Abstracts

Hail hazard and risk assessment in Europe and the relation to orographic and atmospheric characteristics.

Michael Kunz, Heinz Jürgen Punge, Elody Fluck, Manuel Schmidberger, Susanna Mohr, David Piper and Marc Puskeiler

Karlsruhe Institute of Technology (KIT)

Presenter: Kunz, Michael

Hailstorms are among the most common and costly natural disasters in several regions of Europe. In Germany in 2013, for example, two supercells on 27 and 28 July 2013 bearing hailstones with diameters of up to 10 cm caused economic losses in the order of 4.0 billion EUR, representing the costliest natural catastrophe worldwide in 2013 (insured losses) and one of the most expensive loss events in Germany. Therefore, detailed knowledge about the probability and severity of hail events and the related risk is highly demanded by several stakeholders such as insurance companies, public authorities, vulnerable industry branches or power producers.

A major challenge for such hazard and risk assessments is the reconstruction of past hail footprints from remote sensing instruments. Depending on the purpose, we apply different multi-criteria approaches based on either satellite or radar data in combination with reanalysis. The applied methods are calibrated and verified with both insurance loss data and the few available observations as provided, for example, by the European Severe Weather Database (ESWD). In combination with a stochastic modeling approach, assessments of the hail probability for Germany, France, Belgium, and also for the whole of Europe are integrated in different hail risk models applied in the insurance industry.

Apart from the application relevance, the analyses give important insights in the processes how hailstorms are preferably triggered. The spatial variability of the frequency of hail signals is supposed to be caused by the superposition of large- scale climatology and local-scale flow dynamics that is somewhat related to the orography. Out finding that most of the hailstorms occur preferably over and downstream of mountain peaks support the hypothesis that orographically-induced flow convergence in the lee is a dominant triggering mechanism over complex terrain. These convergence zones tend to occur especially for low Froude number flows, where the flow is expected to go partly around the mountains. Sensitivity studies with the numerical weather prediction model COSMO in different setups provided further evidence of the importance of this flow effect related to severe convective storms.


Statistics of US tornado reports

Michael K. Tippett, Joel E. Cohen, John T. Allen, Adam H. Sobel, Suzana J. Camargo

Columbia University

Presenter: Tippett, Michael

Each year torndoes kill people and damage property in the US where they are relatively common. Despite records that have been collected for more than 50 years, even the annual statistics of US tornado occurance remain somewhat uncertain, due in large part to the reporting process which requires observers and has changed over time. A warming climate and its expected impacts on severe thunderstorm frequency are a further confounding factor. We review some notable  aspects of the US tornado report record with an emphasis on changes in variability and the use of atmospheric proxies which provide an independent source of evidence.


What do we know about tornadoes in Europe?

Bogdan Antonescu, David M. Schultz, Pieter Groenemeijer, and Thilo Kühne

University of Machester

Presenter: Antonescu, Bogdan

Our current understanding of the climatology of tornadoes in Europe is based on collections of historical events, cases studies and regional climatologies, all of which are limited by the inconsistencies in reporting practices and observational networks across Europe. Recently, this situation began to change with more tornadoes reported in the last decade compared to the previous period and with reports now coming from the majority of European countries, which allow us to build more accurate and complete climatologies of European tornadoes.

What do we know about tornadoes in Europe? There are two main answers to this question. First, there is not a widespread recognition of the threat of tornadoes to Europe and, as a result, few European meteorological services forecast tornadoes. This lack of recognition results in an underestimate of the tornado threat to Europe. Second, understanding the influence of climate change on severe convective storms and their associated phenomena (e.g., tornadoes) remains a great challenge. When considering this influence of a future warmer climate, the starting point is to consider the observational data. Unfortunately, there is a lack of long organized and centralized tornado records. Although the first collections of tornado records were published during the first half of the 19th Century, no system was in place until very recently to collect tornado reports at a European scale (i.e., the European Severe Weather Database). Thus, without knowing what is the spatial and temporal distribution of tornadoes in the current climate, all the analyses of the influence of climate change on tornadoes are premature.

For these reasons we present a synthesis of tornado observations for 30 European countries between 1800 and 2014, based mainly on regional tornado-occurrence datasets and articles published in peer-reviewed journals, with the aim of producing a pan-European climatology. Thus, 9549 tornadoes were included in the synthesis with the majority of the reports coming from northern, western and southern Europe, and to a lesser extent from eastern Europe where tornado databases were developed after the 1990s. The annual average number of tornado reports increased from 8 tornadoes yr-1 between 1800–1850 to 242 tornadoes yr-1 between 2000–2014. The tornado season has a maximum in June–August for the majority of the European countries analysed, and in August–November over southern Europe. The intensity distribution for the 5187 tornado reports for which an estimate of the F scale was possible shows that the majority of tornado reports, 74.7% of all tornado reports, were for weak tornadoes (F0 or F1), 24.5% for strong tornadoes (F2 or F3) and 0.8% for violent tornadoes (F4 and F5).


