Abstract
Tornado hazards have caused frequent and severe damages to built environment and lives. The severity of the hazards directly relates to the strength of the tornadic events in terms of meteorological parameters, as well as the resistance of buildings to the strong tornadic impacts. The tornado hazards are reviewed from both meteorological and engineering perspectives in this study. For each of the perspectives, current status of tornado hazard studies including the observation, monitoring, analysis, numerical and laboratory simulations, and limitations and future demands of development are discussed. While the two disciplines are fast advancing individually, their interaction and interdependence are deemed prominent and imperative in the future interdisciplinary studies.
Similar content being viewed by others
1 Introduction
This paper reviews the tornado hazard from both meteorological and engineering perspectives. The advancements in meteorology and engineering structural analysis have improved tornado monitoring, prediction, and understanding of potential impacts on the built environment, while certain proactive measures have been taken in improving building code design. The primary focus of this paper is on reviewing the development in each individual discipline as well as their interdisciplinary studies, which leads to discussions of demands for an enhanced interdisciplinary collaboration, so that the advancements in each discipline can be fully integrated to reach the highest potential in mitigating tornado hazards and sustainability of the built environment.
Tornado disasters have been occurring with increasing frequency in the USA and are projected to become more frequent and severe in the coming decades (IPCC, 2014). Tornados in the USA are as deadly and costly as hurricanes. Between 1980 and 2011, 43 % of windstorm losses were attributed to severe thunderstorms and tornados, which is equivalent to 50 % for hurricanes (Lloyd’s 2013). The year of 2011 was a record year for tornados. There were 1,718 tornados reported with a total loss of $25 billion (NOAA National Climatic Data Center 2011). Tornados damage civil structures, claim lives, and deprive people of their basic necessities. Among recent U.S. deadly disasters were the 2011 tornado in Tuscaloosa, Alabama (64 deaths, 1,500 injured, estimated damage $2.4 billion), the 2011 tornado in Joplin, Missouri (158 deaths, 1,150 injured, estimated damage $2.8 billion), and the 2013 tornado in Moore, Oklahoma (24 deaths, 377 injured, estimated damage $2 billion).
Limited knowledge of tornado structure and tornado wind fields resulted from the traditional wind speed estimation by damage-indicator assessment, which depended largely on the assumed horizontal straight-line winds (Senguipta et al. 2008; Wurman et al. 2012a). Moreover, the acquisition and evolution of the Storm Data (McDonald 2001; Verbout et al. 2006) implied its limitations in data quality because of factors such as the complexity of the wind speed–damage relationship, unavailability or heterogeneity of damage indicators, distribution and change of human population, and observation and communication technology (Doswell et al. 2009). Therefore, different methods and statistical models have been proposed in tornado climatology studies to estimate potential tornado damage from the probability of occurrence, the probability of damaging wind exceedance, and the spatial distribution, e.g., (Edwards 2013; Refan 2014). These statistical climatology studies take the perspective at scales that are much larger than that of tornados.
In addition to the statistical studies and with the consideration of buildings as surface-mounted obstacles, numerous numerical studies, such as the finite element method (FEM), have been used to investigate building structures under the impacts of wind loads and flow patterns around the surface-mounted obstacles (Gu et al. 2010; Hunt et al. 1978; Martinuzzi and Tropea 1993; Natarajan and Chyu 1994; Yakhot et al. 2006). Laboratory-based tornado-like vortex simulators are another area of research to investigate the characteristics of vortex flows and how they affect the microscale building blocks put in wind tunnel or vortex chambers (Refan 2014). While these laboratory and numerical efforts have revealed a great deal about the wind loads and flow structures around buildings, most of the previous studies were conducted with the assumption of straight-line boundary-layer flows or steady-state conditions. In particular, the vortex simulators investigate the tornado-like features at scales that are much smaller than that of tornados. Consequently, very limited knowledge about tornado impacts has been used in developing building codes.
Tornados are strong vortices with a significant tangential component, radial inflow/outflow, and vertical updrafts/downdrafts. It is desirable that more comprehensive tornadic wind fields be studied based on recent advancement in technology of observation, numerical simulation of tornados, and potentially realistic building load quantifications. In addition to the traditional parameters including wind speed and pressure, new tornadic parameters such as rotational wind speed, central pressure change and gradient, and vertical suction wind loads need to be derived from high-resolution tornado simulations. Then, detailed loads added onto buildings can be calculated and analyzed, and eventually, a more realistic tornado impacts on buildings can be quantified and used in building design code design.
2 Tornado hazard from a meteorological perspective
2.1 Meteorological characteristics of tornados
Meteorologically, tornados are able to generate strong rotational winds that can result in gust wind speeds greater than about 140 m s−1 (313 mph) at 100 m above ground level (AGL) when combined with translational movement of tornados (Wurman et al. 2014). Regardless of the small spatial scale (width ranges from 100 to 4000 m), tornados can cause abrupt central pressure drop up to 194 mb (Karstens et al. 2010). Tornados spawn within a rotational vortex of airflow called mesocyclone in a thunderstorm. After the formation, tornados move along with the parent thunderstorm, resulting in a tornado path that has an average width of a few hundred meters and an average length of a few kilometers. The rotational vortex also involves vertical motion where an upward suction wind velocity of up to 14 m s−1 (31 mph) was estimated for an EF2 tornado (Wurman and Kosiba 2013) and up to 60 m s−1 (134 mph) from model simulations (The Bryan Cloud Model results (Bryan and Fritsch 2002)). Although these are the main factors that may induce loads on building structures and cause damages, most of the existing tornado intensity studies focus on wind speed only, which is not usually available in direct observation and, instead, relies on aftermath estimation from damage assessments (Edwards et al. 2013).
2.2 Observations and datasets
2.2.1 Observations
Tornados occur mostly at small spatial scales that average at 600 m in diameter. Due also to the nature of the infrequent occurrence and its violent intensity, very few direct instrumental observations of tornados, including winds, its internal structure, and formation, are available. With the very recent 2009–2010 field program, the second Verification of Origins of Rotation in Tornadoes Experiment (VORTEX2 (Wurman et al. 2012a)), a couple of medium strength tornados have been captured by Doppler on Wheels (DOW) radar (about 1 % out of ~ 1,200 tornados reported annually), photographic video filming, and in situ observations inside a tornado (fewer than a few tens total) (Ashley 2007; Doswell et al. 2009; Edwards et al. 2013; Standohar-Alfano and van de Lindt 2014; Standohar-Alfano et al. 2014; Verbout et al. 2006). These invaluable observations provided support to earlier theories about tornado formation and structure of damaging wind flow in some degree.
2.2.2 Damage assessments and surveys
For most of the tornados, their intensity were estimated based on aftermath damage survey and assessment according to the Fujita Scale (F-scale), which was developed by T. Theodore Fujita (1971) and adopted in late 1970s by the US National Weather Service (NWS). All reported tornados were assigned an F-scale (F0–F5) by NWS personnel (McDonald 2001; Verbout et al. 2006), and the resulting Storm Data were archived at the National Climatic Data Center (NCDC, http:/www.ncdc.noaa.gov/). The dataset contains reports of tornado location by latitude and longitude coordinates of starting and ending points, intensity (as an F-scale), and damage path length and width. Based on the F-scale, records of historical tornados from 1950 to 1976 were also assigned F-scale ratings (Kelly et al. 1978). Historical records of F2 and stronger tornados dated back to 1871 and killer tornados back to 1680, were also compiled (Edwards et al. 2013; Grazulis 1993; Grazulis et al. 1993). In February 2007, NWS implemented an Enhanced Fujita Scale (EF-scale) considering mainly the improvement in building materials and technologies (Doswell et al. 2009; Potter 2007).
Although the official Storm Data at NCDC contains data from 1950 up to current, studies in the past two decades often used different sub-periods of the data. The separating points corresponded to the adoption of F- and EF-scales, as well as the time when almost full coverage of Doppler weather radar was available in the USA, which coincided the change from recording the maximum path width to average path width in 1995 (Brooks 2004). One other obvious change in the Storm Data was the increasing trend in the number of EF0 tornados, which related to the changes in population, reporting technologies, public awareness, and increased tornado chasing interests (Elsner et al. 2014; Elsner and Widen 2014). Nonetheless, unreported tornados in the earlier years do not significantly affect study of strike probability and maximum wind speed when only EF1+ tornados are considered (Elsner et al. 2014; Ramsdell and Andrews 1986).
Edwards et al. (2013) thoroughly reviewed the tornado intensity estimation studies of past and present and contemplated on future research directions. Currently, the wind damage assessment has been traditionally dependent on horizontal straight-line winds although attempts have been made in studying rotational wind impacts (e.g., Senguipta et al. 2008; Wurman et al. 2012a). As a result, the impacts from rotational winds, pressure drop, and vertical suction winds have not been considered in any existing building codes.
