Historical Tornado Research Articles

Using Historical Data to Dynamically Route Post-Disaster Assessment Unmanned Aerial Vehicles in the Context of Responding to Tornadoes

Responding to tornado disasters resides at a unique intersection of search and rescue operations: it has attributes of wilderness and maritime search and rescue operations and search and rescue operations in the aftermath of earthquakes and hurricanes. This paper presents a method of attempting to leverage historical data to more efficiently identify the extent of the area damaged by a tornado. To assist in building and understanding the historical data, we also develop a method to generate tornado areas that react similarly to the limited historical data set. The paper successfully demonstrates the method of creating artificial tornado instances that can be used as a testing sandbox for the further development of tools when responding to tornado-type disasters. These artificial instances perform similarly in some important metrics to the historical database of tornado instances that we produced. This paper also shows that the use of historical tornado trends has an impact on the response method outlined in this article, typically reducing the standard deviation of the time it takes to fully identify the extent of the damage.

Digitizing and Geocoding Historical Records to Improve an Understanding of Tornado Climatology

Tornadoes have the potential to cause catastrophic destruction and mass casualties. What influence a changing climate has on tornado behavior remains a largely unanswered question. Much of the current knowledge of tornado climatology is linked to historical tornado reports that date back to 1950 and are maintained by the National Weather Service and the Storm Prediction Center. Here, we digitize and geocode significant and killer tornado reports over the period from 1880 to 1989 as a means of improving and expanding the data readily available to researchers. In total, 12,191 tornado reports were included to create a spatial data set that includes tornado characteristics consistent with the severe report database. Existing as both a straight-line track and an initial point model, the data set is broad enough to provide statistical analysis on tornado characteristics, geospatial analysis at the grid and county level, and point-based analysis at fine spatial scales. This flexibility in the data set allows for the application of methodologies across a range of disciplines that can improve studies related to tornado climatology and curriculum rooted in quantitative and geospatial analysis.

The Hunt for Missing Tornadoes: Using Satellite Imagery to Detect and Document Historical Tornado Damage in Canadian Forests

Abstract Historical tornado events from 1982 to 2020 were documented within Canada’s forested regions using high-resolution satellite imagery. Tornado forest disturbances were identified using a three-step process: 1) detecting, 2) assessing, and 3) dating each event. A grid of 120 km × 120 km boxes was created covering Canada (excluding the extreme north). Of the 484 boxes, 367 were manually searched. Once a long, narrow region of tree damage was detected, it was first cross-referenced with known tornado databases to ensure it was a unique event. Once events were classified as either tornadic or downburst, the coordinates of the start, worst damage, and end locations were documented, as well as the direction of motion, damage indicators, degree of damage, estimated maximum wind speed, and F/EF-scale rating. In total, 231 previously unknown tornadoes were identified. In Ontario, 103 events were discovered, followed by 98 in Quebec, 9 in Manitoba, 6 in Saskatchewan, 9 in Alberta, 5 in British Columbia, and 1 in New Brunswick. The largest number of discovered tornadoes occurred in 2015, and the largest number of strong F2 tornadoes occurred in 2005. Most of the discovered tornadoes occurred in July for both F/EF1 and F/EF2 ratings. Most tornado tracks had widths between 200 and 400 m, and more than 50% of the tornadoes had a pathlength of less than 10 km. Of all the events that were discovered, 125 events could be fully dated, 19 were dated only by month, 41 were dated only by year, and 46 remained undated.

Location of disaster assessment UAVs using historical tornado data

Preparing and responding to disasters is a complicated task. One must balance coverage of SAR resources versus preparation cost. This article presents a method and solution to prepositioning UAV damage assessment and search teams in Oklahoma using historical tornado data. The approach is based on set covering and multi-station vehicle routing models. It also presents a method to robustify the solution in the event a UAV team cannot be activated to respond to the disaster. This can simulate a team unable to respond. Results show 70% more stations and teams being required when chance of a depot failure goes from 0 to 5% and 90% more stations required when 0–10%. We find that when trying to use a solution that does not account for depot failure, the system of UAVs cannot meet search completion targets in 3–4% of cases. These results demonstrate accounting for the chance of teams not being able to respond to domestic disasters is important and failing to do so means an increased chance of not being able to respond adequately to disasters and incorporating the chance of station failure has a profound impact on the number of stations needed.