Continental Subtropical Anticyclones and Warm-Season Progressive Derechos

Lance F. Bosart and Corey T. Guastini

University at Albany/SUNY

Presenter: Bosart, Lance

Subtropical anticyclones exhibit a strong preference for continental locations during the warm season where they are usually associated with extended heat waves and droughts. Subtropical anticyclones over North America can also serve as a pathway for a category of damaging windstorms known as progressive derechos along their poleward periphery. Progressive derechos often occur in conjunction with upper-level disturbances, known colloquially as ridge rollers, and may produce surface wind gusts > 40 m s-1 along a swath > 1000 km. Progressive derechos typically move eastward around the poleward side of subtropical anticyclones and often form in the equatorward entrance region of strong jet streams.

An 18-year Doppler radar-based climatology of warm-season progressive derechos reveals that they occur preferentially east of the Rockies and are most frequent along a corridor from the Dakotas east-southeastward to the Ohio Valley. A typical progressive derecho formation environment is characterized by the presence of weak surface boundary with high surface dew points (> 20 °C) in the warm, humid air south of the boundary, moderate surface-to-6 kilometer shear (10–20 m s-1) that is quasi unidirectional above veering winds in the boundary layer, and strong instability (convective available potential energy values > 4000 J kg-1). This instability is often enhanced by the presence of elevated mixed layers that originate over the Rockies and are advected eastward in the stronger flow poleward of the subtropical anticyclones. On progressive derecho days, convection initially may be elevated just poleward of a weak surface boundary in a warm, moist, unstable environment. As the convection organizes further and grows upscale it becomes surface-based, shifts to the warm side of the aforementioned surface boundary, and develops a strong surface-based cold pool.

A distinguishing characteristic of strong warm-season progressive derechos is that once a deep surface-based cold pool is in place the storms move eastward at a faster rate than the wind at any level in the immediate storm environment. This rapid movement, which can pose predictability challenges, can catch people unaware because the typical “canary-in-the-coal-mine” severe-storm anvil trails the leading, high-wind edge of the storm because the cold-pool driven derecho is moving faster than the mean wind at any level. Illustrative examples of derechos will be shown and the climate implications of derechos and subtropical anticyclones will be discussed.


Impact of US Temperature Anomalies on Tornado and Hail Occurrence

Harold E. Brooks

NOAA/NSSL

Presenter: Brooks, Harold

The relationship between monthly and 3-monthly US national temperature anomalies on national tornado and hail occurrence is explored. For F1+ tornadoes, warm winter (summers) are associated with more (fewer) tornadoes, with the significance higher in the summer. For 1.25 inch+ hail on a grid, the signal is qualitatively similar, but weaker. It’s possible that the skill of forecasting temperature anomalies could allow this to provide some information on seasonal tornado and hail occurrence, as well as providing hints about future occurrence under climate change scenarios.


The Contribution of ENSO to Hail and Tornado Seasonal Variability

John Allen

IRI, Columbia University

Presenter: Allen, John

Relating hail and tornado occurrence to the climate system is an important step on the road to seasonal and sub-seasonal forecasts of severe thunderstorm activity for the United States (U.S.). Hail and tornado indices for the probability of occurrence as a function of convective parameters have been derived using the National Climatic Data Center’s Storm Data and environmental data from the North American Regional Reanalysis (NARR) for the period 1979-2014. The value of using these indices, along with carefully controlled observations, to identify links between the climate system and severe thunderstorms will be illustrated using the El Niño Southern Oscillation (ENSO). The phase of ENSO has long been hypothesized to influence severe thunderstorm occurrence over the U.S. However, limitations in the severe thunderstorm observation record, combined with large year-to-year variability have made demonstrating such a relationship difficult, particularly during spring, the peak hail and tornado season. We show that fewer hail and tornado events occur over the central United States during El Niño and conversely more occur during La Niña, but this is modulated by the strength of the event. It remains poorly understood what climate signals contribute to severe thunderstorm activity during when ENSO is marginal, or weak.

In this presentation the influence of ENSO on tornado and hail frequency will be discussed. Based on this relationship, a seasonal forecast was made for 2015 via a simple statistical model. Given the current strong El Niño conditions the two-faceted seasonal forecast for the spring of 2016 will be presented. Finally, the implications for marginal or weak non-ENSO years will also be considered.


