Severe Thunderstorm Warning Research Articles

Machine Learning Investigation of Downburst-Prone Environments in Canada

Abstract Thunderstorms are recognized as one of the most disastrous weather threats in Canada because of their power to cause substantial damage to human-made structures and even result in fatalities. It is therefore essential for operational forecasting to diagnose thunderstorms that generate damaging downdrafts of negatively buoyant air, known as downbursts. This study develops several machine learning models to identify environments supportive of downbursts in Canada. The models are trained and evaluated using 38 convective parameters calculated based on ERA5 reanalysis vertical profiles prior to thunderstorms with (306 cases) and without (19 132 cases) downbursts across Canada. Various resampling techniques are implemented to adjust data imbalance. An increase in the performance of the random forest (RF) model is observed when the support vector machine synthetic minority oversampling technique is utilized. The RF model outperforms other tested models, as indicated by model performance metrics and calibration. Several model interpretability methods highlight that the RF model has learned physical trends and patterns from the input variables. Moreover, the thermodynamic parameters are deemed to have higher impacts on the model outcomes compared to parcel, kinematic, and composite variables. For example, a considerable rise in the downburst probability is detected with an increase in cold pool strength. This study serves as one of the earliest attempts toward the fledgling field of machine learning applications in weather forecasting systems in Canada. The findings suggest that the developed model has the potential to enhance the effectiveness of issuing severe thunderstorm warnings in Canada, although further assessment with operational meteorologists is needed to validate its practical application. Significance Statement Severe thunderstorms demand particular attention in forecasting as their outflow can pose a serious threat to both structures and human life. This study uses machine learning techniques to predict whether or not a thunderstorm generates a damaging outflow in Canada. Atmospheric conditions that could trigger a severe thunderstorm are identified and discussed. Results show that the models have the potential to assist forecasters in better analyzing and predicting thunderstorms that generate destructive winds. Consequently, taking advantage of promising machine learning tools can yield more reliable forecasts of damaging thunderstorms, thereby mitigating the economic and societal burdens of these storms on Canadian communities.

The Influence of WSR-88D Intra-Volume Scanning Strategies on Thunderstorm Observations and Warnings in the Dual-Polarization Radar Era: 2011–20

Abstract The Weather Surveillance Radar-1988 Doppler (WSR-88D) network has undergone several improvements in the last decade with the upgrade to dual-polarization capabilities and the ability for forecasters to rescan the lowest levels of the atmosphere more frequently through the use of Supplemental Adaptive Intra-volume Scanning (SAILS). SAILS reduces the revisit period for scanning the lowest 1 km of the atmosphere but comes at the cost of a longer delay between scans at higher altitudes. This study quantifies how often radar volume coverage patterns (VCPs) and all available SAILS options are used during the issuance of 148 882 severe thunderstorm and 18 263 tornado warnings, and near 10 474 tornado, 58 934 hail, and 127 575 wind reports in the dual-polarization radar era. A large majority of warnings and storm reports were measured with a VCP providing denser low-level sampling coverage. More frequent low-level updates were employed near tornado warnings and reports compared to severe thunderstorm warnings and hail or wind hazards. Warnings issued near a radar providing three extra low-level scans (SAILSx3) were more likely to be verified by a hazard with a positive lead time than warnings with fewer low-level scans. However, extra low-level scans were more frequently used in environments supporting organized convection as shown using watches issued by the Storm Prediction Center. In recent years, the number of midlevel radar elevation scans is declining per hour, which can adversely affect the tracking of convective polarimetric signatures, like ZDR columns, which were found above the lowest elevation angle in over 99% of cases examined.