Assessment of NWS County Warning Area Tornado Risk, Exposure, and Vulnerability

AbstractThis study investigates the interrelationships between National Weather Service (NWS) county warning area (CWA) tornado risk, exposure, and societal vulnerability. CWA climatological tornado risk is determined using historical tornado event data, and exposure and vulnerability are assessed by employing present-day population, housing, socioeconomic, and demographic metrics. In addition, tornado watches, warnings, warning lead times, false alarm warnings, and unwarned tornado reports are examined in relation to CWA risk, exposure, and vulnerability. Results indicate that southeastern U.S. CWAs are more susceptible to tornado impacts because of their greater tornado frequencies and larger damage footprints intersecting more vulnerable populations (e.g., poverty and manufactured homes). Midwest CWAs experience fewer tornadoes relative to Southeast and southern plains CWAs but encompass faster tornado translational speeds and greater population densities where higher concentrations of vulnerable individuals often reside. Northern plains CWAs contain longer-tracked tornadoes on average and larger percentages of vulnerable elderly and rural persons. Southern plains CWAs experience the highest tornado frequencies in general and contain larger percentages of minority Latinx populations. Many of the most socially vulnerable CWAs have shorter warning lead times and greater percentages of false alarm warnings and unwarned tornadoes. Study findings provide NWS forecasters with an improved understanding of the relationships between tornado risk, exposure, vulnerability, and warning outcomes within their respective CWAs. Findings may also assist NWS Weather Forecast Offices and the Warning Decision Training Division with developing training materials aimed at increasing NWS forecaster knowledge of how tornado risk, exposure, and vulnerability factors influence local tornado disaster potential.

Tornado wind hazard mapping and equivalent tornado design wind profile for Canada

Tornadoes cause property damage and fatalities. A systematic assessment of tornado wind velocity hazard is presented in the present study for Canada based on reported historical tornado catalogue. For the assessment, the Poisson model, zero-inflated Poisson model, and negative binomial model, as well as the adaptive Gaussian kernel smoothing technique, are used to model tornado occurrence. The modeling takes into account the tornado reporting bias due to population density and uses the cloud-to-ground lightning flash density and the thunderstorm days as the explanatory variables. Statistical analysis results indicate that the negative binomial model is preferable.The tornado wind velocity hazard maps for Canada, in terms of wind velocity at 10 m height above the ground surface and return period T, VT(10), are developed based on an available probabilistic tornado wind field model. Also, a simple equivalent tornado design wind profile is developed based on the return period value of tornado wind speed at different heights. The developed tornado design wind profile only depends on VT(10), which greatly facilitates its potential use for structural design subjected to tornado wind load and its possible implementation in structural design code and standards.

U.S. Weather Highlights 2020: The Most Extreme Year on Record?

All the weather cliches applied this year. Yes, this really was “one for the record books” and a “year of weather extremes.” Notable events included winter storms, historical tornado outbreaks, Wes...