Spring Tornado Activity in the United States and the GWO

Victor Gensini

College of DuPage

Presenter: Gensini, Victor

Prediction of tornado activity beyond one week presents a challenging task due to the space and time scales involved. It is shown that revisiting a prior global dynamic framework with new perspective explains nearly an order of magnitude of the variability in U.S. tornado occurrence. Tornadoes are more (less) likely to occur in low (high) angular momentum base states. Composite environmental analysis suggests increases/decreases in tornado occurrence are associated with anomalies in tropospheric ingredients necessary for tornadic storms. Results herein provide a significant contribution to our understanding of U.S. peak-season tornado variability and a uniting framework for extended range forecasting of these extreme weather events. Given that the GWO includes the MJO (and other extratropical forcing such as mountain and frictional torque), using the GWO may serve to increase the predictive capability of the MJO for endeavors.


Subseasonal variability of severe storms in the U.S.: what we know and don’t know

Bradford Barrett, Victor Gensini

U.S. Naval Academy

Presenter: Barrett, Bradford

There is a growing desire to understand how the climate system influences severe thunderstorms over the United States. Due to the relatively small spatial (~10^-1 km) and temporal (~10^4 s) scales of tornadoes and hailstorms, understanding not only why their frequency varies from year-to-year, but also why these events tend to cluster in space and time, remains a challenging task. This is especially true of tornado and hail activity in the subseasonal (30-60—day) time scale, which is beyond the extent of most medium range (up to two weeks) numerical weather prediction solutions. Research on subseasonal variability of U.S. severe weather remains in its infancy, owing to a relatively poor understanding of the processes by which the climate scale interacts with severe thunderstorms. Furthermore, subseasonal variability may result from the projection of both demonstrated and as yet unknown climate teleconnections on U.S. severe thunderstorms. One of the leading candidates for modulating subseasonal variability of severe storm activity is the Madden-Julian Oscillation (MJO), which is the dominant mode of atmospheric variability in the 30-60—day window. Three recent studies have explored modulation of tornadoes and hail by the MJO by creating atmospheric composites and anomalies by phases of a leading MJO index. Both tornadoes and hail were found to vary by MJO phase, and this variability was found to be somewhat supported by corresponding dynamic (e.g. wind shear) and thermodynamical (e.g. instability) characteristics of the atmosphere in the same phases. In this presentation, strengths and weaknesses of these results are discussed and used to frame a plan for future research in subseasonal severe storm variability.


An Introduction of Reinsurance for Scientists

Kelly Hereid

Chubb Tempest RE

Presenter: Hereid, Kelly

Insurance is a basic way that individuals can protect themselves from the financial consequences of being impacted by a natural disaster.  But how does the insurance industry maintain its ability to pay claims in the biggest events?  Reinsurance is “insurance for insurance companies” – a method used by insurance companies to diversify their risk portfolios and protect themselves against catastrophic losses.  Reinsurers’ focus on high-impact disasters puts the industry on the front lines of climate change response in the business community.

I will discuss the fundamentals of how reinsurance works, covering the types of treaties that influence our view of natural hazards.  I will also touch on emerging methods of risk transfer, and highlight a few case studies that demonstrate the overlap between reinsurance business cases and emerging climate research in the meteorological community.


A Broker View of Cat. Risk

Rick Thomas

Willis Towers Watson

Presenter: Thomas, Rick

In the current reinsurance market there are a wide range of tools and potential methods available to derive a view of Catastrophe risk for most perils. Willis takes the view that model adjustment based on detailed understanding of all components is the best approach when good models already exist. However for some regions and perils the only valid way forward is to build new models.  Either way there is a persistent need to improve all aspects of our knowledge of Catastrophe perils, either by gathering more complete data or creating better models for simulating them. This is especially true for Severe Convective Storms, which are generally poorly modelled from an insurance industry perspective.


The Winds of Change: How CAT models can help us understand the changing landscape of severe thunderstorm risk

Eric Robinson

AIR Worldwide

Presenter: Robinson , Eric

 


The RMS Approach to Severe Convective Storm Modeling

Kevin Van Leer

Risk Management Solutions, Inc.

Presenter: Van Leer, Kevin

This presentation will highlight the components of the RMS North America Severe Convective Storm Model along with future research topic ideas for improvement of severe convective storm modeling overall. General details of the modeling approach to hazard, vulnerability, and financial loss for this peril will be discussed. Future details on the presentation will be provided at a later date.


Using science to improve our lives: Anticipating, measuring and reacting to losses from SCS.

Tom Larsen

CoreLogic

Presenter: Larsen, Tom

In 2013, $30 out of every $100 collected in homeowners’ premiums went towards hail and wind claim payments – the disruption to people’s lives is significant, and the monetary costs are ultimately borne by the insurance rate payer. Paramount concerns to insurers include aligning risk pricing with the localized risk for the building and a more efficient claims assessment process that leverages real-time weather footprints. Weather modeling technology and analytics are enabling insurers to deliver better and less costly products to their clients.


