 101 Tennessee had a maximum increase of 105 from Pe- in Period II), however. Figures 6a–d show the expected riod I to Period II, while north-central Texas and northward movement of tornado activity for the

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1690 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

FIG.9.AsinFig. 8, but for spring (MAM). summer (JJA) season, but Fig. 6e shows a substantial but for the significant tornadoes. There are substantial decrease in EF1–EF5 tornadoes of 80% (from 101 decreases over much of the domain, as seen in Fig. 6f, down to 20 tornadoes) in central Oklahoma for Period except for the increase in southern Minnesota. The II minus Period I; this summertime decrease was also maximum gridbox decrease in east-central Oklahoma noted by Brooks et al. (2014b).WesternMinnesota is 44 counts (Fig. 6f). Figures 7a–d show increases for shows gridbox increases of 37 and 40, and eastern counts and significant tornadoes from Period I to Pe- Colorado has an increase of 48. Figures 6b, 6d, and 6f riod II. In particular, Fig. 7e shows a 68% increase in present results similar to those for Figs. 6a, 6c, and 6e, tornado counts (from 74 to 124) in southern Mississippi

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1691

FIG. 10. As in Fig. 8, but for summer (JJA). for SON, with the maximum increase in western Ten- supported in part by the downscaling results shown in nessee of 350% (from 18 to 81). Figures 7b, 7d, and 7f Gensini and Mote (2015,theirFig.4). show the counts and differences for the significant Winter-season counts for (E)F1–(E)F5 tornado days tornadoes for the autumn season with noted increases and significant-tornado days are presented in Figs. 8a–f from Georgia up through the Tennessee and Ohio for Period I, Period II, and Period II minus Period I. For valleys into northern Indiana and notable decreases DJF the counts are largely confined to the Dixie Alley west of the Mississippi River. These results are states for both periods, with a notable decrease along the

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1692 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

FIG. 11. As in Fig. 8, but for autumn (SON).

Gulf Coast. Figures 9a–f show a general spring-season for the southern tier of states as well as in portions of the decrease in tornado days and significant-tornado days Midwest and the Ohio valley, with continued evidence for Period II minus Period I, with the largest MAM of decreases west of the Mississippi River (for both decreases in Oklahoma and north-central Texas. tornado days and significant-tornado days). In general, it Figures 10a–f for JJA show the continued decline of is noted that the autumn season makes the greatest both tornado days and significant-tornado days. contribution to the annual increase in Tennessee/Alabama Figures 11a–f for SON show an increase in tornado days and that the summer contributes the greatest decrease

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1693

FIG. 12. The bootstrapping method of resampling is performed by sampling with replacement 30 times from the annual counts for a selected eastern region and western region, calculating eastern region minus western region for each pair of annual counts sampled, and finding the mean of the 30 difference values. This process is repeated 10 000 times to create a PDF. across Oklahoma and north-central Texas, which has The field significance test performed in this study implications for seasonal prediction and climate change follows the method proposed by Elmore et al. (2006). projections. These results also show agreement with First, a block bootstrap that resamples 10 000 times Gensini and Mote (2015). with a block size of five values is used to test the sig- nificance of Period II minus Period I for all grid boxes in the domain at a 5 0.05. The proportion of statisti- 3. Field significance and bootstrapping (annual cally significant grid boxes N is recorded and stored counts): Period I versus Period II for later use. Next, a Monte Carlo process with 10 000 As large as the spatial changes seem to be from Pe- trials calculates the correlation coefficient between riod I to Period II, suggesting a new center of maximum the annual means of Period II minus the annual means annual tornado counts (as well as significant torna- of Period I and a series of values randomly selected does), additional analysis is required. The approach from a standard normal distribution, and then it de- used now to further solidify these results is to apply a termines the proportion of correlation coefficients field significance test, which addresses any high spatial that are statistically significant at a 5 0.05. If the correlation present in the data, as well as the classical proportion of statistically significant grid boxes is bootstrapping method through resampling (10 000 greater than the proportion of statistically significant times) to examine the corresponding probability den- correlation coefficients, then the domain is ‘‘field sity functions (PDF) and confidence intervals that can significant.’’ For the (E)F1–(E)F5 tornado counts, N is be determined (Diciccio and Romano 1988). calculated to be 27.5% and the threshold is 8.75%.