2.2.3 Modern technologies
In recent years, close observation of tornado wind speeds has been attempted. Mobile radar systems including DOW radar, pulsed Doppler lidar, and high-resolution X-band polarimetric radar have been used in field observations and obtained some detailed tornado wind fields (Wurman et al. 2007; Wurman and Kosiba 2013; Bluestein et al. 2014; Snyder and Bluestein 2014). The integrated radar–lidar systems take the advantage of radar and lidar capabilities (Bluestein et al. 2014). Lidar provides high-resolution near-distance wind observations in clear air, while radar is capable of observing far-distance features even when precipitation or insects exist. However, ground obstacles hinder the acquisition of near surface winds from radar (Wurman et al. 2014). The lack of dense coverage of radars limits such observations to a very limited number of tornado case studies. In addition to radar observations, instrumented and armored tornado intercept vehicles can be built and take observations inside tornados (Wurman et al. 2007; Wurman and Kosiba 2013). At larger scales, satellite and space-borne observations have the potential to provide tornadic environments (e.g., Akiya et al. 2014; Gravelle et al. 2016). Indirect observations including photometric and video analyses have been used in tornado studies (e.g., Wurman and Kosiba 2013). With the fast-advancing technologies, low-cost wireless sensors and global positioning system (GPS)-based instruments might be developed for use in observing tornados (Krehbiel et al. 2000; Otsuka et al. 2014), as well as development of people-centric sensing (Phillips and Sankar 2013) network that collects information from the public.
2.3 Tornado climatology
Study of tornado climatology advances the understanding about the characteristics of tornado occurrences and potential threats to human society and economy. It also provides guidance for long-term planning and the welfare of human life, including preventive and protective measures that can reduce the tornado impact to a minimum. The tornado climatology has encompassed studies of tornado intensity, path length and width, probability distribution of damaging winds, as well as tornado frequencies of different temporal scales from daily, monthly, seasonal, to annual. Different statistical models have been used to describe the characteristics of these parameters. It has been found that the Weibull distribution can be applied to the distributions of tornado intensity and wind speed (Dotzek et al. 2005; Feuerstein et al. 2005), as well as path length and width (Brooks 2004). Elsner et al. (2014) and Elsner and Widen (2014) found that seasonal tornado occurrence could be fitted by negative binomial distribution; while Tippett et al. (2012) noticed that monthly tornado frequency follows Poisson distribution. More recent studies showed interest at daily time scales. Brooks et al. (2003) studied the probability of tornado touch down on any day at a location regardless of the number or intensity of the tornados. Elsner et al. (2014) found that the frequency of days with tornados could be described by power law and that the frequency of tornados by category in a day follows exponential law. Malamud and Turcotte (2012) found that the daily frequency of tornado path length also follows power law.
For the purpose of engineering application, Ramsdell et al. (2007) generated tornado strike probability maps and maximum wind speed maps for use in the development of engineering design-basis tornado wind speeds for nuclear power plant sites in the USA. The strike probability and maximum wind speeds were estimated based on NCDC Storm Data. Wind speed estimation was based on EF-scale (3 s average) with consideration of wind speed variation in the damage area both across and along the tornado path. The design-basis wind speeds were based on tornado strike probabilities for point structure and lifeline (finite) structure and the probability of wind exceeding certain wind speed when a tornado strike occurs.
Due to the nature of the tornado intensity estimation, the Storm Data recorded the maximum intensities. However, as many researchers noticed, within the areal coverage of a tornado, the most intensive stage of a tornado covers only a small portion of the impact area and different levels of intensity are found across the tornado footprint (e.g., Paul and Stimers 2014; Refan 2014; Simmons et al. 2013; Strader et al. 2015). Figure 1 shows an example of a well-assessed and studied tornado path with different area for each intensity level resulted from 4222 damage indicators (Atkins et al. 2014; Burgess et al. 2014). In dealt with the complexity of damage–wind relationship, many models have been developed and used to calculate the area of each intensity level within a tornado impact area (e.g., Abbey and Fujita 1975; Ashley et al. 2014; Ramsdell and Andrews 1986; Strader et al. 2015). The Rankine vortex model (Wood and Brown 2011) was a widely used model to simulate the intensity (wind speed) distribution within a tornado footprint along and across the tornado path (e.g., Reinhold and Ellingwood 1982). Others combined the Rankine vortex model with Weibull distribution of wind speed along and across the path to calculate the intensity distribution (e.g., Refan 2014).
Based on the fact that different methods resulted in inconsistent tornado intensity estimations, Strader et al. (2015) selected a set of 114 well-recorded tornados from post-damage survey, structure-by-structure analysis, and mobile radar observations to establish synthetic tornado path climatology. Tornado intensity distribution, which was defined for each EF category as the total area of the damage swath by the corresponding level of intensity, along with land use/land cover information that were used in the categorization of the tornado intensities. They also established a synthetic tornado path climatology based on the most intense and well-observed 1 km segment of each tornado. The result was compared to the pre-1994 period and modern (1995–2014) period. It was suggested that this synthetic tornado path could be used in guidance of engineering design and other related applications.
These aforementioned models are primarily statistical models involving probability distribution functions and an idealized Rankine vortex model in an attempt to reproduce the spatial and temporal distributions of tornado damage areas and frequencies for each intensity level (Simmons et al. 2013; Strader et al. 2015; Ramsdell and Andrews 1986; Ramsdell et al. 2007; Ashley et al. 2014; Paul and Stimers 2014; Refan 2014; Strader et al. 2015). The resulting tornado climatology has provided mostly information about tornado occurrence and damage potentials at scales that are much larger than tornados themselves.
2.4 Physical laboratory simulators of tornado-like vortex
Tornado simulation has been attempted in laboratory settings. These tornado simulators include the Purdue University tornado simulator (Church et al. 1977), Texas Tech University tornado simulator (Mishra et al. 2008a, b), Iowa State University tornado simulator (Haan et al. 2008), and the Model WindEEE Dome at Western University (Natarajan 2011), as summarized by (Refan 2014). By controlling the inflow of air and suction strengths to adjust parameters such as aspect and swirl ratios, tornado-like vortexes can be generated in-house, and wind impacts on objects can be observed and analyzed. However, these likenesses of tornados are still far from the complexity of real tornados, and the objects in the impact studies are at microscales that are hundreds of times smaller than real buildings (Refan et al. 2014), as it is for the tornado-like vortex to real tornados.
2.5 Numerical simulations of tornados
2.5.1 Aerodynamic simulations
A computational fluid dynamics (CFD) model for small-scale tornados was developed by Hangan and Kim (2008). This model simulates tornado’s tangential velocity, radial velocity, and axial velocity fields in steady-state conditions. The CFD analysis was conducted using swirl ratio as a variable, which is the ratio between the tangential and radial velocity at the inlet boundary. The wind field produced by this CFD analysis was compared to the experimental data from Baker (1981) to validate the CFD model. Hangan and Kim (2008) found a good F4 tornado wind field model, and (Hamada et al. 2010) indicated a good simulation of a F2 tornado wind field. These simulations are also at scales much smaller than real tornados and buildings.
Through an experimental study, Yang et al. (2011) investigated the characteristics of wake vortex and flow structures around a high-rise building model that is placed in tornado-like winds. The resultant wind loads (both forces and moments) acting on the test model were measured and correlated to the flow field measurements in order to explicate the underlying physics of the flow–structure interactions between the tested high-rise building model and tornado-like vortex. The results revealed clearly that the evolution of the wake vortex and turbulent flow structures around the test building model as well as the resultant wind loads induced by tornado-like winds were significantly different from those in straight-line winds.
2.5.2 Atmospheric simulations
The most recent, promising, and exciting advancement in tornado modeling lies in fluid dynamics-based atmospheric modeling. Numerical weather prediction has advanced in the past half-century and is now able to generate 1–3 day weather forecasts with high accuracy. Numerical modeling of small-scale weather events including thunderstorms, downbursts, gravity waves, and tornados has evolved tremendously with the availability of high-performance computers (e.g., Bryan and Fritsch 2002; Markowski et al. 2014; Markowski and Richardson 2014; Naylor et al. 2012; Naylor and Gilmore 2012a, b; Naylor and Gilmore 2014; Nowotarski et al. 2015). Tornado simulation has been limited to weaker ones until very recently when Orf et al. (2014) successfully simulated a long-lasting EF5 tornado at 30 m spatial resolution for the first time. Figure 2 shows a recently simulated EF5 tornado (24 May 2011 in Calumet-El Reno-Piedmont-Guthrie, OK) with detailed structures at 30 m spatial resolution by using an atmospheric storm-scale model (Orf et al. 2014). Volume-rendered vorticity magnitude (red to green) and trajectories (blue) are depicted. This modeled tornado indicates a strong cyclonic vortex accompanied by numerous supplying positive and negative vorticities. The inlet is volume-rendered cloud and rainwater showing a realistic looking of a tornado as would be seen in the field (https://ams.confex.com/ams/27SLS/webprogram/Paper255451.html, accessed on 05 Feb 2015). Obviously, tornado-induced wind loads and flow structures around buildings would be very complex and quite different from that of straight-line boundary-layer flows (Markowski et al. 2002).