Weathering The Storm: Bringing Clarity To The Unknowns Of Severe Thunderstorm Modeling

Steve Drews

Aon Benfield

Presenter: Drews, Steve

Severe thunderstorm catastrophe models have not met the expectations of insurance companies since the dawn of severe thunderstorm modeling.  Insurance companies have the responsibility to make sure the premiums that they charge are in line with the level of risk, and this process has long been stable for the perils of hurricanes and earthquakes.  For the peril of severe thunderstorms, however, insurance companies have experienced extreme difficulty in modeling this peril and pricing their risk accordingly due to a lack of accurate understanding of peril frequency and severity, so much so that many companies have abandoned their usage of the models and have reverted back to using their claims experience to determine their premium levels. This approach can be very dangerous, as years like 2011 can catch insurance companies off-guard with extremely high retained losses. Additionally, companies are growing into areas where they lack detailed claims experience to accurately price the level of risk being underwritten and may also be changing the way they underwrite these premiums for hail risk, especially for roofs. Aon Benfield Impact Forecasting listened to insurance companies’ issues with the models, developed a unique historical model, and updated its own probabilistic model to accurately reflect the frequency, severity, and loss costs of the subperils of tornadoes, hail, and convective winds.


The realization of extreme tornadic storm events under future anthropogenic climate change.

Jeff Trapp

University of Illinois at Urbana-Champaign

Presenter: Trapp, Jeff

Here we seek to answer the basic question of how current-day extreme tornadic storm events might be realized under future anthropogenic climate change. The “pseudo-global warming” (PGW) methodology was adapted for this purpose. Three contributions to the CMIP5 archive were used to obtain the mean 3D atmospheric state simulated during May 1990–1999 and May 2090–2099. The future-past differences in temperature, relative humidity, pressure, and winds were added to NWP analyses of three high-end tornadic storm events, and this modified atmospheric state was then used for initial and boundary conditions for real-data WRF model simulations of the events at high resolution. Comparison of an ensemble of these simulations with control simulations (CTRL) facilitated assessment of PGW effects.

In contrast to the robust development of supercellular convection in each CTRL, the combined effects of increased convective inhibition (CIN) and decreased parcel lifting under PGW led to a failure of convection initiation in many of the experiments. Those experiments that had sufficient matching between the CIN and lifting tended to generate stronger convective updrafts than CTRL, because of the relatively higher convective available potential energy (CAPE) under PGW. And, the experiments with enhanced updrafts also tended to have enhanced vertical rotation. In fact, such supercellular convection was even found in simulations that were driven with PGW-reduced environmental wind shear. Notably, the PGW modifications did not induce a change in the convective morphology in any of the PGW experiments with significant convective storminess.


What drives the increase in CAPE that drives the increase in severe weather?

David M. Romps

UC Berkeley

Presenter: Romps, David

Global climate models (GCMs) predict large increases in CAPE with global warming, and this is the driving force behind predicted increases in the frequency of severe weather and lightning strikes. Before we can have any confidence in these predictions, however, we must first develop a theory for CAPE. The standard lore is that CAPE increases with warming because water vapor, which is the fuel for thunderstorms, exhibits Clausius-Clapeyron (CC) scaling with temperature. Of course, this lore must not be mistaken for fact; at best, it is a hand-waving placeholder for a to-be-developed theory. Fortunately, during the past three years, substantial progress has been made towards a theory for CAPE. This has led to the first analytical expressions for CAPE, based on solutions to the coupled equations for moist convection and its environment. These equations manifest CC scaling up to a critical temperature (about 310 K at the cloud base), above which CC scaling fails. Additional insights include the reason behind the top-heavy buoyancy profiles of tropical CAPE, and the surprising insensitivity of CAPE to the latent heat of fusion.


Trends in hail and severe weather activities in China over the past 30 years under the changing monsoon climate

Qinghong Zhang, Fuqing Zhang, Xiang Ni and Mingxin Li

Peking University

Presenter: Zhang, Qinghong

Based on exclusive station-by-station hourly-to-daily records of hails, lightning, and damaging winds, we examine the frequency changes of severe, extreme and convective weather activities in China over the past 30 years. In particular, it is found that there is a decline trend in the total number of hail and severe weather events over China over the past few decades but the changes vary from region to region. In particular, hail frequency in northern China and Tibetan Plateau decreases by 42 and 53 percent, respectively since the early 1980s while no clear trend in southern China. It is also found that hail duration, as well as intensity (in terms of maximum hail diameter) in China showed a significant decreasing trend in the northern China and Tibetan Plateau, while no trend in southern China from 1982 to 2012. The variation in hail activity is found to be associated with the variation of severe weather events across China. Preliminary analysis shows that the overall decreasing hail activity and severe convective weather over northern China might be attributable to the weakening Asia summer monsoon while the reduction of hail activity in Tibet Plateau might also be due to the decrease of convective instability that is associated with global warming.