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1694 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55

and east grid boxes with the greatest change are iden- tified as GW (decrease) and GE (increase), re- spectively. Figure 14a shows the resampling results for the annual tornado counts over box b minus box a. Inspection of the accompanying table shows that the two regions are mutually exclusive at the 99% confi- dence level. Figure 14b is similar to Fig. 14a but is for grid box GE minus grid box GW. The comparison of these PDFs from resampling also shows 99% confi- dence level for the differences. Results for the signifi- cant tornadoes for grid box GE minus grid box GW are presented in Fig. 14c, which also supports, at the 99% confidence level, the different PDFs for the observa- tional data versus the resampled data. On the basis of all of the results presented in Fig. 14, FIG. 13. Locations of box a and box b as well as the grid boxes there is statistical support for major temporal changes in with the greatest increase (grid box GE) and greatest decrease the spatial climatology of U.S. tornadoes in the annual (grid box GW) in tornado counts between Period I and Period II. counts both for (E)F1–(E)F5 tornadoes and for significant tornadoes (E)F2–(E)F5. Furthermore, these shifts show a Because N exceeds the threshold, the (E)F1–(E)F5 new region of maximum annual tornado (and significant tornado counts are field significant at the 95% confi- tornado) occurrence for 1984–2013 identified by box dence level. In a similar way, the (E)F2–(E)F5 tor- b (located in the Dixie Alley) and not by box a (located in nado counts are found to be field significant at the 95% the Period-I traditional Tornado Alley region). confidence level with an N of 37.5% and a threshold of 8.75%. 4. Summary and conclusions Next, the classical bootstrapping method of resampling is performed to show the statistical significance of the A statistical assessment of changes in the U.S. tornado difference between the two regions of most extreme climatology for two consecutive 30-year time periods change for each time period. Figure 12 shows the method over the domain bounded by 308–508N, and 808–1058 has for bootstrapping used to calculate the PDF for the dif- been completed. These two time periods of equal length ference between two grid boxes in a single period. First, were characterized by changes in the mean surface air two values are randomly selected with replacement, one temperature from cold in Period I (1954–83) to warm in from the set of annual mean counts for the eastern region Period II (1984–2013). The years 1950–53 are not con- and one from the set of annual mean counts for the sidered to be a homogeneous part of the modern-day western region. After the difference (eastern region mi- tornado record. Further, 2014 is not included because nus western region) of the values is calculated, two more doing so would have resulted in time periods of unequal values are randomly selected in the same manner; this length. The chosen domain was divided into grid boxes process is repeated until a set of 30 difference values is that were 2.5832.58, which corresponds to the resolution obtained. Then the mean of the 30 difference values is of the NCEP–NCAR reanalysis data. Tornado counts for calculated. This entire process of sampling and calculat- each grid box were made for all (E)F1–(E)F5 tornadoes, ing the mean is repeated 10 000 times to obtain a set of including various subsets of these data for significant, 10 000 mean difference values. From this set of mean strong, and violent tornadoes. The gridbox counts were difference values, a PDF is created that can then be used made for each event according to the following criteria: to test the significance of the counts at the 99% confi- 1) tornado starts in the grid box and ends elsewhere, dence level for a given period. 2) tornado starts elsewhere and ends in the grid box, The bootstrapping method described above is done 3) tornado starts and ends in the grid box, or 4) tornado for the annual tornado counts and the annual starts elsewhere and ends elsewhere but the straight-line significant-tornado counts with a focus on the regions path crosses though the grid box. Similar considerations of extreme change. Figure 13 highlights two of the most were given to tornado days, as well as seasonal partitions extreme regions of change, consisting of four grid for all data. Tornado days are defined as a day with at boxes each and labeled as box a (decrease) and box least one (E)F11 tornado. Statistical field significance b (increase), that are prime targets for bootstrap testing, along with classical bootstrap resampling of analysis. Within these two regions, the individual west selected datasets, has been introduced to support

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1695

FIG. 14. The difference in (a) (E)F1–(E)F5 tornado counts for box b minus box a and the difference in (b) (E)F1–(E)F5 and (c) (E)F2– (E)F5 tornado counts for grid box GE minus grid box GW. All PDFs are derived from 10 000 times of resampling, and all of the PDFs are mutually exclusive at the 99% confidence level.