Initialized with regional weather model analysis data, the Cloud Model 1 (CM1, Bryan and Fritsch 2002) is capable of simulating small-scale storms and hurricanes (e.g., Bryan and Morrison 2012; Kilroy and Smith 2015; Orf et al. 2012). Although very challenging, this capability of tornado simulation provides an opportunity to study realistic tornado-generated loading on building structures and then simulate and analyze the structural response and resistance to the wind and pressure impact. The atmospheric modeling of tornados is different from the statistical modeling by climatologists, physical vortex simulators or aerodynamic modeling by engineers. It is based on fluid dynamics with consideration of physical processes such as cloud microphysics, radiation, and turbulence. The CM1 model bears some similarity to most mesoscale weather models including the Weather Research and Forecasting (WRF) model (Skamarock et al. 2005). It is a numerical model governed by a three-dimensional, time-dependent, and non-hydrostatic equation set and is run on high-performance supercomputers. The model is designed for the purpose of simulating small-scale atmospheric phenomena such as thunderstorms and tornados at very high resolution (e.g., Bryan and Morrison 2012; Kilroy and Smith 2015; Orf et al. 2012; Bryan and Fritsch 2002; Markowski et al. 2014; Markowski and Richardson 2014; Naylor et al. 2012; Naylor and Gilmore 2012a, b, 2014; Nowotarski et al. 2015; Orf et al. 2014). A spatial resolution of as high as 30 m (or finer) plus sub-second time scale simulations depict detailed features within tornados, which have an average diameter of hundreds to thousands of meters. For the purpose of studying tornado impacts on building structures, this model is capable to provide unprecedented, very detailed three-dimensional wind, pressure, and vorticity fields as well as many other meteorological elements for engineering studies.
3 Tornado hazard from an engineering perspective
Tornados induce huge losses to human societies. Insurance data for tornado damages during 1949–2006 revealed 793 tornadic events that each caused more than $1 million in losses in the USA (Changnon 2009). The average annual loss of tornado catastrophes is $982 million. Tornado catastrophes and losses were found most frequent in Texas, Oklahoma, and Kansas, and moderately frequent in other Midwestern states.
These losses are primarily because of the building and infrastructure damages. For example, Wurman et al. (2007) evaluated the potential impacts of intense tornados crossing densely populated urban areas. A large and intense tornado with winds in excess of 76 m s−1 crossing through residential portions of Chicago, Illinois, would result in tragic consequences. It would impact 99 km2, destroy up to 239,000 single- and dual-family housing units occupied by up to 699,000 people, result in 4500–45,000 deaths, and cause substantial damage to over 400,000 other homes occupied by over 1,100,000 people. Tornados with winds exceeding 102 m s−1 over a broad area with high-rise office and apartment buildings would cause widespread permanent structural damages. Wurman et al. (2008) pointed out that the death toll for a given scenario was a function of several variables. The first one was the population in low-rise, single-family residential structures that face winds >76 m s−1; and the second variable was the population in high-rise structures exposed to winds ≥102 m s−1. Even with warnings of long lead times, a violent tornado in Chicago or any other major cities would certainly result in severe losses. The challenge of providing the denominators critical to calculating probability of death (POD = No. of fatalities/No. of survivors) in these various scenarios was daunting.
To mitigate the tornado-induced damages and losses from the engineering perspective, structural engineering researchers have been primarily focused on: (1) retrospective observational studies on the tornado-induced structural damages, (2) laboratory experiments and/or computer simulations on the tornado impact on building structures, and (3) development and improvement of the structural analysis and design methodology for tornado loadings.
3.1 Observational studies on tornado-induced structural damages
Currently, most of the tornado intensity studies focus on wind speed, which is very limited in direct observation and relies on aftermath estimation from damage assessments (Edwards 2013). For example, De Silva et al. (2006) observed the destruction caused by the 3 May 1999 Oklahoma City Tornado and indicated that the greatest opportunity to reduce monetary losses from tornados was for low-intensity tornados. Further, at damage ratings of F1 and higher, research should be directed toward life safety measures. Bienkiewicz (2008) visited ten sites in Missouri and Kansas after the 2003 Missouri–Kansas tornados and presented a description of the tornado events. The visiting report describes tornado hazards for residential homes, multi-family housing complex, college dormitory, houses under construction, commercial and industrial structures, historic structures, emergency facilities, transmission line, etc. Jordan (2007) conducted several case studies depicting damages from tornados, including the 2004 Utica and Joliet, Illinois tornado that killed 8 people and the 2006 Caruthersville, Missouri tornado. Fratinardo and Schroeder (2013) compared field damage observations on residential and commercial structures of the 15 March 2012 Dexter, Michigan Tornado, the 14 April 2012 Wichita, Kansas Tornado, and the 22 April 2011 Saint Louis, Missouri Tornado. It was indicated that, to assess tornado-damaged buildings, it requires knowledge of the building’s structural system and materials, an understanding of the construction, and the effect of wind forces. The results can serve as a guide in determining the extent of damage to an individual structure, although not always accurate. Extensive field examination of the structure is required to completely assess the damage and to estimate wind speeds.
To obtain deeper understanding of the observational studies on tornado-induced building damages, the most popularly used field damage observation methodology and four typical and representative observational case studies are described in detail, which demonstrate the four observational study methods in structural engineering field including field validation and documentation, development of design provisions, improvement of construction details, and comparison study of current and previous technologies.
3.1.1 Field investigation methodology
Lee et al. (2010) reported that the significant factors contributing to the extent of structural damages incurred from a windstorm included the quality and type of construction, the age, the presence or lack of a good maintenance program, design, construction and material defects, and the orientation of the building. It is critical that proper structural forensic investigation procedures be used to assess damage and collect accurate data. Further, the assessment of damage patterns from site-specific data and documentation, combined with weather data, allow for an accurate determination of the cause and extent of damage at a structure.
The original F-scale and the Enhanced EF-scale are the widely accepted measures for tornado-induced damage in the field. The F-scale was introduced as a means of estimating the intensity of tornados by relating appearance of damage to wind speeds (McDonald et al. 2009). The relationships in F-scale were never independently verified, which were developed only based on the extensive experience of Dr. Fujita. In addition, the F-scale did not consider the importance of construction quality and had limited damage indicators. The EF-scale is an improved F-scale, which defined 28 damage indicators and a number of degrees of damage for each indicator. The EF-scale was correlated with the F-scale by means of a regression equation (McDonald et al. 2009). In its present form, the EF-scale is applicable only to tornado intensity. Womble et al. (2009) examined two major components of the EF-scale: (1) damage levels and (2) ranges of tornado wind speeds related to these damage levels. Generally, the EF-scale describes properly the overall progression of damage with the increase in tornado wind speeds.
The American Society of Civil Engineers (ASCE) 11–99, Guideline for Structural Condition Assessment of Existing Buildings (Guideline) served as a guideline for the completion of a structural damage condition assessment of an existing structure in the field investigation procedures. The procedures and guidelines, however, are slanted toward single-structure assessments with the intention of determining structural adequacy with respect to increased loading, change in use, etc. Bracken and Roda (2007) proposed protocols that can be used when conducting multi-structure damage assessments in response to catastrophic events.
Being different from traditional damage surveys, Crandell and Kochkin (2005) conducted a scientific experiment for assessing the performance of a population of buildings following a natural disaster. The main feature in the scientific damage assessment methodology pertained to the manner by which a representative sample of buildings was selected for study. Using this feature, the statistical analyses were conducted in relation to wind loading parameters, building characteristics, and observed performance. Using mean and variance as well as confidence levels for the sampled building population, the robustness of cause–effect relationships and frequency of various types and levels of the damage can be characterized.
3.1.2 Case 1: Field validation of EF-scale determinations and the associated wind speeds of the 4 May 2007 Greensburg, Kansas tornado
Coulbourne (2008) conducted two site visits to the 4 May 2007 Greensburg, Kansas, tornado. The task was to assess the damages in Greensburg and to classify the tornado using the EF-scale. He described the detailed facts for the tornado hazard. Several supercell thunderstorms formed across parts of the Midwestern USA and spawned tornados in several states on 4 May 2007. The strong supercell developed southwest of Greensburg, Kansas that evening and resulted in 12 tornados. Formed in northwest Comanche County at about 9:00 pm LDT, one of these tornados moved northeastward through Kiowa County. This tornado reached Greensburg, Kansas, a small community of approximately 1,400 people at approximately 9:45 pm. The tornado passed through the town from the southern edge to the northwestern border. The tornado was rated EF5 with an estimated wind speed greater than 89 m s−1 and a damage path of 2.7 km across, and the funnel cloud itself was estimated to be 1.6 km in diameter. The tornado destroyed or severely damaged many of the buildings in Greensburg and caused ten fatalities. Coulbourne (2008) reported that, from the observed damage, there appears to be good agreement between the EF-scale determinations and the associated wind speeds. Because the wind speed changed rapidly in the storm, the mapping of the approximate wind speeds showed wide fluctuations for tornados. Wind designs based on the building code for hurricane-prone areas would provide protection for tornados classified EF3 and weaker. In tornado-prone areas, the high wind design needs to consider the windborne debris protection because the numerous high-speed missiles generated by a tornado event may induce more severe damages. However, the primary protection from tornado events should be the shelter, either the residence or community shelters.
