An Investigation of Spatial Redistribution of USA Tornado Activity and Its Relationship to Climate Change

The effects of global warming on a regional scale can potentially lead to changes in weather phenomena, such as more drought, floods, heat waves, and extreme weather in general. However, the limited information of any climate change impacts on USA tornado activity warrants further research investigation. This study begins by considering the change in surface air temperature over the CONUS to determine possible climatic shifts, with an emphasis on the 60-year period from 1954-2013. More specifically, two successive 30-year periods for a 20-state region within a domain encompassing tornado alley (30-50°N, 80-105°W), have been investigated. Two statistically different climatic regimes have been found, consisting of a Cold Period I (1954-1983) and a Warm Period II (1984-2013). Warming of 1.5°F for the northern half of the region and 0.8°F for the southern half is noted, and the northern half also had a much larger increase in the variance compared to the south. An R-statistic has been introduced to quantify these temperature patterns. Next, a state-by-state count of (E)F1-(E)F5 tornadoes is compared for each period, and large differences can be noted. The greatest increase in counts is seen across the Southeast in states such as Tennessee (+182) and Mississippi (+174), with the greatest decreases for Oklahoma (-383) and Missouri (-153). Seasonal tabulations for these changes reveal that the summer months (JJA) account for much of the reduction, while the months of (MAM) and (SON) are the seasons that contribute most to TN and MS increases. These state-by-state results were further refined by the use of NCEP/NCAR reanalysis data, applied to the gridded domain, which covers the entire data record (1954-2013). The even more refined NARR reanalysis data only covers a portion of the study period (1979-2013), but a comparison of surface temperature for the two data sets for Period II reveals similar patterns, which encourages the use of the NCEP/NCAR data for the entire period. When applied to tornado climatology, a similar geospatial trend is found, namely increasing tornadoes in the southeast and decreasing tornadoes to the west and north. Each grid box (2.5° x 2.5°) receives a count of one tornado for each of the following: a) tornado entirely contained in the grid box, b) tornado starts in the grid box and ends elsewhere, and c) tornado comes from elsewhere and ends in the grid box. Six grid boxes showing the largest change between the two periods were tested for statistical significance and four of the six showed significant change, namely those encompassing central and southern Oklahoma and central Tennessee.

It is important that statistics support the above results, but it is even more important that appropriate observational analyses of meteorological parameters that produce tornadoes be considered and found supportive. These parameters include CAPE, CIN, stability indices, moisture fields, shear and helicity. Accordingly, mean values from the NCEP/NCAR reanalysis data were considered for the two periods for some of these parameters. Lifted Index (LI) and Precipitable Water (PWAT) patterns and their changes have been determined, which support the existence of more convective instability and available moisture across the Southeast. Wind shear between the 925mb and 850mb levels has been compared between the two periods, and virtually no change is noted. Further, CAPE was examined using the NARR dataset for two decades (1980-89 and 2000-09), since it was not readily available in the NCEP/NCAR dataset. Comparison of these two decades also shows little to no change in CAPE. The results for both shear and CAPE support the general concept that anticipated changes with global warming in the 21st century has not yet appeared on this temporal and spatial scale. Further investigation of changes in the USA tornado climatology that have been identified in this study will require targeted computations that show the variability of CAPE and shear on a much smaller time scale than monthly mean values. Such could show more meteorological consistency with declining counts of tornado days, as well as violent tornado day trends due to corresponding episodic peaks in CAPE and shear. In summary, considerable evidence has been found to indicate a geospatial shift in USA tornado activity from the heart of tornado alley to the Southeast, based on both observational and statistical analyses for two distinctively different time periods that show evidence of climate change.