Statistical modelling of thunderstorms in the present and future climate

Anja Westermayer (1,2), Pieter Groenemeijer (2), Tomas Pucik (2), Robert Sausen (3), and Eberhard Faust (1)

(1) Munich Re, München, Germany (Anja.Westermayer@physik.lmu.de), (2) European Severe Storms Laboratory e.V. (ESSL), Wessling, Germany, (3) Deutsches Zentrum für Luft und Raumfahrt (DLR), Institut für Physik der Atmosphäre, Oberpfaffenhofen, Germany

Munich Re and ESSL

Presenter: Westermayer, Anja

A statistical model was developed for the occurrence of electrified convection across Central Europe, based on ERA-Interim reanalysis data and lightning detection data from the European Cooperation for Lightning Detection. The model was developed by fitting an additive logistic regression to multiple selected physical parameters, whose individual relation to lightning occurrence was studied a priori. In doing so, it was found that lightning occurrence is strongly dependent on mid-tropospheric humidity.

A re-application of the model to ERA-Interim reanalysis data shows that it can reproduce the annual cycle of thunderstorms in northern and southern half of the Central European domain rather accurately. Applying the model to the entire ERA-Interim dataset, which starts in 1979, reveals long-term changes in thunderstorm occurrence and variability.

Subsequently, the model was applied to EuroCORDEX regional climate simulations to create predictions of changes in thunderstorm probability until 2100 according to the rcp 4.5 and rcp 8.5 climate scenarios. These changes and the processes responsible causing them are studied by an evaluation of the changes of individual parameters included in the model.


High-resolution regional climate simulations of warm season convection in the United States

Kristen Rasmussen and Roy Rasmussen

National Center for Atmospheric Research

Presenter: Rasmussen, Kristen

Convective storms and mesoscale convective systems are vital to hydrologic and energy cycles on Earth. As the global climate changes, patterns of severe weather are likely to shift. In order to accurately represent all types of storms in numerical forecasts and climate models, it is becoming increasingly important to incorporate the effects of physical mechanisms and specific details involving clouds and mesoscale processes, storm life cycle, mesoscale organization, topographic and diurnal effects, latent heating, and precipitation from such storms. High-resolution regional climate simulations (i.e., the pseudo-global warming technique; Rasmussen et al. 2011) that model the current and future climate scenarios in a convection-resolving framework have recently been completed for ten years of current and future climate in the United States. We will present results from these simulations focused on the warm season in the U.S. In addition, changes in the environment supporting convection and the resulting intensity and frequency of convective storms will be shown. Understanding how future changes in climate will impact convective storms of all scales and intensities is an important research topic that may define the weather and climate intersection.


Our current understanding of tornadic storm dynamics

Yvette Richardson

Penn State University

Presenter: Richardson, Yvette

Our current understanding of tornado genesis, maintenance, and demise based on theory, observations, and numerical modeling studies will be reviewed.  In particular, we examine the possible origins of near-surface rotation, its contraction to tornado strength, the subsequent maintenance of the tornadic vortex, including the possible role of internal cold pool surges, and processes leading to tornado demise, including the displacement of the near-surface rotation from the midlevel mesocyclone.


Paving the way for High-Resolution Ensemble Prediction: Recent Activities and future plan for NOAA/Hazardous Weather Testbed Spring Forecasting Experiments.

Adam J. Clark

NOAA/OAR/National Severe Storms Laboratory

Presenter: Clark, Adam

Annual Spring Forecasting Experiments (SFEs), co-organized by NOAA’s National Severe Storms Laboratory (NSSL) and Storm Prediction Center (SPC), aim to accelerate the transfer of promising new tools and concepts from research to operations, inspire operationally relevant research, and identify and document sensitivities and performance of state-of-the-art convection-allowing (1 to 4-km grid-spacing) experimental modeling systems. The 5-week experiments are held annually during the climatological peak of the severe weather season in NOAA’s Hazardous Weather Testbed in the National Weather Center building in Norman, OK. The HWT is situated between the operations areas of the SPC and OUN local Weather Forecasting Office and equipped with workstations similar to those used to generate operational products. Thus, the HWT is ideally suited to SFEs, which bring together a wide range of meteorologists, including forecasters (both government and private sector), researchers, and academics in a simulated, real-time forecasting environment. Main emphases of recent SFEs include testing the feasibility of new types of severe weather outlooks that provide more specific information than current operational products, and use and evaluation of new convection-allowing ensembles contributed by NSSL, the Center for Analysis and Prediction of Storms (CAPS), the National Center for Atmospheric Research (NCAR), and the Air Force Weather Agency (AFWA). In the coming years, the HWT will also be used to test NSSL’s Warn-on-Forecast initiative, which aims to dramatically reduce severe weather warning lead time using short-term, rapidly-updating, high-resolution ensemble NWP with advanced ensemble data assimilation.