3.1.3 Case 2: Development of tornado design and recovery provisions though studying the 27 April 2011 Tuscaloosa, Alabama tornado
The month of April, 2011 was the most active tornado month on record. On the single day of April 27, 164 tornados traveled across areas of Mississippi, Alabama, Georgia, and Tennessee (Clem and Hall 2012), including the devastating EF5 tornado striking Tuscaloosa, Alabama. This tornado caused significant personal injury and loss of life and destroyed facilities and residences completely by spanning 16.8 km2 representing 12 % of the total city area. Combined with another devastating tornado at Joplin, Missouri on 22 May 2011, the two tornados caused 223 fatalities and damaged more than 13,000 buildings in the two cities, which were two of the most violent and costly tornados on record (Prevatt et al. 2012a, b). For infrastructures, a total of 353 transmission structures were damaged and 108 transmission lines were out of service during the tornado outbreak in April 2011, including fifteen 500 kV transmission lines (Smith et al. 2012).
The Building Science Branch of the Federal Emergency Management Agency’s Federal Insurance and Mitigation Administration arranged a Mitigation Assessment Team (MAT) to assess the damage caused by the tornados in Alabama, Georgia, Mississippi, and Tennessee on 25–28 April 2011 (Walsh and Tezak 2012). The MAT provided suggestions to codes and standards by assessing residential and commercial buildings, infrastructure, critical facilities, hardened areas, tornado refuge areas, best available refuge areas, storm shelters, safe rooms, etc. The observations, conclusions, and recommendations on building performance presented in FEMA P-908 were highlighted. The MAT found that poor building performance resulted in occupant deaths and injuries. MAT developed Tornado Recovery Advisories for owners and designers of critical facilities in tornado-prone regions to facilitate improved performance of critical facilities and enhanced occupant protection.
By using the 2011 Tuscaloosa, Alabama tornado field observations as an example to systematically explain the concept, van de Lindt et al. (2013) proposed a dual objective-based design philosophy, which focused on both building damage and loss reduction in the low-to-moderate wind speed tornados and building occupant life safety in more damaging wind speed events (EF4 and EF5 tornados). The new philosophy was focused on separating tornado intensity levels. The performance of buildings at EF0 and EF1 wind speeds can be improved at the component level (i.e., connections), at the wind speed design for the EF2 and EF3 tornados can be improved at the system levels, such as, shear wall and load path that is direction in which each consecutive load will pass through connected members, and at EF4 and EF5 wind speed life safety can be provided using alternate means, such as safe rooms.
In addition, van de Lindt et al. (2012) analyzed a light-frame wooden house with an unconventional roof system based on the field observation data of the 2011 Tuscaloosa, Alabama tornado. Using a fragility methodology as a tool, the loss of the roof system was probabilistically investigated. The authors reported that the use of nail connections in a roof-to-wall connection resulted in a weaker link. In comparison, a hurricane clip and the fragility approach (van de Lindt et al. 2012) can be used as a supplement to the EF rating when unusual conditions in either the structure or the surroundings exist. Amini and van de Lindt (2014) applied the fragility methodology to provide insight for tornado design wind speeds using typical wood-frame construction methods. Five archetype structures were selected to represent the residential construction design space in the mid-south, Midwest, and central plains regions of the USA. Fragilities were developed for the main components along the vertical load path, roof sheathing, and roof-to-wall connection. These fragilities provided information regarding the critical components and controlling factors in overall system load path performance. It was concluded that an EF1 tornado can be resisted by conventionally designed light-frame wooden buildings (using nails and typical hardware) with light to moderate damage if hurricane clips and tighter nail schedules for roof sheathing are applied. However, an EF2 tornado would cause moderate to considerable damages; and an EF3 tornado would cause failure of the load path.
Graettinger et al. (2012) utilized geo-referenced data, as the supplement to the traditional field observations, for the analysis and fast dissemination of wood-frame construction (primarily residential houses) data collected in the aftermath of the Tuscaloosa, Alabama tornado. A data fusion method based on time and space synchronization was employed to geo-reference the field-collected data. Based on such fusion, the collected data were uploaded on a Geographical Information System (GIS) database and disseminated through a web portal to the research community and anyone with an internet connection. The web portal enables end-user functions to display and analyze the data. The GIS attribute of the collected data has been actively used to analyze the response of the house/infrastructure to the tornado and to develop the aforementioned dual objective-based design philosophy for structural engineering.
The City of Tuscaloosa, Alabama developed a strategic citywide plan for renewal and rebuilding in response to the devastating effects of the direct tornado hit on 27 April 2011 (Edward Back et al. 2012). The city engaged in an inclusive, multi-disciplinary planning process to address land use, housing, public facilities, infrastructure, and sustainability, etc.
3.1.4 Case 3: Improvement of construction details through investigating the 22 May 2011 Joplin, Missouri tornado
An EF5 tornado struck Joplin, MO on 22 May 2011. A team of building and construction professionals carried out the investigation and documentation of the performance of residential and critical buildings in Joplin (Prevatt et al. 2012a, b). The team reported that the Joplin tornado created an 11-km-long damage path, destroyed over 5000 buildings, and killed over 150 people. From this single tornado, it was nearly 25 % of the total economic loss caused by the 1400 tornados during spring 2011. Combined with the devastating Tuscaloosa, Alabama tornado in April 2011, the two tornados caused over $13 billion in economic losses, counting approximately 3 % of the combined annual gross domestic product (GDP) of the two states. The research team found that new ways of analysis and design are needed to improve the general level of engineering and construction for residential structures. The findings of the field study can help validate recent experimental and analytical results on tornado loads from laboratory. In a series of transects perpendicular to the tornado path, a gradual reduction in severity of damage from the center of the path toward the edges was found. Many failures were not only attributed to wind speeds that exceeded building code levels, but also to the lack of continuity of vertical and/or lateral load paths. Many load path failure examples were observed where good detailing in compliance with current building codes for tornado-prone regions would have reduced the damages and injuries.
Coulbourne and Miller (2012) specifically studied the two large schools in Joplin, which were significantly damaged by the Joplin tornado. Those schools together had over 2700 students, in which, the high school was built in 1968 and the middle school was built in 2009. The damage to the high school buildings was serious, especially in high bay areas, such as the gymnasium and the auditorium. The exterior walls in courtyard facing single-story classrooms were collapsed, and the exterior walls of the center section of the school were destroyed with wind-borne debris. However, the roof and concrete floors of the center section of the school were intact because of their sufficient strength. The middle school was also badly damaged in the high bay areas, i.e., the gymnasium and auditorium. The roofs were lifted or shifted from their connection to the walls, and the walls were collapsed. However, the one-story sections of the school performed very well with only some broken windows. None of the roof or exterior walls of those one-story sections were damaged. It was pointed out that the lack of wind load path continuity in both the older and newer schools might be one of the major reasons for the damages.
3.1.5 Case 4: Comparison study of current and previous building design/construction technology through tornados in Moore, Oklahoma
The tornado in Moore, Oklahoma, and the surrounding area on 20 May 2013 traveled 23 km, and its damage path was up to 1.7 km wide. In the tornado, 24 people were killed, over 200 others were injured, and many structures were damaged. Extensive field damage surveys were performed by Burgess et al. (2014). The surveys were aided by use of high-resolution aerial and satellite imagery. During the survey, the EF-scale was utilized and, in addition, the software package (Damage Assessment Toolkit) was used to facilitate the process. Survey results documented 4253 objects damaged by the tornado, 4222 of them with EF-scale damage indicators (DIs). Approximately 50 % of the total DIs were associated with EF0 ratings. Exclusive of EF0 damage, 38 % were associated with EF1, 24 % with EF2, 21 % with EF3, 17 % with EF4, and only 0.4 % associated with EF5.