Tornado loss model of Oklahoma and Kansas, United States, based on the historical tornado data and Monte Carlo simulation

This paper introduces a new tornado loss assessment model for one-family and two-family wood frame residential buildings in the states of Kansas (KS) and Oklahoma (OK), United States. The model belongs to the group of Monte Carlo simulations and consists of five main modules: (1) tornado climatology; (2) tornado intensity and tornado track geometry; (3) exposure map; (4) fragility and vulnerability functions; and (5) financial module that is used to calculate the resulting damages. In addition, two analytical formulations of radial distribution of tangential velocity in tornado vortex were implemented in the tornado intensity module—Rankine Vortex (RV) and Modified Rankine Vortex (MRV) models. Using the MRV tornado model, the resulting mean aggregated annual expected losses for the return period of 1000 years are US$ 392,080,526.8 and US$ 641,281,336.0 for KS and OK, respectively. Although house prices in KS are higher than in OK, the higher occurrence of tornadoes in OK combined with the larger population of OK (i.e., larger density of houses) result in the increased tornado losses for OK. The losses derived using the MRV module are higher than the losses from the RV module due to the RV module underestimating the tornado tangential velocity. Furthermore, the losses predicted by the current model using the MRV formulation are less than the estimated losses from the previous version of the model (Romanic et al., 2016). The model is also tested for the case of a single tornado track positioned over suburban area of Oklahoma City, OK.

Public perception of climatological tornado risk in Tennessee, USA.

The southeastern United States experiences some of the greatest tornado fatality rates in the world, with a peak in the western portion of the state of Tennessee. Understanding the physical and social characteristics of the area that may lead to increased fatalities is a critical research need. Residents of 12 Tennessee counties from three regions of the state (N = 1804) were asked questions about their perception of climatological tornado risk in their county. Approximately half of participants underestimated their local tornado risk calculated from 50 years of historical tornado data. The percentage of participants underestimating their climatological risk increased to 81% when using model estimates of tornado frequencies that account for likely missed tornadoes. A mixed effects, ordinal logistic regression model suggested that participants with prior experience with tornadoes are more likely to correctly estimate or overestimate (rather than underestimate) their risk compared to those lacking experience (β = 0.52, p < 0.01). Demographic characteristics did not have a large influence on the accuracy of climatological tornado risk perception. Areas where more tornadoes go unreported may be at a disadvantage for understanding risk because residents' prior experience is based on limited observations. This work adds to the literature highlighting the importance of personal experiences in determining hazard risk perception and emphasizes the uniqueness of tornadoes, as they may occur in rural areas without knowledge, potentially prohibiting an accumulation of experiences.

Structural Property Losses from Tornadoes in Florida

Abstract Property losses from tornadoes in Florida are estimated by combining a 1-km spatial grid of structural values from the Department of Revenue’s 2014 cadastral database with historical tornado events since 1950. There are 91 180 grid cells in the state with at least some structural value. Total and residential structural values total $942 billion and $619 billion, respectively. Over the period 1950 through 2015 there were 3233 individual tornado reports in the state with a peak frequency during June. The property value exposed to tornadoes is estimated using a geometric model for the path. Annual statewide total and residential structural property exposure to tornadoes is estimated at $171 million and $103 million, respectively. Property exposure to tornadoes peaks in February. A regression model quantifies the relationship between actual losses since 2007 and exposures. A doubling of the residential exposure increases actual recorded losses by 26% since 2007, and a doubling of nonresidential exposure increases losses by 21%, controlling for changes over time. Randomization of the historical tornado paths provides alternative exposure scenarios that are used to determine the probability of extreme loss years. Results from the Monte Carlo algorithm indicate a 1% chance that the annual loss will exceed $430 million and a 0.1% chance that it will exceed $1 billion. These findings, and the procedure to obtain them, should help property insurance and reinsurance companies gauge their risk of losses and prioritize their management actions.

Restoration and functionality assessment of a community subjected to tornado hazard

The loss of infrastructure functionality for a community from a tornado can significantly affect economic vitality and result in substantial social disruption. The duration of the loss of functionality is a decisive factor in the magnitude of the economic disruption caused by a tornado. Therefore, quantitative models for the post-tornado restoration processes are needed in order to provide risk-informed decision support. In this study, the water network, electric power network, school buildings, residential buildings and businesses were modelled with their relative spatial distribution along with select dependencies and cross-dependencies in order to investigate the restoration of a community in the aftermath of a disaster. A tornado path was simulated using statistics from a historical tornado database. Based on the damage level of the components and their connectivity throughout the community, the functionality of the supplier nodes was assessed through a Monte Carlo simulation. Finally, the repair time of the damaged components was modelled to capture how the demands may be satisfied over time until eventually full restoration is achieved. Then, two metrics were defined to highlight business and social disruption following a tornado. The approach presented herein can be implemented in future studies to quantify indirect economic loss and social disruption resulting from a tornado.