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC 1696 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 55 conclusions regarding spatial changes in tornado clima- The greatest increase (105) is in central Tennessee, tology between the two periods. and the greatest decrease (172) is located in north- Tornado counts for Period I captured the classical, central Texas and southwestern Oklahoma. Similar well-known center of Tornado Alley with a gridbox results are noted for the significant-tornado counts. maximum of 477 located in southeastern Oklahoma The summer season (JJA) shows the expected shift and northeastern Texas. In Period II the new maximum northward for both Period I and Period II, with the gridbox value (also 477) is now located in northern greatest decrease (81) located in western Oklahoma. Alabama. Differences in counts from Period I to Period Similar results are shown for the significant-tornado II show respective changes of 2217 and 1154 for these counts. The autumn season (SON) shows increases in grid boxes. Similar compilations for significant torna- tornado counts from Period I to Period II for the Dixie does again show the maximum count (271) in Okla- Alley states extending into northern Indiana, with a homa for Period I, but the new maximum in Period II maximum gridbox value (63) located in western Ten- (185) is located in northern Alabama. Equally impor- nessee. Similar results are shown for the significant tant is the overall decline in significant tornadoes, with tornadoes, with a new value of maximum change (22) the largest decrease (2159) located in southeastern located in western Tennessee, western Kentucky, and Oklahoma and northeastern Texas. Similar results southern . Seasonal tornado days have also have also been shown for the (E)F3–(E)F5 tornado been determined for both tornadoes and significant counts. The field significance test and the bootstrapping tornadoes, and results are consistent and supportive of method of resampling (10 000 times) for both the tor- the findings for the seasonal tornado counts. Changes nado counts and the significant-tornado counts support in seasonal tornado activity from Period I to Period II this geographical shift in the relocation of the center of have accounted in part for the relocation of the center U.S. tornado activity at the 95% confidence level and of annual maximum tornado activity. the 99% confidence level, respectively. Although sev- Although this study has shown a temporal change in eral studies have shown evidence of the Dixie Alley of spatial patterns of tornado activity, no results have tornado events, the results here reveal that there is a been presented to relate this to climate change. It is temporal shift of maximum activity away from the noteworthy, however, that the two periods studied are traditional Tornado Alley. characterized by differences in surface air temperature Equallyimportantinthisstudyhasbeentheex- that may be related to parameters that can influence amination of the number of tornado days, since it can tornado activity. Climate-model predictions of in- be argued that the statistics are dominated by a few big creasing CAPE and weaker shear raise interesting outbreak events. The maximum number of tornado questions regarding the role of climate change in cur- days in Period I (246) is located in southeastern rent and future U.S. tornado climatology. Considerably Oklahoma and northeastern Texas, but in Period II more investigation into the meteorological parameters the new maximum (208) is located in southwestern responsible for the patterns of change in the new tor- Mississippi and eastern Louisiana. The greatest de- nado climatology is warranted, with particular atten- crease (137) is located in southwestern Oklahoma and tion given to the agreement (or lack thereof) with north-central Texas, with the greatest increase (29) in climate-model simulations. central Tennessee. Similar results are noted for the significant and violent tornado days with a general Acknowledgments. The authors are grateful to Purdue overall decline in numbers. University doctoral candidate Kimberly Hoogewind Seasonal considerations of tornado counts and tor- for assistance with the gridbox tornado counts and nado days have been made that show interesting re- their accuracy. Purdue University professor Michael sults that affect the annual totals discussed above. Baldwin is recognized for suggesting use of the classical The winter season (DJF) shows substantial increases bootstrapping method and for assistance in imple- in tornado counts from Period I to Period II, with menting the field significance test. The reviewers are the maximum increase (56) in central Tennessee. also recognized for their many valuable comments and Significant-tornado counts were largely unchanged constructive suggestions that have allowed the authors except for decreases along the Gulf Coast and in- to improve the manuscript. creases across Tennessee and western Kentucky. The spring season (MAM) shows a bifurcation in maxi- mum counts from a single center in central Oklahoma REFERENCES (308) for Period I to two centers in Period II located in Agee, E., and S. Childs, 2014: Adjustments in tornado counts, northern Alabama (283) and central Oklahoma (263). F-scale intensity, and path width for assessing significant