Dao et al. (2014) analyzed building damage patterns for the 2013 Moore tornado and reported that the failure progression of residential structures within a tornado wind field depends on the virtual location and orientation of the house along the tornado path. Specifically, even if similar damage was observed, two different structures might experience different wind speeds if their relative distances from the centerline of the tornado path are different. Also, different damage levels were seen for structures at the same relative location to the centerline of the tornado. It could be caused by the difference in house orientation (garage door/windows on the windward versus leeward side) and/or windborne debris impact, which led to a change in internal wind pressure. Dao et al. (2014) recommended that the distance from the centerline of a tornado and the orientation of a building should be considered when making predictions of tornado wind speed from the observed residential structural damages.
In history, the city of Moore was struck by two similar devastating tornados with similar paths in 1999 and 2003, respectively. Fratinardo and Schroeder (2014) compared the observed residential damages caused by the 2013 Moore tornado and those documented during the 1999 and 2003 Moore tornados. The study showed that residential buildings could not withstand the “disastrous” damage during their respective tornado attacks. However, their damages could have been reduced by using some simple techniques during their construction, such as bolt anchors at wall-to-foundation joints, hurricane clips at wall-to-ceiling interfaces, and straps and hangers placed where applicable in lieu of face-nail fasteners at certain framing joints. A cost-effective and simple change to existing building code was recommended to enhance the structural integrity of the built environment. Similarly, Ramseyer et al. (2014) compared the 1999 and the 2013 Moore tornados to study the effects of different lateral load bracing systems and different roof slope. It was found that: (1) improved bracing system construction methods significantly narrowed the damage path of a tornado, increased the safety of occupants, and reduced financial loss due to property damage; (2) the intermittent bracing resulted in early wall failure at the stud-to-sill plate connection, while the continuous sheathing bracing resulted in a wall system that could sustain a higher ultimate load capacity; (3) anchorage failures with continuously sheathed homes, but these occurred after the walls had failed and tended to be caused by the prying action of the wall rotating or tipping as it fails; and (4) failure of single-family home with a low slope roof system was most likely to begin in the garage and homes with high pitch roofs were more likely to have a failure mechanism within which the roof system fails first;
In addition to studying residential houses, residential shelter performance in the aftermath of the devastating 2013 Moore tornado was specifically investigated by Standohar-Alfano et al. (2014). Seventy-five residential shelters along the tornado damage path were observed and documented. It was noted that the use of shared shelters saved many lives. All belowground shelters performed well, and no evidence of structural compromise or shelter breach from debris missiles was observed. However, occasionally it was observed that the debris blocked the opening of the garage slab residential shelters, and flooding was also found in a few cases.
3.2 Laboratory experiments and computer simulations of tornado impacts on building structures
3.2.1 Laboratory experiments
To analyze quantifiably tornado-induced aerodynamic loads on civil engineering structures, the laboratory tornado simulator was specifically designed to generate tornado-like vortices in the Iowa State University (Haan et al. 2008). A vortex can be generated by this simulator, which can travel along a ground plane to interact with structure models on the plane. Using a 1.83 m diameter fan and an annular duct suspended from an overhead crane that can travel along a 10.4 m ground plane, the rotation-forced downdraft is generated. Tornado vortex radii can be from 0.23 to 0.56 m and the maximum tangential velocities range from 6.9 to 14.5 m s−1. Both single-celled vortices and two-celled vortices with corresponding swirl ratios ranging from 0.08 to 1.14 can be produced. The wide range of tornado sizes together with model structures of 1:100 to 1:500 scale can be used extensively to examine the tornado-induced wind loads on terrestrial structures.
Using the tornado simulator, Haan et al. (2010) simulated tornado winds to quantify the aerodynamic loading on one-story gable-roof buildings. A 1:100 scale model of a building with dimensions of 9.1 × 9.1 × 6.6 m3 and gable-roof angle of 35° was used in the study. The result showed that the peak values of side force coefficients exceeded the building design standards (ASCE 7-05) by factors as high as 1.5, and the peak values of the external uplift force coefficients exceeded ASCE 7-05 provisions by factors between 1.8 and 3.2. It is evident that buildings constructed based on ASCE7-05 standards or related codes may be under-designed to withstand the wind loads induced by tornados.
Haan et al. (2014) summarized how the simulator’s results compared with the acquired damage data from the 2013 Moore, Oklahoma tornado. First, the radial distance from the center of the damage path was used as the factor to plot a structural damage function. Then this function was compared with the laboratory simulator-predicted values. The comparison indicated that the shape of the damage function curves was reasonably consistent, but the damages for most radial positions were overpredicted by the laboratory simulator. The shape of the curve could be improved with a more rational connection between predicted forces and the resulting damages.
3.2.2 Computational modeling for tornado impacts on structures
Tornados are violently rotating columns of air extending from thunderstorms to the ground. The evolution of the wake vortex and turbulent flow around building structures induced by tornadic winds is significantly different from those in conventional straight-line winds (Yang et al. 2011). The tornado vortex is a very complex, highly turbulent, three-dimensional vortex flow. In addition to strong rotational flows, there are areas of strong vertical motions as well as cyclonic and anticyclonic vortices. For example, a study for wind loads and surface pressure on small building models in swirling winds and straight-line winds was carried out by Bienkiewicz and Dudhia (1993). It was discovered that the wind loads acting on the tested models were significantly higher (3–5 times) in swirling winds compared to that in straight-line winds. It might not be correct when use the conventional straight-line wind tunnel to run with maximum tornado wind speed for estimating tornado-induced wind loads on built structures.
Unfortunately, there is still no widely accepted tornado computational modeling technique for structural engineering applications. The aforementioned CFD and CM1 models for meteorology have not yet been applied in structural engineering field.
3.3 Development and improvement of structural design code/provisions for tornados
Tornado was classified as one of the important causes of death and injuries among natural disasters in the USA (Whalen et al. 2004). As aforementioned, the damages caused by an EF3 or weaker tornado can be controlled by the hurricane wind design provisions in the existing building codes and by improved construction detailing. Based on the post-event forensic investigation data across several wind regions of the USA, VanDerostyne et al. (2013) concluded that the currently used building codes do not adequately recognize the inherit risks from tornados. While it is unfeasible for buildings to be designed to endure all levels of tornados, it is not unfeasible for the minimum design requirements for better performance during strong wind events, particularly those in the “Tornado Alley”. Specific recommendations (VanDerostyne et al. 2013) for all building types include better detailing at the roof-to-wall connections as well as better quality control during the course of construction. For wood-framed buildings, improvement of the minimum fastening requirements, requirement for better anchorage to the foundation, and requirement for plywood sheathing around the entire perimeter were also recommended. For concrete masonry unit (CMU) construction, it was recommended to improve minimum reinforcing and grouting for all exterior walls.
The life safety during EF4 or EF5 tornados has to be protected by shelters or safe rooms. Because tornado-related mortality and injury rates are higher when tornado shelters are not available, shelters have been used as one of the most effective techniques in tornado disaster mitigation efforts. Provisions in FEMA P-361, Design and Construction Guidance of Community Safe Rooms (FEMA 2008a) and the International Council Code (ICC) 500, the Standard for the Design and Construction of Storm Shelters (ICC/NSSA 2008) provide designers and engineers with suggestions of design considerations of shelter structures. These design and structural analysis provisions of the shelters were based on simplified and conservative analytical methods and on the results of numerous impact tests on shelter components at the Texas Technology University (Budek et al. 2006). Zhou et al. (2014) proposed a new tornado-safe room design using Carbon-Fiber reinforced hybrid Matrix Composite (CHMC), or CarbonFlex, to withstand tornado-borne debris impacts. Test results showed that the CarbonFlex wall composite had excellent impact resistance in comparison with the conventional residential construction.
In addition to the shelters or safe rooms, tornado-safe design is required for nuclear facilities. The American Nuclear Society in 1983 (ANS 1983) published a tornado design standard, providing figures for tornado wind speeds, pressure drop, and tornado missile wind speeds. Expanded from the ANS 1983, the ANS 2.3-2011 Standard (ANS 2.3-2011) provides design guidance for tornado missile as a function of wind speed and type of wind loading.
4 Bridging meteorological and engineering studies for a sustainable built environment
4.1 Demands for understanding of tornados and more accurate simulations
The limitation of existing knowledge of detailed tornado structure and tornado wind fields resulted in the traditional wind damage assessment, which depended largely on horizontal straight-line winds although attempts have been made in studying rotational wind impacts (e.g., Senguipta et al. 2008; Wurman et al. 2012b). The impacts from rotational winds, pressure drops, and vertical suction winds have not been considered in any existing building codes. Moreover, the acquisition and evolution of the Storm Data (McDonald 2001; Verbout et al. 2006) implied its limitations in data quality because of factors such as the complexity of the wind speed–damage relationship, unavailability or heterogeneity of damage indicators, distribution and change of human population, and observation and communication technologies (Doswell et al. 2009).
Most of the engineering-oriented tornado simulation or modeling focused on a wind tunnel (or chamber) modeling and simulated the wind impacts on microscale objects that are put in the simulators. Refan (2014) provides a detailed summary about several existing vortex simulators and aerodynamic modeling of wind impacts. These types of modeling are limited to the controlled conditions that are far too simple than the real atmosphere.