A Few GIS Clarifications on Tornado Density Mapping

AbstractA geographical information system (GIS) perspective is taken to examine conceptual and methodological complications present in tornado density and probability mapping. Tornado density is defined as the inverse-distance-weighted count of tornado touchdown points or tornado-affected cells within a neighborhood area. The paper first adds a few geographic elements into the tornado definition and then characterizes tornado density as a density field in GIS that depends on predefined, modifiable areas to exist. Tornado density is therefore conceptually distinguished from both individual tornadoes and tornado probability. Three factors are identified to be vital in tornado density mapping: the neighborhood size, the distance decay function, and the choice of tornado properties. Correspondingly, 12 neighborhood sizes ranging from 20 to 360 km are tested, four distance decay functions are compared, and two tornado properties—tornado touchdown locations and pathlengths—are separately incorporated in mapping. GIS interpretations, clarifications, and demonstrations are provided for these factors to reach a thorough understanding of how the factors function and affect the resultant tornado density maps. Historical tornado data of the eastern half of the United States from 1973 to 2013 are used in these demonstrations. Uncertainty and propagation analyses are recommended for future tornado density and probability mapping, and a Monte Carlo simulation using tornado pathlength data is conducted as an example of uncertainty modeling. In all, tornado density mapping is diagnosed as a largely subjective activity, and the mapper needs to make multiple choices according to the mapping purpose, scale, and the involved tornado record data.

Climate analysis of tornadoes in China

Based on analysis of historical tornado observation data provided by the primary network of national weather stations in China for the period from 1960 to 2009, it is found that most tornadoes in China (85%) occurred over plains. Specifically, large numbers of tornado occurrences are found in the Northeast Plain, the North China Plain, the middle-lower Yangtze Plain, and the Pearl River Delta Plain. A flat underlying surface is conducive to tornado occurrence, while the latitudal variation of tornado occurrence in China is not so obvious. Tornadoes mainly occur in summer, and the highest frequency is in July. Note that the beginning and the time span of tornado outbreaks are different in North and South China. Tornadoes occur during May–September in South China (south of 25°N), June–September in Northeast China (north of 40°N), July–September in the middle-lower Yangtze Plain, and July–August in North China (between 25° and 40°N). More than 80% of total tornadoes occurred during the above periods for the specific regions. The 1960s and 1970s have seen about twice the average number of tornadoes (7.5 times per year) compared to the mean for 1960–2009. The most frequent occurrence of tornado was in the early and mid 1960s; there were large fluctuations in the 1970s; and the number of tornadoes in the 1980s approached the 50-yr average. Tornado occurrences gradually decreased in the late 1980s, and an abrupt change with dramatic decrease occurred in 1994. The decrease in the tornado occurrence frequency is consistent with the simultaneous climatic change in the meteorological elements that are favorable for tornado formation. Tornado formation requires large vertical wind shear and sufficient atmospheric moisture content near the ground. Changes in the vertical wind shear at both 0–1 and 0–6 km appear to be one important factor that results in the decrease in tornado formation. The changing tendency of relative humidity also has contributed to the decrease in tornado formation in China.

Central American biomass burning smoke can increase tornado severity in the U.S.

