Fig 1 - uploaded by Walker S. Ashley
Content may be subject to copyright.
Source publication
Most U.S. metropolitan regions have experienced urban “sprawl,” or the outward spreading of urban development from city centers.
For cities lying in areas prone to severe weather, the sprawl phenomenon exposes greater numbers of developed areas and inhabitants
to a variety of thunderstorm hazards. This study’s principal goal is to determine how urb...
Citations
... The year 2020 recorded 13 of these events, easily surpassing the previous record of 9 in 2011. The frequency and magnitude of these costly severe storm events has dramatically increased since 1980 and can be primarily attributed to rapidly expanding developed land uses and subsequent increase in exposure to natural hazards (e.g., Hall and Ashley 2008;Changnon 2009;Bouwer 2011;Paulikas and Ashley 2011;Ashley et al. 2014;Rosencrants and Ashley 2015;Strader and Ashley 2015;Strader et al. 2015;Ashley and Strader 2016;Strader et al. 2017a,b;Strader and Ashley 2018;Strader et al. 2018;Childs et al. 2020;Ash et al. 2020). In addition to the expanding footprint of the human-built environment, climatological analyses of the spatial distributions of tornado and severe hail reports, and their favorable environment frequencies, reveal increasing trends in recent decades for both hazards across portions of the midwestern, southeastern, and northeastern United States. ...
Previous studies have identified environmental characteristics that skillfully discriminate between severe and significant-severe weather events, but they have largely been limited by sample size and/or population of predictor variables. Given the heightened societal impacts of significant-severe weather, this topic was revisited using over 150 000 ERA5 reanalysis-derived vertical profiles extracted at the grid-point nearest—and just prior to—tornado and hail reports during the period 1996–2019. Profiles were quality-controlled and used to calculate 84 variables. Several machine learning classification algorithms were trained, tested, and cross-validated on these data to assess skill in predicting severe or significant-severe reports for tornadoes and hail. Random forest classification outperformed all tested methods as measured by cross-validated critical success index scores and area under the receiver operating characteristic curve values. In addition, random forest classification was found to be more reliable than other methods and exhibited negligible frequency bias. The top three most important random forest classification variables for tornadoes were wind speed at 500 hPa, wind speed at 850 hPa, and 0–500-m storm-relative helicity. For hail, storm-relative helicity in the 3–6 km and -10 to -30 °C layers, along with 0–6-km bulk wind shear, were found to be most important. A game theoretic approach was used to help explain the output of the random forest classifiers and establish critical feature thresholds for operational nowcasting and forecasting. A use case of spatial applicability of the random forest model is also presented, demonstrating the potential utility for operational forecasting. Overall, this research supports a growing number of weather and climate studies finding admirable skill in random forest classification applications.
... Overall, this research is motivated in part by previous studies and analyses (e.g., Ashley & Strader, 2016;Ashley et al., 2014;Hall & Ashley, 2008;Paulikas & Ashley, 2011;Rae & Stefkovich, 2000;Rosencrants & Ashley, 2015;Strader et al., 2016b;Wurman et al., 2007). This particular study asks similar questions, but in the context of tornado disaster outcomes across a variety of land use patterns. ...
... Most notably, researchers have investigated how land use is linked to the risk of urban flooding (e.g., Pottier, Penning-Rowsell, Tunstall, & Hubert, 2005;Shepherd, 2005;Brath, Montanari, & Moretti, 2006;O'Connell et al., 2007;Ferguson & Ashley, 2017), landslides (e.g., Leighton, 1976;Sidle & Ochiai, 2006;Sidle, Pearce, & O'Loughlin, 1985), and coastal inundation (Wheater & Evans, 2009). In addition, studies (Ashley & Strader, 2016;Ashley et al., 2014;Hall & Ashley, 2008;Paulikas & Ashley, 2011;Rae & Stefkovich, 2000;Rosencrants & Ashley, 2015;Strader et al., 2016b;Wurman et al., 2007) have investigated the role large population centers, population growth, and urban sprawl serve in influencing tornado impacts. Others (i.e., Hall & Ashley, 2008;Ashley et al., 2014;Rosencrants & Ashley, 2015;Ashley & Strader, 2016;Strader et al., 2016b) have focused on how changes in population and land use, especially in the form of suburban and exurban sprawl, is leading to greater numbers of people and homes potentially in harm's way and, moreover, increasing tornado disaster potential. ...