Atmospheric simulation, on the other hand, considers all fluid dynamics and physics and is able to simulate and predict the weather phenomena. Tornados are at microscale (in meteorology) and lack of direct observations, which have hindered the advancement in atmospheric simulations. However, recently, atmospheric models, especially the Cloud Model (Bryan and Fritsch 2002), showed the great potential to simulate the tornados (Naylor et al. 2012; Naylor and Gilmore 2014). The atmospheric modeling has revealed the very complex flow patterns that exist in tornados (Orf et al. 2012). Figure 2 shows an example of atmospheric model-simulated EF5 tornado wind patterns. These suggest that tornado impacts on building structures are far more complex than an in-house tornado simulator could handle.
Atmospheric models are able to simulate the actual atmosphere and have reached high accuracy in less-than-three-day weather prediction. These models generate all weather-related physical parameters/variables, including three-dimensional winds, pressure, temperature, humidity, rain, hail, etc. Currently, many studies are focusing on the study of tornado genesis, which ultimately will lead to more accurate understanding of the factors that cause the tornado to occur and consequently allow us to more accurately simulate and predict tornados.
4.2 Demands in structural analysis for high-resolution tornadic wind fields
With the consideration of buildings as surface-mounted obstacles, numerous numerical studies, such as the finite element method (FEM), have been used to investigate building structures under wind loads and flow structures around the surface-mounted obstacles (Hunt et al. 1978; Martinuzzi and Tropea 1993; Natarajan and Chyu 1994; Yakhot et al. 2006 and Gu et al. 2010). While those numerical efforts have revealed a great deal about the wind loads and flow structures around buildings, all of the previous studies were conducted with the assumption of straight-line boundary-layer flows and steady-state conditions. However, tornados are strong vortices with a significant tangential component, radial inflow/outflow, and vertical updraft/downdraft. It is desirable that more comprehensive tornadic wind fields be generated. In addition to the traditional parameters including wind speed and pressure, new tornadic parameters such as rotational wind speed, central pressure change and gradient, and vertical suction wind loads need to be derived from tornado simulations.
The impacts from rotational winds, pressure drop, and vertical suction winds have not been considered in any existing building codes. With the up-to-date tornado observation data and recent efforts in field experiments, it is desirable to develop a new tornadic impact map and design parameters for building design codes. The map and parameters are useful for designing independent or in-house tornado shelters and important building structures such as nuclear facilities and hospitals. In addition, with projected climate change in the future, it becomes possible to consider future potential climate change-induced extreme weather conditions and wind impact.
4.3 Demands in systematic structural analysis and design provisions for tornado impacts
As described in this paper, building damages under an EF3 or weaker tornado impact may be controlled by the hurricane-related design provisions. And life safety under EF4 and EF5 tornado impact needs to be protected by shelters and safe rooms. For engineered buildings, even important building structures such as hospital, there are still no systematic and effective analysis methodology and design provisions in existence. Such a methodology and provision will depend on a more comprehensive understanding of tornado characteristics and better modeling techniques of tornado loading.
5 Conclusion
This paper reviewed the tornado hazard from both meteorological and engineering perspectives. On the basis of the development in the two individual disciplines, the paper intended to identify future demands in advancement of each discipline, and more importantly to identify the potential areas to bridge and link the two so that the tornado impacts on civil structures can be better understood.
Meteorologically, the historical observation data have the limitation of providing detailed characteristics in terms of tornado wind speed, coverage area, central pressure, and vertical suction strength, thus leading to vague estimates of stress loads on buildings. The laboratory simulations of tornados are only tornado-like vortices and are at spatial scales too small and strengths too weak compared to real tornados. Atmospheric simulations of tornados have gained advancement, but still lacking the verification with real observations, making the simulation of a real tornado not able to be realized yet.
In the viewpoint of engineering, the tornado damage on building structure relies on aftermath surveys. However, such result is affected by the inconsistency of building codes and consequently the inconsistent building structures that are under investigation. Field methods have been used in various tornado surveys, and yet detailed load analysis of each building structure is not available. The laboratory simulations of tornado impacts on model building blocks are also at scales too small in comparison to real buildings and tornados. Numerical aerodynamic simulation of the loads exerted on buildings is also limited due to the unrealistic tornado wind and pressure fields applied.
In dealing with the above limitations, demands in both disciplines are discussed. In meteorological perspectives, there is a continuing need in close observation of tornados. While the fast development in atmospheric simulation of tornados foresees the potential capability of successful simulation of real tornados, verification against real observations is still a challenge. Remote-sensing technologies in tornado observation is deemed a practical way in future observations.
From the engineering perspectives, more detailed structural analysis in the realistic tornado wind field is needed in order for the potential capability of building code development. Both laboratory experiments and numerical modeling are viable approaches in conducting building load analysis and simulation, based on which the reliable building codes can be designed.
References
Abbey R, Fujita T (1975) Use of tornado path lengths and gradations of damage to assess tornado intensity probabilities. Bull Am Meteorol Soc Am Meteorological Soc 45 Beacon St, Boston, MA 02108–3693:753
Akiya Y, Saito A, Sakanoi T, Hozumi Y, Yamazaki A, Otsuka Y, Nishioka M, Tsugawa T (2014) First spaceborne observation of the entire concentric airglow structure caused by tropospheric disturbance. Geophys Res Lett 41(19):6943–6948
Amini MO, van de Lindt JW (2014) Quantitative insight into rational tornado design wind speeds for residential wood-frame structures using fragility approach. J Struct Eng 140(7):04014033
Ashley WS (2007) Spatial and temporal analysis of tornado fatalities in the United States: 1880–2005. Weather Forecast 22(6):1214–1228
Ashley WS, Strader S, Rosencrants T, Krmenec AJ (2014) Spatiotemporal changes in tornado hazard exposure: the case of the expanding bull’s-eye effect in Chicago, Illinois. Weather Clim Soc 6(2):175–193
Atkins NT, Butler KM, Flynn KR, Wakimoto RM (2014) An integrated damage, visual, and radar analysis of the 2013 Moore, Oklahoma, Ef5 Tornado. Bull Am Meteorol Soc 95(10):1549–1561
Baker GL (1981) Boundary layers in laminar vortex flows. Ph.D Thesis, Purdue University
Bienkiewicz B (2008) Lessons learned from structural damage investigations—a case study of 2003 Missouri-Kansas tornadoes. In: Structures congress-structures congress, pp 24–26
Bienkiewicz B, and Dudhia P (1993) Physical modeling of tornado-like flow and tornado effects on building loading. In: Proceedings 7th US national conference on wind engineering, pp 95–106
Bluestein HB, Houser JB, French MM, Snyder JC, Emmitt GD, PopStefanija I, Baldi C, Bluth RT (2014) Observations of the boundary layer near tornadoes and in supercells using a mobile, collocated, pulsed doppler lidar and radar. J Atmos Ocean Technol 31(2):302–325
Bracken WC, and Roda TA (2007). Establishing protocols for catastrophe damage assessments of multiple buildings. Forensic Engineering (2007) ASCE pp 1–10
Brooks HE (2004) On the relationship of tornado path length and width to intensity. Weather Forecast 19(2):310–319