In this talk, I’ll discuss the philosophy and history behind the annual SFEs, and I’ll highlight key results from recent experiments. In addition, I’ll discuss future plans for upcoming SFEs that involve heightened coordination efforts among collaborators to identify the most effective ensemble configuration strategies to help guide implementation of the first operational convection-allowing ensemble by NCEP’s Environmental Modeling Center in 2017.


Measurements of In-Cloud and Cloud-to-Ground Lightning and the relationship with Severe Storms.

Kristin Calhoun

OU/CIMMS & NOAA/NSSL

Presenter: Calhoun, Kristin

The majority of lightning produced by storms, particularly within severe storms, remains in-cloud. Trends in total lightning (in-cloud and cloud-to-ground) have been well correlated with trends in graupel volume, updraft mass flux, and other changes in storm intensity. Through this relationship, total lightning data provides a particularly attractive addition to the current, radar-only analyses and subsequent forecasts of severe convective storms. However, until recently, the detection and monitoring of total lightning has been primarily utilized only within the research community or for unique events such as space missions. Additionally, total lightning data may be able to provide extra lead-time over traditional radar data, highlighting which storms are electrically active and growing quickly as opposed to those that are not. Cloud-to-ground (CG) lightning appears to have little relationship to storm severity, but may be related to formation and decent of precipitation at lower heights of a storm. CG lightning is its own a significant hazard, however, and many users of lightning data are mainly concerned with where a CG flash has already occurred or has the potential to connect to ground.

Multiple systems have been developed specifically to detect lightning activity, each one with its own capabilities and limitations. Most modern networks are able to determine some subset of frequency, location, and intensity. Ground-based networks operating at very low/low frequencies (VLF/LF) are better at detecting the radiation produced by the return strokes associated with CG lightning over a wide area, whereas networks operating within very high frequency (VHF) bands are more able to detect the breakdown processes associated with lightning channels in-cloud across a small region. Satellite-based detection of lightning (e.g., the Lightning Imaging Sensor or the upcoming Geostationary Lightning Mapper) relies on the optical identification of lightning; these systems detect roughly 70-90% of total lightning, but at lower resolution than most ground-based systems. Which system or combination of systems should be used depends greatly on the user’s needs. In addition to reviewing the relationship between lightning activity and severe weather, this presentation will compare the various lightning location systems and highlight the strengths and weaknesses of each.


Automated Detection of Hazardous Storm Cells Using Long-Term Databases of Satellite Imager Observations

Kristopher Bedka, Konstantin Khlopenkov, Cecilia Wang, Heinz Jurgen Punge, and Wim Thiery

NASA Langley Research Center

Presenter: Bedka, Kristopher

Thunderstorms can produce a variety of hazards such as hail, damaging winds, tornadoes, heavy rain, lightning, aircraft icing, and turbulence, each of which represent a significant threat to life and property. Many hazardous storms have updrafts of sufficient intensity to penetrate through the tropopause and into the lower stratosphere, transporting tropospheric aerosols, chemical species, water vapor, and ice which have a significant impact on the Earth’s climate. Studies have suggested that global hazardous storm activity will increase in association with projected climate change, as warmer surface temperatures will lead to greater atmospheric instability favorable for hazardous storm formation. Unfortunately very few consistent datasets currently exist to define global historical hazardous storm activity throughout the full diurnal cycle, leaving the community unable to determine if future activity has changed relative to the past or if change has already occurred.

Hazardous storm updraft regions appear anomalously cold and turbulent in infrared and visible wavelength satellite imagery, signatures that can be detected using automated pattern recognition algorithms developed at NASA Langley Research Center (LaRC) in support of the GOES-R Advanced Baseline Imager program. Multispectral geostationary (GEO) and polar-orbiting imager and reanalysis data are used to identify clouds that have penetrated through the tropopause region. These so-called “overshooting cloud top” (OT) signatures are highly correlated with the hazards and cross-tropopause transport described above.

This OT pattern recognition approach paired with LaRC’s immediate access to the entire GEO and AVHRR imager records enables establishment of long-term (up to 35 years) high temporal and spatial resolution databases of hazardous storm events. These datasets have recently been applied to determine hail risk over select regions and to understand the diurnal evolution and spatial distribution of hazardous storms over regions lacking operational radar and lightning sensors. This presentation will describe the automated OT detection approaches developed at LaRC and will highlight recent applications and future opportunities for long-term satellite-based OT databases.