... The synthetics are employed to simulate possible hurricane impacts across a spatially diverse risk landscape. Previous studies have exemplified the benefits of hazard synthetics in assessing potential disasters to the human-built environment, primarily for the tornado hazard (Wurman et al. 2007;Hall and Ashley 2008;Paulikas and Ashley 2011;Ashley et al. 2014;Strader et al. 2014). Hurricane synthetics have been used to evaluate simulated lifecycle and intensity through historical storm data (Casson and Coles 2000;Vickery et al. 2000;Emanuel 2006;Emanuel et al. 2006;Hall and Jewson 2007;Hallegatte 2007;Rumpf et al. 2007Rumpf et al. , 2009Yonekura and Hall 2011;Nakamura et al. 2015;Ellis et al. 2015Ellis et al. , 2016. ...
This research appraises how residential built environment growth influences coastal exposure and how this component of societal vulnerability contributes to tropical cyclone impact and disaster potential. Historical housing unit data and future demographic projections from a high-resolution, spatial allocation model illustrate that the area within 50 km of the US Atlantic and Gulf Coastlines has the greatest housing unit density of any physiographic region in the USA, with residential development in this region outpacing non-coastal areas. Tropical cyclone exposure for six at-risk metropolitan statistical areas (MSAs) along the US Atlantic and Gulf Coasts are assessed. All six MSAs evaluated are distinct in their development character, yet all experienced significant growth from 1940 through the contemporary period; projections from the model under various socioeconomic pathways reveal that this growth is anticipated to continue during the twenty-first century. Using a worst-case scenario framework, the historical and future residential data for the six MSAs are intersected with synthetic hurricane wind swaths generated from contemporary landfalling events. The New York City MSA contains the greatest residential built environment exposure, but Miami is the most rapidly changing MSA and has the greatest potential for hurricane disaster occurrence based on the juxtaposition of climatological risk and exposure. A disaster potential metric illustrates that all six MSAs will experience significant increases in disaster probability during the twenty-first century. This analysis facilitates a detailed spatiotemporal assessment of US coastal region vulnerability, providing decision makers with information that may be used to evaluate the potential for tropical cyclone disasters, mitigate tropical cyclone hazard impacts, and build community resilience for these and other hazards in the face of environmental and societal change.
... The region has undergone steady growth in population since 1990. As people have flocked to Atlanta, the boundaries of the urban area have been pushed outward at a rapid pace, giving the city a sprawling character (Yang and Lo 2003) with attendant increases in impervious surface (Rose et al. 2008) where forest and cropland were previously located (Yang and Lo 2002), amplifying exposure to thunderstorm hazards (Paulikas and Ashley 2011). Additionally, precipitation patterns have changed due to urban effects (Shepherd et al. 2002;Bentley et al. 2010;Ashley et al. 2012;Haberlie et al. 2015). ...
This research examines changes in residential built-environment flood exposure within the current boundaries of the Atlanta, Georgia metropolitan statistical area, by estimating the number of housing units that are located within the floodplains of the region. Housing unit data at the block level from the 1990 to 2010 decennial censuses are used to estimate housing unit exposure to floods using a binary dasymetric and proportional allocation method. Three different representations of the 100-year (1 percent annual chance) and 500-year (0.2 percent annual chance) floodplain are employed: the generally more conservative floodplains created using the Federal Emergency Management Agency’s Hazus-MH software, the generally more extensive floodplains included in the proprietary Flood Hazard Data product from KatRisk LLC and the regulatory floodplains from the National Flood Insurance Program. The number of housing units within both the 100- and 500-year floodplain increased from 1990 to 2010 throughout the Atlanta region. Housing unit growth within the regulatory 100-year flood zone was slower than growth elsewhere, suggesting that the National Flood Insurance Program may have been marginally effective overall. Results using the KatRisk product reveal both greater overall and a greater increase in housing units at risk within the 100-year floodplain than the regulatory product suggests. The results argue that heightened flood exposure, particularly in areas experiencing new development, is an important factor to consider when addressing the impact of the flood hazard over time.
... In addition, the US has become more urbanized over this time frame with greater percentages of residents being concentrated around city population centers (U.S. Census 1995Census , 2015. Multiple studies (Hall and Ashley 2008;Paulikas and Ashley 2011;Ashley et al. 2014;Ashley and Strader 2015) have illustrated that this urbanization of America could potentially place greater numbers of lives at risk from a singular tornado event today than in years past. ...