AbstractTornadoes in the Southeast and central U.S. are episodically accompanied by smoke from biomass burning in central America. Analysis of the 27 April 2011 historical tornado outbreak shows that adding smoke to an environment already conducive to severe thunderstorm development can increase the likelihood of significant tornado occurrence. Numerical experiments indicate that the presence of smoke during this event leads to optical thickening of shallow clouds while soot within the smoke enhances the capping inversion through radiation absorption. The smoke effects are consistent with measurements of clouds and radiation before and during the outbreak. These effects result in lower cloud bases and stronger low‐level wind shear in the warm sector of the extratropical cyclone generating the outbreak, two indicators of higher probability of tornadogenesis and tornado intensity and longevity. These mechanisms may contribute to tornado modulation by aerosols, highlighting the need to consider aerosol feedbacks in numerical severe weather forecasting.

Population Influences on Tornado Reports in the United States

AbstractThe number of tornadoes reported in the United States is believed to be less than the actual incidence of tornadoes, especially prior to the 1990s, because tornadoes may be undetectable by human witnesses in sparsely populated areas and areas in which obstructions limit the line of sight. A hierarchical Bayesian model is used to simultaneously correct for population-based sampling bias and estimate tornado density using historical tornado report data. The expected result is that F2–F5 compared with F0–F1 tornado reports would vary less with population density. The results agree with this hypothesis for the following population centers: Atlanta, Georgia; Champaign, Illinois; and Des Moines, Iowa. However, the results indicated just the opposite in Oklahoma. It is hypothesized that the result is explained by the misclassification of tornadoes that were worthy of F2–F5 rating but were classified as F0–F1 tornadoes, thereby artificially decreasing the number of F2–F5 and increasing the number of F0–F1 reports in rural Oklahoma.

A Simple and Flexible Method for Ranking Severe Weather Events

Abstract The notion of an “outbreak” of severe weather has been used for decades, but has never been formally defined. There are many different criteria by which outbreaks can be defined based on severe weather occurrence data, and there is not likely to be any compelling logic to choose any single criterion as ideal for all purposes. Therefore, a method has been developed that uses multiple variables and allows for considerable flexibility. The technique can be adapted easily to any project that needs to establish a ranking of weather events. The intended use involves isolating the most important tornado outbreak days, as well as important outbreak days of primarily nontornadic severe convective weather, during a period when the number of reports has been growing rapidly from nonmeteorological factors. The method is illustrated for both tornadic and primarily nontornadic severe weather event day cases. The impact of the secular trends in the data has been reduced by a simple detrending scheme. The effect of detrending is less important for the tornado outbreak cases and is illustrated by comparing rankings with and without detrending. It is shown that the resulting rankings are relatively resistant to secular trends in the data, as intended, and not strongly sensitive to the choices made in applying the method. The rankings are also consistent with subjective judgments of the relative importance of historical tornado outbreak cases.

The Determinants of Tornado Casualties and the Benefits of Tornado Shelters

Several recent powerful tornadoes have raised interest among the public and policymakers in tornado shelters. We estimate casualty functions (fatalities, injuries) based on factors like storm intensity, county population density, and time of day, using historical data from Oklahoma and then estimate expected tornado casualties for Oklahoma using historical tornado frequencies assuming no further market penetration of shelters. We then calculate the cost per life saved and injury prevented for shelters. The cost per life saved for shelters in permanent homes without discounting is around $30 million. Shelters will then provide cost effective protection for residents of mobile homes.

A Minimum Assumption Tornado-Hazard Probability Model

Abstract One of the principle applications of climatological tornado data is in tornado-hazard assessment. To perform such a hazard-potential determination, historical tornado characteristics in either a regional or tom area are complied. A model is then used to determine a site-specific point probability of a tornado greater than a specified intensity occurring. Various models require different climatological input. However, a knowledge of the mean values of tornado track width, tornado track width, tornado affected area and tornado occurrence rate as both a function of tornado intensity and geographic area, along with a violence frequency distribution, enable Mod of the models to be applied. The NSSFC-NRC tornado data base is used to supply input for the determination of these parameters over the United States. This climatic data base has undergone extensive updating and quality control since it was last reported. For track parameters, internally redundant data were used to cheek consistency. Further, r...