Prospects for Future Radar-based Nowcasting of Tornado Formation and Dissipation

Michael M. French

Stony Brook University

Presenter: French, Michael

In the past 5 years, data obtained by mobile phased-array and dual-polarization radars have been used to gain significant insight into supercell and tornado behavior. The former system allows for tornado behavior to be tracked over very short time scales and the latter system is used to gather additional information about supercell hydrometeors through remote sampling. By using mobile platforms, these two technologies have been utilized in a manner such that fine spatial resolution is emphasized.

Data from a mobile phased-array radar with volumetric updates every ~10 s have supported previous hypotheses about the prevalence of “bottom-up” tornado development. In several cases these data also have provided a unique look at rapid changes in tornado vertical orientation and tornado storm-relative motion prior to tornado dissipation. Tornado dissipation has been found uniquely to occur in a vertically “inside-out” manner. Separately, data from a mobile dual-polarization radar have provided additional evidence to support hypotheses about the relationship between storm processes thought to be important for tornado development and changes in drop-size distribution (DSD) proxies. In addition, these data have been used to introduce possible relationships between DSD proxies and tornado dissipation.

Together, this type of radar information may be thought of as one contributor, in addition to efforts such as Warn-On-Forecast, to a future probabilistic system designed to forecast the development of a tornado and project when an existing tornado may dissipate. This talk will summarize this past radar work to motivate these nowcasting possibilities. In addition, preliminary results of efforts to validate case study results using a larger number of tornadic storms will be presented. Also discussed will be pros and cons of such an approach and its ultimate utility to both the meteorological community and the public.


Quantifying hail hazard from convective overshooting using peril-specific environmental conditions

Heinz Jürgen Punge, Kristopher M. Bedka, and Michael Kunz

Karlsruhe Institute of Technology

Presenter: Punge, Heinz Jürgen

Satellite derived datasets of cloud overshooting tops (OT) can serve as a proxy for severe convection. However, the occurrence and intensity of specific hazards associated to these severe storms, such as hail, heavy rain, or severe wind, may depend on the local and regional atmospheric conditions. In this study we try to take such dependencies into account in order to derive a reliable climatology of hail.

The environmental conditions associated to large hail events as reported in a severe weather database are determined from the ERA-INTERIM reanalysis. Whenever the conditions associated to an OT fall outside the range of reported hail events, hail is considered unlikely. This way, a filter for hail is constructed. For example, we find that hail is less likely in severe thunderstorms in Southern Europe and North Africa than in Central Europe due to a higher freezing levels. Further relevant variables include the dew point and the equivalent potential temperature.

Using a total of 10 atmospheric variables a hail hazard map for Europe has been constructed and is found in good agreement with major regional and national scale climatological studies.


A review of 2015 CFS anomaly forecasts of precipitation and severe weather.

Greg Carbin

NOAA/NWS Storm Prediction Center

Presenter: Carbin, Greg

Output from version 2 of the Climate Forecast System (CFSv2) has been developed at the Storm Prediction Center to aid forecasters in anticipating severe weather events several days in advance. The post-processed CFSv2 output, made available through the CFS Severe Weather Guidance Dashboard (http://wxvu.net/spc/cfs_scp/), will be described as part of this presentation along with updated visualization techniques allowing forecasters to review past model performance with respect to daily severe weather activity. Preliminary severe weather verification measures for 2015 will also be reviewed. These measures show how initial attempts at daily severe weather prediction, beyond about 10 days, have shown little skill over climatology. Nonetheless, when a number of CFSv2 forecasts from different runs are combined into an ensemble, some indications suggest useful guidance ahead of highly anomalous events may still be attainable. An example will be shown using the anomalous precipitation forecast produced by the CFSv2 valid during the late spring of 2015. A probability-matched mean of quantitative precipitation for the month of May, derived from CFSv2 forecasts generated in the last days of April, will be verified by the observations of highly anomalous precipitation, occurring primarily across the U.S. Southern Great Plains. While the placement of the forecast anomaly was of limited value in this case, the CFSv2 ensemble precipitation forecast for May provided a strong anomaly signal of similar magnitude to the observed precipitation anomaly. We will discuss how this same approach could be applied to the forecast of severe weather activity in the three-to-four-week time frame.


 

Challenges in Severe Convective Storm Prediction for the Coastal-Urban New York City-Long Island Region on All Time Scales

Brian A. Colle, Kelly A. Lombardo (UConn-Avery Point), and Harrison Li (Harvard U.)