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC AUGUST 2016 A G E E E T A L . 1697

tornado destruction. J. Appl. Meteor. Climatol., 53, 1494– Elmore, K. L., M. E. Baldwin, and D. M. Schultz, 2006: Field sig- 1505, doi:10.1175/JAMC-D-13-0235.1. nificance revisited: Spatial bias errors in forecasts as applied to Ashley, W. S., and S. M. Strader, 2016: Recipe for disaster: How the the Eta Model. Mon. Wea. Rev., 134, 519–531, doi:10.1175/ dynamic ingredients of risk and exposure are changing the MWR3077.1. tornado disaster landscape. Bull. Amer. Meteor. Soc., 97, 767– Elsner, J. B., S. C. Elsner, and T. H. Jagger, 2015: The increasing 786, doi:10.1175/BAMS-D-15-00150.1. efficiency of tornado days in the United States. Climate Dyn., Brooks, H. E., G. W. Carbin, and P. T. Marsh, 2014a: Increased 45, 651–659, doi:10.1007/s00382-014-2277-3. variability of tornado occurrence in the United States. Science, Gensini, V. A., and T. L. Mote, 2015: Downscaled estimates of late 346, 349–352, doi:10.1126/science.1257460. 21st century severe weather from CCSM3. Climatic Change, ——, ——, and ——, 2014b: Monthly US temperatures and tor- 129, 307–321, doi:10.1007/s10584-014-1320-z. nado occurrence. 27th Conf. on Severe Local Storms, Madison, Trapp, R. J., B. A. Halvorson, and N. S. Diffenbaugh, 2007: Tele- WI, Amer. Meteor. Soc., 4A.1. [Available online at https:// scoping, multimodel approaches to evaluate extreme convec- ams.confex.com/ams/27SLS/webprogram/Paper254220.html.] tive weather under future climates. J. Geophys. Res., 112, Diciccio, T. J., and J. P. Romano, 1988: A review of bootstrap D20109, doi:10.1029/2006JD008345. confidence intervals. J. Roy. Stat. Soc., 50, 338–354. ——, E. D. Robinson, M. E. Baldwin, N. S. Diffenbaugh, Diffenbaugh, N. S., M. Scherer, and R. J. Trapp, 2013: Robust in- and B. R. J. Schwedler, 2011: Regional climate of hazardous creases in severe thunderstorm environments in response to convective weather through high-resolution dynamical greenhouse forcing. Proc. Natl. Acad. Sci. USA, 110, 16 361– downscaling. Climate Dyn., 37, 677–688, doi:10.1007/ 16 366, doi:10.1073/pnas.1307758110. s00382-010-0826-y. Dixon, P. G., A. E. Mercer, J. Choi, and J. S. Allen, 2011: Tornado risk Verbout, S. M., H. E. Brooks, L. M. Leslie, and D. M. Schultz, 2006: analysis: Is Dixie Alley an extension of Tornado Alley? Bull. Evolution of the U.S. tornado database: 1954–2003. Wea. Amer. Meteor. Soc., 92, 433–441, doi:10.1175/2010BAMS3102.1. Forecasting, 21, 86–93, doi:10.1175/WAF910.1.

Unauthenticated | Downloaded 10/04/21 12:32 PM UTC