Radar-Based Comparison of Thunderstorm Outflow Boundary Speeds versus Peak Wind Gusts from Automated Stations

AbstractStraight-line winds are arguably the most challenging element considered by operational forecasters when issuing severe thunderstorm warnings. Determining the potential maximum surface wind gust prior to an observed, measured gust is very difficult. This work builds upon prior research that quantified a relationship between the observed outflow boundary speed and corresponding measured wind gusts. Whereas this prior study was limited to a 30-case dataset over eastern Colorado, the current study comprises 943 cases across the contiguous United States and encompasses all times of day, seasons, and regions while representing various convective modes and associated near-storm environments. The wind gust ratios (WGRs), or the ratio between a measured wind gust and the associated outflow boundary speed, had a nationwide median of 1.44, mean of 1.68, 25th percentile of 1.19, and 75th percentile of 1.91. WGRs varied considerably by region, season, time of day, convective mode, near-storm environment, and outflow boundary speed. WGRs tended to be higher in the plains, Intermountain West, and southern coastal regions, lower in the cool season and during the morning and overnight, and lower in linear convective modes relative to supercell and disorganized modes. Environments with stronger mean winds and low- to midlevel shear vector magnitudes tended to have lower WGRs, whereas those with steeper low-level lapse rates and other thermodynamic characteristics favorable for momentum transfer and evaporative cooling tended to have higher WGRs. As outflow boundary speed increases, WGRs—and their variability—decrease. Applying these findings may help operational meteorologists to provide more accurate severe thunderstorm warnings.

Affective Normative Data for English Weather Words

The research in this article examines the emotional associations people have to common weather words and to selected terms that appear in weather communications (e.g., severe thunderstorm warning). A sample of 420 university students provided ratings for each term along four dimensions: 1. Valence (unhappy vs. happy), 2. Arousal (calm vs. excited), 3. Dominance (in control/dominant vs. controlled/passive), and 4. Surprise (unsurprising/predictable vs. surprising/unpredictable). The results of this research provide descriptive statistical data for the 141 weather words along the four dimensions. The author also examined the correlations of the four dimensions across the terms and observed a high degree of association between the rated arousal and surprise characteristics of terms. In addition, the results revealed the clustering of weather words according to shared similarities across the four affective dimensions (illustrating affective-based synonymy). The results of the research are significant because they reveal a deeper understanding of the subjective and emotional experiences of the atmosphere that people may have when describing the weather of a place. Similarly, the normative data from this research may be used in the analysis of weather- or climate-based communications to characterize the emotional significance or impact of a message.

Development of a Human–Machine Mix for Forecasting Severe Convective Events

Abstract Providing advance warning for impending severe convective weather events (i.e., tornadoes, hail, wind) fundamentally requires an ability to predict and/or detect these hazards and subsequently communicate their potential threat in real time. The National Weather Service (NWS) provides advance warning for severe convective weather through the issuance of tornado and severe thunderstorm warnings, a system that has remained relatively unchanged for approximately the past 65 years. Forecasting a Continuum of Environmental Threats (FACETs) proposes a reinvention of this system, transitioning from a deterministic product-centric paradigm to one based on probabilistic hazard information (PHI) for hazardous weather events. Four years of iterative development and rapid prototyping in the National Oceanic and Atmospheric Administration (NOAA) Hazardous Weather Testbed (HWT) with NWS forecasters and partners has yielded insights into this new paradigm by discovering efficient ways to generate, inform, and utilize a continuous flow of information through the development of a human–machine mix. Forecasters conditionally used automated object-based guidance within four levels of automation to issue deterministic products containing PHI. Forecasters accomplished this task in a timely manner while focusing on communication and conveying forecast confidence, elements considered necessary by emergency managers. Observed annual increases in the usage of first-guess probabilistic guidance by forecasters were related to improvements made to the prototyped software, guidance, and techniques. However, increasing usage of automation requires improvements in guidance, data integration, and data visualization to garner trust more effectively. Additional opportunities exist to address limitations in procedures for motion derivation and geospatial mapping of subjective probability.