... In our Introduction, we note that tornado risk is, in part, a function of US demographics patterns. It has been well established that tornadoes striking populated areas pose a greater risk to more lives and property (Paulikas and Ashley 2011;Ashley et al. 2014;Ashley and Strader 2015). Beyond this observation, it is unclear which other socioeconomic factors may have likely contributed to the death outcomes assessed for this study. ...
Motor vehicles historically have been dangerous locations to shelter in during tornado events. Throughout the twentieth century, motor vehicle design has become safer while tornado forecasting has become better understood. Despite such advances, tornado fatalities in motor vehicles still occur today, and some events periodically result in high numbers of deaths (e.g., ten motor vehicle occupants were killed by a single tornado in Garland, Texas, in 2015). We seek to examine all US tornado-induced motor vehicle fatalities documented between 1991 and 2015. Our findings indicate that motor vehicle fatalities have not significantly changed during this study period. We attribute annual fatality totals to persons lacking awareness of impending dangers coupled with numbers of significant tornado events for a given year. We find most fatalities result when vehicles are lofted or passengers are ejected, and this most typically occurs at the EF3–EF5 intensity thresholds. Fatalities that occur at weaker tornado winds (EF0–EF2) are most often attributed to collapsing debris (mostly trees) on vehicles. Spatially, motor vehicle fatalities are greatest in the Deep South and southern Great Plains regions where overall tornado and nighttime tornado frequencies are greatest. Some of the largest motor vehicle fatality events have resulted from tornadoes not being distinctly visible to motorists; such events have been characterized by tornadoes occurring at night or by tornadoes not appearing as “classic funnels.”
... To test the precision of both methodological approaches, a potential follow-up to this study may involve assessing residential tornado-induced exposure and fatality projections exclusively on the basis of population and housing data originating from U.S. census blocks which intersect the synthetic damage path (as displayed in Fig. 3.15). Such an analysis would largely adopt the methodologies of previous tornado simulation studies (WUR 2007;Hall and Ashley 2008;Paulikas and Ashley 2010;Ashley et al. 2014), and would test the efficiency of the population distribution methods employed by this study. ...
Tornado vulnerability may be regarded as the likelihood of fatalities that may result from a tornado strike. This likelihood is partly based on basement availability and the types of structures individuals may congregate in during the time of a tornado event, as weaker structures and limited basement availability have been associated with higher tornado fatality rates in the past. Tornado fatality statistics are also a function of wind speeds, structural damage patterns, and the total numbers of persons exposed to such dangerous weather conditions. This study first assesses urban population patterns for three Chicago suburbs to determine the likeliest locations where individuals may be present over the course of a 24-hour daily period; the damage path of the 2011 Joplin, MO tornado is subsequently replicated over these three municipalities to project the numbers of individuals exposed to such an event. The identification of population location patterns, coupled with basement availability assumptions and projected tornado wind damage patterns, provide a basis for simulating fatality estimates under the hypothesis that a violent tornado strikes one populated area at different times of day. The results of this study suggest that the simulated tornado events of this project may be about as lethal as the Tuscaloosa-Birmingham, AL tornado event of 2011. In addition, findings from this study indicate that basement availability in the Chicago area is far more prevalent than in areas such as Joplin, MO,
and fatality estimates would fall under the 158 killed in Joplin in 2011. The methodological elements of this study may serve to identify communities that may be particularly vulnerable to violent tornado events, as high death toll outcomes in the past have resulted, in part, from a limited availability of basements and public shelters.
... Compounding this issue, the atmospheric effects caused by developed land use include an increased probability of thunderstorm activity over and around a city (e.g. Dixon and Mote, 2003;Shepherd, 2005;Mote et al, 2007;Bentley et al, 2010;Paulikas and Ashley, 2011;Ashley et al, 2012;Coquillat et al, 2012;Ganeshan et al, 2013;Stallins et al, 2013). The combination of these factors results in areas of enhanced exposure in urban areas to thunderstorm hazards. ...