Brooks HE, Doswell CA, Kay MP (2003) Climatological estimates of local daily tornado probability for the United States. Weather Forecast 18(4):626–640
Bryan GH, Fritsch JM (2002) A benchmark simulation for moist nonhydrostatic numerical models. Mon Weather Rev 130(12):2917–2928
Bryan GH, Morrison H (2012) Sensitivity of a simulated squall line to horizontal resolution and parameterization of microphysics. Mon Weather Rev 140(1):202–225
Budek A, Zain M, Qiao L, Phelan RS (2006) Validation of finite-element analyses for storm shelters. J Archit Eng 12(2):64–71
Burgess D, Ortega K, Stumpf G, Garfield G, Karstens C, Meyer T, Smith B, Speheger D, Ladue J, Smith R, Marshall T (2014) 20 May 2013 Moore, Oklahoma, Tornado: damage survey and analysis. Weather Forecast 29(5):1229–1237
Camp J, Rothfusz L, Anderson A, Speheger D, Ortega K, Smith B (2014) Assessing the Moore, Oklahoma (2013) tornado using the National Weather Service Damage Assessment Toolkit. In: Proceedings of the 94th American Meteorological Society Annual Meeting. Special symposium on severe local storms: the current state of the science and understanding impacts
Changnon SA (2009) Tornado losses in the United States. Nat Hazards Rev 10(4):145–150
Church C, Snow J, Agee E (1977) Tornado vortex simulation at Purdue University. Bull Am Meteorol Soc 58(9):900–908
Clem CL, and Hall B (2012) Systematic plan for re-constructing the TVA transmission system, April 27, 2011. In: Electrical transmission and substation structures 2012@ solutions to building the grid of tomorrow. ASCE, pp 1–13
Coulbourne WL (2008) Wind speed analysis of Greensburg, KS tornado. In: Structures congress-structures congress, pp 24–26
Coulbourne WL, and Miller J (2012) Performance of school buildings in the Joplin, MO, Tornado. In: Structures congress 2012, ASCE, pp 989–998
Crandell JH, and Kochkin V (2005). Scientific damage assessment methodology and practical applications. In: ASCE/SEI structures congress, forensics symposium, New York
Dao TN, Graettinger AJ, Alfano C, Gupta R, Haan FL, Prevatt D, Richardson J, Kashani AG (2014) Failure progression analysis of observed residential structural damage within a Tornado Wind Field. Proc Struct Congress 2014:1448–1459
De Silva DG, Kruse JB, Wang Y (2006) Catastrophe-induced destruction and reconstruction 1. Nat Hazards Rev 7(1):19–25
Doswell CA III, Brooks HE, Dotzek N (2009) On the implementation of the enhanced fujita scale in the USA. Atmos Res 93(1–3):554–563
Dotzek N, Kurgansky MV, Grieser J, Feuerstein B, Nevir P (2005) Observational evidence for exponential tornado intensity distributions over specific kinetic energy. Geophys Res Lett 32(24):L24813
Edward Back W, Fridley K, Morgan R (2012) Incorporation of recovery planning into a student design course following a Tornado. Leadersh Manag Eng 12(3):169–178
Edwards JL (2013) Post-disaster climatology for hurricanes and tornadoes in the United States: 2000–2009. Kent State University, Kent
Edwards R, LaDue JG, Ferree JT, Scharfenberg K, Maier C, Coulbourne WL (2013) Tornado intensity estimation: past, present, and future. Bull Am Meteorol Soc 94(5):641–653
Elsner JB, Widen HM (2014) Predicting Spring Tornado Activity in the Central Great Plains by 1 March. Mon Weather Rev 142(1):259–267
Elsner J, Jagger T, Widen H, Chavas D (2014) Daily Tornado frequency distributions in the United States. Environ Res Lett 9(2):024018
Feuerstein B, Dotzek N, Grieser J (2005) Assessing a Tornado climatology from global tornado intensity distributions. J Clim 18(4):585–596
Fratinardo VF, Schroeder SA (2013) Accuracy of EF ratings following a tornado event: an engineer’s perspective. In: Structures congress. American Society of Civil Engineers (ASCE), Pittsburgh, PA
Fratinardo VF, Schroeder SA (2014) Lessons learned: how changes in code, construction and preparedness affected Tornado damage in Moore, Oklahoma, 1999 to 2013. In: Structures congress, pp 1383–1391 (ASCE)
Graettinger AJ, Grau D, Van De Lindt J, and Prevatt DO (2012) GIS for the geo-referenced analysis and rapid dissemination of forensic evidence collected in the aftermath of the Tuscaloosa tornado. In: Construction research congress, pp 2170–2179
Gravelle CM, Mecikalski JR, Line WE, Bedka KM, Petersen RA, Sieglaff JM, Stano GT, Goodman SJ (2016) Demonstration of a GOES-R Satellite Convective toolkit to “Bridge the Gap” between severe weather watches and warnings: an example from the 20 may 2013 Moore, OK Tornado Outbreak. Bull Am Meteorol Soc 97:69–84. doi:10.1175/BAMS-D-14-00054.1
Grazulis TP (1993). A 110-year perspective of significant tornadoes. In: The tornado: its structure, dynamics, prediction, and hazards, pp 467–474
Grazulis TP, Schaefer JT, Abbey RF (1993) Advances in tornado climatology, hazards, and risk assessment since tornado symposium II. The tornado: its structure, dynamics, prediction, and hazards, pp 409–426
Gu M, Huang P, Tao L, Zhou X, Fan Z (2010) Experimental study on wind loading on a complicated group-tower. J Fluids Struct 26(7–8):1142–1154
Haan FL Jr, Sarkar PP, Gallus WA (2008) Design, construction and performance of a large tornado simulator for wind engineering applications. Eng Struct 30(4):1146–1159
Haan FL Jr, Balaramudu VK, Sarkar PP (2010) Tornado-induced wind loads on a low-rise building. J Struct Eng ASCE 136(1):106–116
Haan FL, Sarkar PP, Prevatt D, Roueche D, Graettinger A, Dao TN, and Crawford PS (2014) Using tornado damage surveys to improve laboratory tornado simulations. In: Structures congress 2014, ASCE, pp 1472–1483
Hamada A, El Damatty AA, Hangan H, Shehata AY (2010) Finite element modelling of transmission line structures under tornado wind loading. Wind Struct 13(5):451–469
Hangan H, Kim J (2008) Swirl ratio effects on tornado vortices in relation to the fujita scale. Wind Struct 11(4):291–302
Hunt J, Abell C, Peterka J, Woo H (1978) Kinematical studies of the flows around free or surface-mounted obstacles; applying topology to flow visualization. J Fluid Mech 86(01):179–200
Jordan JW (2007). Tornado damage assessment for structural engineers. Forensic Engineering, ASCE, pp 1–17
Karstens CD, Samaras TM, Lee BD, Gallus WA Jr, Finley CA (2010) Near-ground pressure and wind measurements in tornadoes*. Mon Weather Rev 138(7):2570–2588
Kelly D, Schaefer J, McNulty R, Doswell C III, Abbey R Jr (1978) An augmented tornado climatology. Mon Weather Rev 106(8):1172–1183
Kilroy G, Smith RK (2015) Tropical cyclone convection: the effects of a vortex boundary-layer wind profile on deep convection. Q J R Meteorol Soc 141(688):714–726
Krehbiel PR, Thomas RJ, Rison W, Hamlin T, Harlin J, Davis M (2000) GPS-based mapping system reveals lightning inside storms. EOS Trans Am Geophys Union 81(3):21–25
Lee KS, Caffey JM, and Killian DM (2010) Structural evaluation procedures and case studies of damage related to wind storms, tornadoes, and hurricanes. In: Forensic Engineering 2009@ sPathology of the Built Environment, ASCE, pp 749–758
Malamud BD, Turcotte DL (2012) Statistics of severe tornadoes and severe tornado outbreaks. Atmos Chem Phys 12(18):8459–8473
Markowski PM, Richardson YP (2014) The influence of environmental low-level shear and cold pools on tornadogenesis: insights from idealized simulations. J Atmos Sci 71(1):243–275
Markowski PM, Straka JM, Rasmussen EN (2002) Direct surface thermodynamic observations within the rear-flank downdrafts of nontornadic and tornadic supercells. Mon Weather Rev 130(7):1692–1721
Markowski P, Richardson Y, Bryan G (2014) The origins of vortex sheets in a simulated supercell thunderstorm. Mon Weather Rev 142(11):3944–3954
Martinuzzi R, Tropea C (1993) The flow around surface-mounted, prismatic obstacles placed in a fully developed channel flow (data bank contribution). J Fluids Eng 115(1):85–92
McDonald JR (2001) T. Theodore Fujita: his contribution to tornado knowledge through damage documentation and the Fujita scale. Bull Am Meteorol Soc 82(1):63–72
McDonald JR, Mehta KC, Smith DA, and Womble JA (2009) The enhanced fujita scale: development and implementation. In: ASCE 5th congress on forensic engineering, Washington, DC, Nov 10–15 (submitted)
Mishra A, James D, Letchford C (2008a) Physical simulation of a single-celled tornado-like vortex, part B: wind loading on a cubical model. J Wind Eng Ind Aerodyn 96(8):1258–1273
Mishra AR, James DL, Letchford CW (2008b) Physical simulation of a single-celled tornado-like vortex, part A: flow field characterization. J Wind Eng Ind Aerodyn 96(8–9):1243–1257
Natarajan D (2011) Numerical simulation of tornado-like vortices. PhD Thesis, the University of Western Ontario