Nonmeteorological Influences on Severe Thunderstorm Warning Issuance: A Geographically Weighted Regression-Based Analysis of County Warning Area Boundaries, Land Cover, and Demographic Variables

Abstract Studies have shown that the spatial distribution of severe thunderstorm warnings demonstrates variation beyond what can be attributed to weather and climate alone. Investigating spatial patterns of these variations can provide insight into nonmeteorological factors that might lead forecasters to issue warnings. Geographically weighted regression was performed on a set of demographic and land cover descriptors to ascertain their relationships with National Weather Service (NWS) severe thunderstorm warning polygons issued by 36 NWS forecast offices in the central and southeastern United States from 2008 to 2015. County warning area (CWA) boundaries and cities were predominant sources of variability in warning counts. Global explained variance in verified and unverified severe thunderstorm warnings ranged from 67% to 81% for population, median income, and percent imperviousness across the study area, which supports the spatial influence of these variables on warning issuance. Local regression coefficients indicated that verified and unverified warning counts increased disproportionately in larger cities relative to the global trend, particularly for NWS weather forecast office locations. However, local explained variance tended to be lower in cities, possibly due to greater complexity of social and economic factors shaping warning issuance. Impacts of thunderstorm type and anthropogenic modification of existing storms should also be considered when interpreting the results of this study.

The Emergence of Lightning in Severe Thunderstorm Prediction and the Possible Contributions from Spatial Science

Thunderstorms capable of producing dangerous weather are common in different regions throughout the world. Identifying particular thunderstorms that will become hazardous is a considerable forecasting challenge for meteorologists making warning decisions. The goal of more accurate and earlier severe thunderstorm warning has led to the idea of using a storm's lightning behavior – a reflection of storm intensity to predict imminent storm severity. Researchers have found that storms producing severe weather are often preceded by a rapid increase in lightning frequency, or a “jump”, prompting attempts to develop a quantitative lightning-based indicator of storm severity as a forecasting aid. Although recent studies have produced promising results, this initiative is rife with opportunities for geographers as lightning continues to evolve into a reliable operational tool. Technological advances in lightning detection have enabled the research, but the inherent volume of lightning data and their time-space nature represent an obstacle and an opportunity for spatial scientists. Increasingly robust lightning-jump algorithms will require extensive thunderstorm databases, the development of which involves identification of time-space clusters of lightning that represent discrete thunderstorms of varying types, which need to be aligned with geo-referenced severe weather reports. The spatial analytical tools and techniques inherent to geography can facilitate the meteorological analyses that are needed to advance the lightning jump concept toward an operational tool. Forecasters within the USA are particularly hopeful, as they look to capitalize on data generated by the Geostationary Lightning Mapper sensor mounted aboard a key satellite for weather and climate monitoring expected to be launched in 2016.

Stochastic models of severe weather watches and warnings: transition probabilities

To alert the public to the possibility of tornado (T), hail (H), or convective wind (C), the National Weather Service (NWS) issues watches (V) and warnings (W). There are severe thunderstorm watches (SV), tornado watches (TV), and particularly dangerous situation watches (PV); and there are severe thunderstorm warnings (SW), and tornado warnings (TW). Two stochastic models are formulated that quantify uncertainty in severe weather alarms for the purpose of making decisions: a one-stage model for deciders who respond to warnings, and a two-stage model for deciders who respond to watches and warnings. The models identify all possible sequences of watches, warnings, and events, and characterize the associated uncertainties in terms of transition probabilities. The modeling approach is demonstrated on data from the NWS Norman, Oklahoma, warning area, years 2000–2007. The major findings are these. (i) Irrespective of its official designation, every warning type {SW, TW} predicts with a significant probability every event type {T, H, C}. (ii) An ordered intersection of SW and TW, defined as reinforced warning (RW), provides additional predictive information and outperforms SW and TW. (iii) A watch rarely leads directly to an event, and most frequently is false. But a watch that precedes a warning does matter. The watch type $$\{SV$$ , TV, $$PV\}$$ is a predictor of the warning type $$\{SW$$ , RW, $$TW\}$$ and of the warning performance: It sharpens the false alarm rate of the warning and the predictive probability of an event, and it increases the average lead time of the warning.