http://onlinelibrary.wiley.com/doi/10.1002/qj.2499/abstract
http://chubasco.niu.edu/pubs/Haberlie%20et%20al.%202015%20QJRMS.pdf
This study assesses the impact of urban land use on the climatological distribution of thunderstorm initiation occurrences in the humid subtropical region of the Southeast United States, which includes the Atlanta, Georgia metropolitan area. Initially, an automated technique is developed to extract the locations of isolated convective initiation (ICI) events from 17 years (1997-2013) of composite reflectivity radar data for the study area. Nearly 26,000 ICI points were detected during 85 warm-season months, providing the foundation for first long-term, systematic assessment of the influence of urban land use on thunderstorm development. Results reveal that ICI events occur more often over the urban area compared to its surrounding rural counterparts, confirming that anthropogenic-induced changes in land cover in moist tropical environments lead to more initiation events, resulting thunderstorms, and affiliated hazards over the developed area. The ICI risk for Atlanta is greatest during the late afternoon and early evening in July and August in synoptically benign conditions. Greater ICI counts downwind of Atlanta suggest that prevailing wind direction also influences the location of these events. Moreover, ICI occurrences over the city were significantly higher on weekdays compared to weekend days—a result that was not apparent in a rural control region located west of the city. This suggests that the weekly commuting cycle and associated aerosol levels of Atlanta may amplify ICI rates. The investigation provides a methodological framework for future studies that examine the effect of land use, land cover, and terrain discontinuities on the spatiotemporal character of ICI events.
... Research examining the changes in the disaster landscape suggests that the amplification of disasters is driven largely by increases in population, wealth, and inflation (Kunkel et al. 1999;Changnon et al. 2000;Brooks and Doswell 2001;Pielke et al. 2008;Changnon 2010;Bouwer 2011;Simmons et al. 2013). Sprawl development, which has been a dominant growth mode in the USA the past 70 years, has led to the dramatic expansion of most metropolitan regions, placing more people and property in harm's way of geophysical hazards (Changnon and Burroughs 2003;Hall and Ashley 2008;Bouwer 2011;Morss et al. 2011;Paulikas and Ashley 2011;Ashley et al. 2014). Other studies have investigated possible climate change effects of microscale and mesoscale meteorological hazards (Trapp et al. 2007;Peterson et al. 2008;Kunkel et al. 2012;Brooks 2013;Diffenbaugh et al. 2013). ...
... Based on the results from this study and others (Changnon 2003;Changnon and Burroughs 2003;Hall and Ashley 2008;Paulikas and Ashley 2011;Shepherd et al. 2011;Bouwer 2013;Ashley et al. 2014), exposure can be seen as a driving force in the increasing impacts to humans from severe weather. The ''expanding bull's eye effect'' (Ashley et al. 2014) provides a conceptual model of the increasing exposure within metropolitan areas to hazard occurrence. ...
... In comparison with previous macroscale research on exposure (Wurman et al. 2007;Hall and Ashley 2008;Paulikas and Ashley 2011;Morss et al. 2011;Bouwer 2013;Ashley et al. 2014), this research promoted a spatially detailed examination of exposure changes across complex and continually evolving urban landscapes. The results suggest that, while there is generally macroscale expansion in development, the spatiotemporal trends at the microscale are multifaceted and in a continual state of flux. ...
Weather-related disasters and affiliated losses in the USA have amplified over time. However, prior research using normalization schemes on damage tallies suggests that weather hazard losses are not necessarily rising when inflation, changes in wealth, and growth in population are accounted. This study evaluates the latter factor, assessing if population changes and a sprawling development mode have led to increasing potential for tornado disasters in the USA. Specifically, this research shows where and how quickly tornado exposure is growing by appraising spatiotemporal trends in gridded population and housing unit data for five metropolitan statistical areas (MSAs). The macroscale risk to tornadoes is represented by tornado day climatology and is related to the exposure of the five MSAs, which include Atlanta, GA; Chicago, IL; Dallas/Fort Worth, TX; Oklahoma City, OK; and St. Louis, MO. Supplementing the macroscale investigation, an observationally derived, hypothetical violent tornado track is transposed on various development types in each MSA to determine the microscale changes in human and built-environment exposure. Results demonstrate increased exposure in all MSAs at both the macro- and microscale. Of the five MSAs studied, Dallas, TX, had the greatest potential for a tornado disaster due to the higher risk for tornado occurrence comingling with the amount of MSA exposure. These results reveal further that amplifying exposure is a major impetus behind intensifying severe weather impacts and losses.
... Recent studies have illustrated that scenario strategies can be applied to evaluate the potential of tornado disasters by relying on GIS-ready damage surveys and measurements conducted by the NWS (Rae and Stefkovich, 2000;Paulikas and Ashley, 2011;Ashley et al., 2014), engineers (Marshall et al., 2012a;Ashley et al., 2014) and/or wind measurements from mobile Doppler radars (Wurman et al., 2007;Ashley et al., 2014). These investigations have examined scenarios focused on potential 'worst-case' (Clarke, 2005) effects on populations and the built environment (Wurman et al., 2007;Hall and Ashley, 2008;Paulikas and Ashley, 2011;Ashley et al., 2014). ...