Natarajan V, Chyu M (1994) Effect of flow angle-of-attack on the local heat/mass transfer from a wall-mounted cube. J Heat Transfer 116(3):552–560
Naylor J, Gilmore MS (2012a) Convective initiation in an idealized cloud model using an updraft nudging technique. Mon Weather Rev 140(11):3699–3705
Naylor J, Gilmore MS (2012b) Environmental Factors Influential to the Duration and Intensity of Tornadoes in Simulated Supercells. Geophys Res Lett 39:L17802
Naylor J, Gilmore MS (2014) Vorticity evolution leading to tornadogenesis and tornadogenesis failure in simulated supercells. J Atmos Sci 71(3):1201–1217
Naylor J, Gilmore MS, Thompson RL, Edwards R, Wilhelmson RB (2012) Comparison of objective supercell identification techniques using an idealized cloud model. Mon Weather Rev 140(7):2090–2102
Nowotarski CJ, Markowski PM, Richardson YP, Bryan GH (2015) Supercell low-level mesocyclones in simulations with a sheared convective boundary layer. Mon Weather Rev 143(1):272–297
Orf L, Kantor E, Savory E (2012) Simulation of a downburst-producing thunderstorm using a very high-resolution three-dimensional cloud model. J Wind Eng Ind Aerodyn 104:547–557
Orf L, Wilhelmson RB, Wicker LJ, Lee BD and Finley CA (2014) Genesis and maintenance of a long-track EF5 tornado embedded within a simulated supercell. In: 27th Conference on severe local storms. American Meteorological Society, Nov 2–7, 2014, Madison, WI, presentation 3B.3: https://ams.confex.com/ams/27SLS/webprogram/Paper255451.html
Otsuka T, Torii Y, and Ito T (2014) Anomaly detection algorithm for localized abnormal weather using low-cost wireless sensor nodes. In: 2014 IEEE 7th international conference on service-oriented computing and applications (SOCA). IEEE, pp 304–308
Paul BK, Stimers M (2014) Spatial analyses of the 2011 Joplin tornado mortality: deaths by interpolated damage zones and location of victims. Weather Clim Soc 6(2):161–174
Phillips WD, and Sankar R (2013) Improved transient weather reporting using people centric sensing. In: 2013 IEEE consumer communications and networking conference (CCNC). IEEE, pp 920–925
Prevatt DO, Roueche DB, van de Lindt, JW, Pei S, Dao T, Coulbourne W, Graettinger AJ, Gupta R, and Grau D (2012a) Building damage observations and EF classifications from the Tuscaloosa, AL and Joplin, MO tornadoes. In: Structures congress 2012, ASCE Reston, VA, pp 999–1010
Prevatt DO, van de Lindt, JW, Back EW, Graettinger AJ, Pei S, Coulbourne W, Gupta R, James D, Agdas D (2012b) Making the case for improved structural design: tornado outbreaks of 2011. Leadersh Manag Eng 192:254–270
Potter S (2007) Fine-tuning Fujita: after 35 years, a new scale for rating tornadoes takes effect. Weatherwise 60(2):64–71
Ramsdell JV, Andrews G (1986) Tornado Climatology of the Contiguous United States (No. NUREG/CR-4461; PNL-5697). Pacific Northwest Lab., Richland, WA (USA)
Ramsdell JV, Rishel JP, Buslik AJ (2007) Tornado climatology of the contiguous United States. Nuclear Regulatory Commission Rep. NUREG/CR-4461, Rev. 2, p 246. http://www.nrc.gov/docs/ML0708/ML070810400.pdf. Accessed 25 May 2016
Ramseyer C, Floyd R, Holliday L, Roswurm S (2014) Influence of lateral load bracing systems on damage and survivability of residential structures impacted by the Moore, Oklahoma, tornado of May 20, 2013. Proc Struct Congress 2014:1484–1507
Refan M (2014) Physical simulation of tornado-like vortices. Ph.D Dissertation, The University of Western Ontario. Electronic Thesis and Dissertation Repository. Paper 1923
Refan M, Hangan H, Wurman J (2014) Reproducing tornadoes in laboratory using proper scaling. J Wind Eng Ind Aerodyn 135:136–148
Reinhold TA, Ellingwood B (1982) Tornado damage risk assessment (No. NUREG/CR-2944; BNL-NUREG-51586). National Bureau of Standards, Washington, DC (USA). Center for Building Technology
Senguipta A, Haan FL, Sarkar PP, Balaramudu V (2008) Transient loads on buildings in microburst and tornado winds. J Wind Eng Ind Aerodyn 96(10–11):2173–2187
Simmons KM, Sutter D, Pielke R (2013) Normalized tornado damage in the United States: 1950-2011. Environ Hazards Hum Policy Dimens 12(2):132–147
Skamarock WC, Klemp JB, Dudhia J, Gill DO, Barker DM, Wang W, Powers JG (2005) A description of the advanced research WRF version 2. NCAR/TN–468+STR, NCAR TECHNICAL NOTE
Smith TL, Perotin M, and Walsh E (2012) Enhancing tornado performance of critical facilities: findings and recommendations of FEMA’s mitigation assessment team. In: Structures congress 2012, ASCE, pp 977–988
Snyder JC, Bluestein HB (2014) Some considerations for the use of high-resolution mobile radar data in tornado intensity determination. Weather Forecast 29(4):799–827
Standohar-Alfano CD, van de Lindt John W (2014) Empirically based probabilistic tornado hazard analysis of the United States using 1973–2011 data. Nat Hazards Rev 16(1):04014013
Standohar-Alfano CD, Freyne S, Graettinger AJ, Floyd RW, Dao TN (2014) Performance of residential shelters during the May 20, 2013, Tornado in Moore, Oklahoma. J Perform Constr Facil. doi:10.1061/(ASCE)CF.1943-5509.0000636
Strader SM, Ashley W, Irizarry A, Hall S (2015) A climatology of tornado intensity assessments. Meteorol Appl 22(3):513–524
Tippett MK, Sobel AH, Camargo SJ (2012) Association of US tornado occurrence with monthly environmental parameters. Geophys Res Lett 39:L02801
van de Lindt J, Amini MO, Standohar-Alfano C, Dao T (2012) Systematic study of the failure of a light-frame wood roof in a tornado. Buildings 2(4):519–533
van de Lindt JW, Pei S, Dao T, Graettinger A, Prevatt DO, Gupta R, Coulbourne W (2013) Dual-objective-based tornado design philosophy. J Struct Eng ASCE 139(2):251–263
VanDerostyne DA, Hallet SK, and Nichols JA (2013) Post-event forensic investigation of damaged structures from strong wind events. In: Forensic Engineering 2012@ sGateway to a Safer Tomorrow, ASCE, pp 841–850
Verbout SM, Brooks HE, Leslie LM, Schultz DM (2006) Evolution of the US tornado database: 1954–2003. Weather Forecast 21(1):86–93
Walsh E, Tezak S (2012) Findings and Recommendations of FEMA’s Mitigation Assessment Team Investigations of the Spring 2011 Tornado Outbreaks. Forensic Eng 2012:821–830
Whalen TM, Gopal S, Abraham DM (2004) Cost-benefit model for the construction of tornado shelters. J Constr Eng Manag 130(6):772–779
Womble JA, Smith DA, Mehta KC, McDonald JR (2009) The enhanced Fujita Scale: for use beyond tornadoes. Forensic Eng 2009:699–708. doi:10.1061/41082(362)71
Wood VT, Brown RA (2011) Simulated tornadic vortex signatures of tornado-like vortices having one- and two-celled structures. J Appl Meteorol Climatol 50(11):2338–2342
Wurman J, Kosiba K (2013) Finescale radar observations of tornado and mesocyclone structures. Weather Forecast 28(5):1157–1174
Wurman J, Robinson P, Alexander C, Richardson Y (2007) Low-level winds in tornadoes and potential catastrophic tornado impacts in urban areas. Bull Am Meteorol Soc 88(1):31–46
Wurman J, Alexander C, Robinson P, Richardson Y (2008) Low-level winds in tornadoes and potential catastrophic tornado impacts in urban areas reply. Bull Am Meteorol Soc 89(10):1580–1581
Wurman J, Dowell D, Richardson Y, Markowski P, Rasmussen E, Burgess D, Wicker L, Bluestein HB (2012a) The second verification of the origins of rotation in tornadoes experiment: VORTEX2. Bull Am Meteorol Soc 93(8):1147–1170
Wurman J, Dowell D, Richardson Y, Markowski P, Rasmussen E, Burgess D, Wicker L, Bluestein HB (2012b) The second verification of the origins of rotation in tornadoes experiment: VORTEX2. Bull Am Meteorol Soc 93(8):1147–1170
Wurman J, Kosiba K, Robinson P, Marshall T (2014) The role of multiple-vortex tornado structure in causing storm researcher fatalities. Bull Am Meteorol Soc 95(1):31–45
Yakhot A, Anor T, Liu H, Nikitin N (2006) Direct numerical simulation of turbulent flow around a wall-mounted cube: spatio-temporal evolution of large-scale vortices. J Fluid Mech 566:1–9
Yang Z, Sarkar P, Hu H (2011) An experimental study of a high-rise building model in tornado-like winds. J Fluids Struct 27(4):471–486
Zhou H, Dhiradhamvit K, Attard TL (2014) Tornado-borne debris impact performance of an innovative storm safe room system protected by a carbon fiber reinforced hybrid polymeric-matrix composite. Eng Struct 59:308–319
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Huang, Z., Fan, X., Cai, L. et al. Tornado hazard for structural engineering. Nat Hazards 83, 1821–1842 (2016). https://doi.org/10.1007/s11069-016-2392-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11069-016-2392-z