Hail events across the Greater Metropolitan Severe Thunderstorm Warning Area

Abstract. This study addresses the recent climatology of hail occurrence in the Greater Metropolitan Severe Thunderstorm Warning Area (GMSTWA) of New South Wales (NSW). The study area is a sprawling suburban area with a population of nearly 4.7 million and one of Australia's largest metropoles. The main objective is to highlight the recent temporal–spatial fluctuations of hail event frequencies and magnitudes (sizes) for each of recognized and vastly inhabited local government areas (LGAs). The relevant hail event data from 1989 to 2013 were initially derived from the severe storm archive of the Australian Bureau of Meteorology. A climatologically oriented GIS technique was then applied in the examining and mapping procedure of all hail events and hail days reported throughout the study area. By applying a specific criterion, severe hail (defined as 2 cm or more in diameter) was cautiously selected for relevant analysis. The database includes 357 hail events with sizes 2–11 cm which occurred in 169 hail days (a day in which a hail event at least more than 2 cm reported) across the region during the past 25 years. The hail distribution patterns are neither temporally nor spatially uniform in magnitude throughout the study area. Temporal analysis indicated that most of hail events occur predominately in the afternoons with peak time of 1–5 p.m. Australian eastern standard time (EST). They are particularly common in spring and summer, reaching maximum frequency in November and December. There is an average of 14.3 events per year, but a significant decreasing trend in hail frequency and associated magnitude in the recent years has been identified. In turn, spatial analyses also established three main distribution patterns over the study area which include the Sydney metropolitan, the coastal and the most pronounced topographic effects. Based on the understanding of the favorable factors for thunderstorm development in the GMSTWA, the potential impacts from climate variability and future climate change have been briefly discussed.

Observations of the 14 July 2011 Fort Collins Hailstorm: Implications for WSR-88D-Based Hail Detection and Warnings

Abstract The issuance of timely warnings for the occurrence of severe-class hail (hailstone diameters of 2.5 cm or larger) remains an ongoing challenge for operational forecasters. This study examines the application of two remotely sensed data sources between 0100 and 0400 UTC 14 July 2011 when pulse-type severe thunderstorms occurred in the jurisdiction of the Denver/Boulder National Weather Service (NWS) Forecast Office in Colorado. First, a developing hailstorm was jointly observed by the dual-polarization Colorado State University–University of Chicago–Illinois State Water Survey (CSU–CHILL) research radar and by the operational, single-polarization NWS radar at Denver/Front Range (KFTG). During the time period leading up to the issuance of the initial severe thunderstorm warning, the dual-polarization radar data near the 0 °C altitude contained a positive differential reflectivity ZDR column (indicating a strong updraft lofting supercooled raindrops above the freezing level). Correlation coefficient ρHV reductions to ~0.93, probably due to the presence of growing hailstones, were observed above the freezing level in portions of the developing >55-dBZ echo core. Second, data from the National Lightning Detection Network (NLDN), including the locations and polarity of cloud-to-ground (CG) discharges produced by several of the evening’s storms, were processed. Some association was found between the prevalence of positive CGs and storms that produced severe hail. The analyses indicate that the use of the dual-polarization data provided by the upgraded Weather Surveillance Radar-1988 Doppler (WSR-88D), in combination with the NLDN data stream, can assist operational forecasters in the real-time identification of thunderstorms that pose a severe hail threat.

Improving Effectiveness of Weather Risk Communication on the NWS Point-and-Click Web Page

Abstract The National Weather Service's (NWS) point-and-click (PnC) web page is a primary channel through which NWS directly provides routine and hazardous weather information to its users. The research presented here aims to improve risk communication of hazardous weather information on the PnC web page. The focus is on improving communication of threat existence and threat timing because this important information influences how individuals perceive and respond to a weather risk. Experimental presentations of PnC forecast information were designed for two weather scenarios: a severe thunderstorm warning and a flood watch. The experimental presentations were created by adding new textual and graphical pieces of information that were intended to better convey threat existence and timing, and they were evaluated through two rounds of nationwide surveys of PnC web page users. The survey results show that the default presentation of forecast information on the PnC web page was the least effective at conveying hazardous weather threat existence and timing. Adding start-time text and end-time text, when these information pieces were coupled, helped respondents understand the precise time that weather threats were in effect for the rapid-onset, short-duration severe thunderstorm warning and for the delayed-start, longer-duration flood watch. Adding a box graphic placed around the forecast icons further enhanced communication effectiveness by drawing respondents' attention to the weather threat. Other experimental forecast presentations were designed but were less effective at communicating hazardous weather threat existence and timing, illustrating the importance of empirically evaluating weather risk communication prior to providing it operationally.