... Recent studies have illustrated that scenario strategies can be applied to evaluate the potential of tornado disasters by relying on GIS-ready damage surveys and measurements conducted by the NWS (Rae and Stefkovich, 2000;Paulikas and Ashley, 2011;Ashley et al., 2014), engineers (Marshall et al., 2012a;Ashley et al., 2014) and/or wind measurements from mobile Doppler radars (Wurman et al., 2007;Ashley et al., 2014). These investigations have examined scenarios focused on potential 'worst-case' (Clarke, 2005) effects on populations and the built environment (Wurman et al., 2007;Hall and Ashley, 2008;Paulikas and Ashley, 2011;Ashley et al., 2014). A primary weakness of these studies is the overall small sample size of tornadoes used to derive models, as well as the exclusion of varying types of tornado intensity assessment methodologies. ...
An increasing number of significant and violent tornado events in the United States have been documented and mapped at extremely high resolution by government, research and private entities using remotely sensed and post-event damage surveys; however, these assessments often generate inconsistent spatial measures of tornado strength, even for the same event. This investigation assembles a portfolio of contemporary tornado events that contain spatially comprehensive damage and/or wind velocity information from a diverse set of sources. Thereafter, the relationship between land-use/cover and tornado intensity is examined in order to quantify spatial measures of damage indicator bias in post-event tornado damage surveys. A climatology of both significant and violent tornado intensity assessments is then created, promoting the generation of synthetic, or model, paths with observationally constrained damage length and width metrics by the Enhanced Fujita scale magnitude. Results from the climatology and collection of synthetic paths are compared to previous observed, empirical and theoretical assessments, revealing differences in the spatial scale of the overall tornado footprint, as well as percentage contribution of swaths by Enhanced Fujita scale magnitude. The range of synthetic paths produced may be used to assess potential tornado damages to the population, the built environment and insurance portfolios.
... Throughout the ensuing 50 years a series of job booms coupled with a relatively flat landscape allowed for rapid development to occur throughout the region, as 28 counties comprised the Atlanta MSA by the year 2003 (Cromartie, 2001;Squires, 2002;Cavin, 2003;Metro Atlanta Chamber of Commerce, 2005). Findings by Paulikas and Ashley (2010) indicate that this expansion would result in severe thunderstorm hazards affecting more lives and structures per event in later years. ...
... The methodology used by Paulikas and Ashley (2010) consisted of subdividing 28 Atlanta MSA counties into 3 'ring' sections (an Urban Core, Inner Ring and Outer Ring of counties), which revealed different rates of population growth found throughout the study region. The Urban Core of counties is comprised of the most historically developed counties in the Atlanta MSA with the highest population and housing unit density values (Paulikas and Ashley, 2010). ...
... The methodology used by Paulikas and Ashley (2010) consisted of subdividing 28 Atlanta MSA counties into 3 'ring' sections (an Urban Core, Inner Ring and Outer Ring of counties), which revealed different rates of population growth found throughout the study region. The Urban Core of counties is comprised of the most historically developed counties in the Atlanta MSA with the highest population and housing unit density values (Paulikas and Ashley, 2010). The Inner Ring comprises counties lying immediately adjacent to the Urban Core which initially were very rural but have witnessed massive development since 1960 in areas to the north of the city limits (Paulikas and Ashley, 2010). ...
Historic severe thunderstorm hazard reports consist of spatial and temporal disparities in reporting frequencies, which are attributed partially to changes in assessment and reporting techniques witnessed throughout the U.S. over the years. Population bias, however, has been cited as a key underlying factor creating inhomogeneities in severe weather reporting frequencies in the past. Given the nature of the historic population distribution patterns found throughout the Atlanta, GA metropolitan region, this location serves as a focus to determine potential statistical associations that may exist between population numbers and historic severe weather hazard reporting frequencies. To address this potential association, statistical one-way ANOVA, post-hoc and correlation tests are conducted for this study. Results indicate that Atlanta-area severe wind and hail reporting frequency patterns display some significant associations with respect to the region's historic population patterns. Tornado reporting frequencies conversely display the weakest statistical associations, and are least affected by the area's human settlement patterns since 1960.