Quantifying the impact of adding gap filling radar data on forecaster wind assessments, warnings, and confidence

AbstractExperimental weather radars are being developed that could enhance the severe weather warning process by providing higher resolution data sensed closer to the ground and with faster update rates. Because wind speed is an important criterion in the issuance of severe thunderstorm warnings, this research investigates the impact of adding these new data to the forecaster decision‐making process. In a static case review setting, 30 National Weather Service (NWS) forecasters evaluated six convective weather cases under two conditions: (1) using (conventional) WSR‐88D weather radar data, and, (2) using both WSR‐88D and additional data from an experimental four‐radar network. Forecasters' predictions of ground level wind gusts, 2–5 min into the future, were compared to measurements from ground‐based wind sensors. When provided with the additional radar data participants significantly improved the accuracy of their wind speed assessments (absolute error reduced from 5.9 m s−1 to 4.0 m s−1; p < 0.001), increased their assessment confidence ratings (p < 0.001), forecasted significantly greater wind speeds (20.4 m s−1 as opposed to 17.1 m s−1; p < 0.001), and increased the number of affirmative decisions to warn from 15 to 35 (p = 0.001). While the addition of high resolution, low altitude, rapidly updating radar data is shown to have both qualitative and quantitative benefits, training and warning policy implications for the incorporation of new technology must also be carefully considered as increased accuracy, confidence and higher wind speed estimates may lead to more warnings. Copyright © 2011 Royal Meteorological Society

The Relationship between Tornadic and Nontornadic Convective Wind Fatalities and Warnings

Abstract A database of tornado fatalities, nontornadic convective wind fatalities, severe thunderstorm warnings, and tornado warnings was compiled for the period 1986–2007 to assess the spatial and temporal distribution of warned and unwarned fatalities. The time of fatality and location as reported in Storm Data was compared to tornado and severe thunderstorm warnings to determine if a warning was in effect when the fatality occurred. Overall, 23.7% of tornado fatalities were unwarned, while 53.2% of nontornadic convective wind fatalities were unwarned. Most unwarned tornado fatalities occurred prior to the mid-1990s—coinciding with modernization of the National Weather Service—while unwarned nontornadic convective wind fatalities remained at a relatively elevated frequency throughout the study period. Geographic locations with high numbers of unwarned tornado and nontornadic convective wind fatalities were associated with one high-magnitude event that was unwarned rather than a series of smaller unwarned events over the period. There are many factors that contribute to warning response by the public, and the issuance of a severe thunderstorm or tornado warning is an important initial step in the warning process. A better understanding of the characteristics of warned and unwarned fatalities is important to future reduction of unwarned fatalities.

A Case Study of Severe Winter Convection in the Midwest

AbstractBetween 2100 UTC 11 February 2003 and 0200 UTC 12 February 2003, a line of thunderstorms passed swiftly through parts of eastern Iowa and into north-central Illinois. Although this storm somewhat resembled a warm season, line-type mesoscale convective system, it was unique in that the thunderstorm winds exceeded the severe criterion (50 kt; 25.7 m s−1) during a snowburst. While the parent snowband deposited only 4 cm of snow, it did so in a short period and created a treacherous driving situation because of the ensuing near-whiteout conditions caused by strong winds that reached the National Weather Service severe criteria, as the line moved across central Illinois. Such storms in the cold season rarely occur and are largely undocumented; the present work seeks to fill this void in the existing literature.While this system superficially resembled a more traditional warm season squall line, deeper inspection revealed a precipitation band that failed to conform to that paradigm. Radar analysis failed to resolve any of the necessary warm season signatures, as maximum reflectivities of only 40–45 dBZ reached no higher than 3.7 km above ground level. The result was low-topped convection in a highly sheared environment. Moreover, winds in excess of 50 kt (25.7 m s−1) occurred earlier in the day without thunderstorm activity, upstream of the eventual severe thundersnow location. Perhaps of greatest importance is the fact that the winds in excess of the severe criterion were more the result of boundary layer mixing, and largely coincident with the parent convective line. This event was a case of forced convection, dynamically linked to its parent cold front via persistent frontogenesis and the convective instability associated with it; winds sufficient for a severe thunderstorm warning, while influenced by convection, resulted from high momentum mixing downward through a dry-adiabatic layer.

Severe thunderstorm and tornado warnings in Europe

This study summarizes the current severe thunderstorm warning and forecast operations in different European National Hydro-Meteorological Services (NHMSs), and, in doing so, suggests ways for countries developing their own warning service to learn from experiences in other countries, as well as from the warning operations in the United States of America, the longest-lived severe thunderstorm warning operations in the world. This study is based on a questionnaire sent to 39 European NHMSs of which thirty-three (85%) responded. Currently, many European NHMSs are actively developing their severe thunderstorm forecast process and warning services with 26 (79%) of respondent countries issuing severe thunderstorm warnings and 8 (24%) issuing tornado warnings. Both warning criteria and methodologies used in the warning process vary from country to country. Lead-times range from 30 min to 96 h, indicating a range of different warning philosophies for each country. Major challenges toward improving the warning operations include obtaining observations of severe weather for real-time forecasting and post-event verification, educating forecasters, and having access to state-of-the-art forecaster workstations. An additional challenge is in communicating anticipated or ongoing severe thunderstorms, both internationally between NHMSs and nationally with media and emergency authorities.

Warning Operations in Support of the 1996 Centennial Olympic Games

During the summer of 1996, the National Weather Service (NWS) provided weather support for the Centennial Olympic Games in Atlanta, Georgia. This weather support effort presented many challenges, particularly in the area of providing short-term forecast (watch) and warning support. Topping the list of challenges was working with a customer with different weather information needs than the general public. The needs of the venue and competition management were much more detailed than the NWS traditionally is accustomed to and the thresholds for various phenomena were very low (e.g., warnings for the occurrence of any rain were issued rather than the more traditional NWS severe thunderstorm warning). This paper discusses many of the challenges faced and met by the Olympic Weather Support Office (OWSO). Details are provided on the weather warning requirements of the Olympic venue and competition management, the watch/warning strategy utilized by the OWSO, and the resulting performance of the office. More than 1200 watches and warnings were issued during the period of Olympic weather support. These bulletins were for phenomena ranging from dew formation and low visibility to lightning and heavy rain. Several emerging technologies were employed for warning operations in the OWSO, including the Warning Decision Support System and the Watch/Warning/Advisory package. These hardware/software solutions appear to have made a positive impact on the performance of the office in meeting the challenges of this unique warning situation.

Storm Motion as Related to Boundary-Layer Convergence

Data presented indicate that severe thunderstorm areas may become anchored to intense, boundary- layer convergence zones. In such cases. the mean cloud-layer wind vector has little relation to the movement of regions of large hail, wall clouds and tornadoes. Suggestions for improving severe thunderstorm warnings are offered based on this possibility.

Doppler Radar Application to Severe Thunderstorm Study and Potential Real-Time Warning

Significant advances in single-Doppler radar application to severe storm study and identification have been made since 1965. Mesocyclones have been detected by Doppler radar and found to precede severe weather occurrence by several tens of minutes. A typical mesocyclone evolution leading to tornado development has also been documented. The tornado vortex itself has a revealing signature in Doppler radar data, the tornadic vortex signature (TVS). Statistics of both mesocyclone and TVS association with confirmed severe weather are presented in this paper. Doppler radar provides the potential for improving severe thunderstorm warnings. Experiments are underway to test the operational use of this new tool in storm warning and flight advisory services.