Stony Brook University

Presenter: Colle, Brian

There are well-known environmental parameters (e.g., large instability, moisture, lift, vertical shear) and spatial patterns (e.g., drylines, cyclone warm sector, jet streaks, etc…) that favor severe convective storms, which have been useful for short-term forecasting and determining potential future storm changes from climate model output. For the urban-coastal region of New York City (NYC) and Long Island (LI), the upstream terrain, coastal marine layer, sea breeze circulations, and land-surface variability strongly modify convective storms. This mesoscale variability makes now-casting severe storm development in this region difficult, and it poses a challenge for determining longer-term trends/prediction from climate models or downscaling approaches.

This presentation will first highlight some of the spatial patterns that favor severe convection (e.g., tornadoes) in the NYC-LI region. Nearly half (18 of 34 events from 1950 to 2010) of NYC–LI tornadoes developed between 0500 and 1300 EDT, and August is the peak tornado month as compared to July for most of the northeastern United States. A spatial composite highlights the approaching midlevel trough, moderate most unstable convective available potential energy (MUCAPE), and frontogenesis along a low-level baroclinic zone. The shorter-term prediction challenges are highlighted from two NYC-LI tornado events early in the morning 8 August 2007 and late afternoon 16 September 2010. During the 2007 event, a mesoscale convective system intensified in the lee of the Appalachians in a region of low-level frontogenesis and moderate MUCAPE (~1500 J/kg). Warm advection at low levels and evaporative cooling within an elevated mixed layer (EML) ahead of the mesoscale convective system (MCS) helped steepen the low-level lapse rates. Meanwhile, a surface mesolow along a quasi-stationary frontal zone enhanced the warm advection and low-level shear. The late afternoon event on 16 September 2010 was characterized by a quasi-linear convective system (QLCS) that also featured an EML aloft, a surface mesolow just west of NYC, low-level frontogenesis, and a southerly low-level jet ahead of an approaching midlevel trough. These cases also highlight the benefits of additional radars (Terminal Doppler Weather Radars at JFK and EWR airports), real-time mesonet surface observations, and meteorological data from commercial aircraft profiles.

On longer time scales, it is hypothesized that relatively large biases in climate models make future climate prediction of severe storms over the Northeast U.S difficult. Using several models from the Coupled Model Intercomparison Project (CMIP5), there is very large inter-model spread and biases associated with various environmental storm parameters (lapse rate, integrated water, large-scale vertical motion), but there is a fairly robust upward trend in integrated moisture (and therefore CAPE) in all models, thus providing some confidence for at least increased convective storm frequency over the Northeast U.S. during the later 21st century using linear discriminant analysis.


 

Seasonal Prediction of Tornadoes Using a Space-Time Statistical Model

James B. Elsner and Thomas H. Jagger

Florida State University

Presenter: Elsner, James

The destructive impact tornadoes have on communities has sparked interest in predicting the risk of impacts on seasonal time scales. Here the authors demonstrate how to build statistical models for predicting tornado rates. They test the models with tornado counts accumulated over a 45-year period aggregated to counties in the State of Oklahoma and to cells in a latitude/longitude grid across a large portion of the south-central United States. The spatial model provides a fit to the counts that includes terms for the spatial correlation and the population effect. A space-time model provides a similar fit to annual counts but also includes a term for a time varying climate factor. This work contributes to methods for forecasting severe convective storms on the seasonal time scale.


 

Seasonal Prediction of Lightning Activity in North Western Venezuela: Large-Scale versus Local Drivers

Á.G. Muñoz, J. Díaz-Lobatón, X. Chourio, J. Stock

NOAA, IRI-Columbia U, Zulia U

Presenter: Munoz, Angel G.

The Lake Maracaibo Basin in North Western Venezuela has the highest annual lightning rate of any place in the world (∼ 200 flashes km−2 yr−1), whose electrical discharges occasionally impact human and animal lives (e.g., cattle) and frequently affect economic activities like oil and natural gas exploitation. Lightning activity is so common in this region that it has a proper name: Catatumbo Lightnings. Although short-term lightning forecasts are now common in different parts of the world, to the best of the authors’ knowledge, seasonal prediction of lightning activity is still non-existent. This research discusses the relative role of both large-scale and local climate drivers as modulators of light- ning activity in the region, and presents a formal predictability study at seasonal scale.

Analysis of the Catatumbo Lightning Regional Mode, defined in terms of the second Empirical Orthogonal Function of monthly Lightning Imaging Sensor (LIS-TRMM) and Optical Transient Detector (OTD) satellite data for North Western South America, permits identification of potential predictors at seasonal scale via a Canonical Correlation Analysis. Lightning activity in North Western Venezuela responds to well defined Sea Surface Temperature patterns (e.g., El Niño-Southern Oscillation, Atlantic Meridional Mode) and changes in the low-level meridional wind field that are associated with the Inter Tropical Convergence Zone migrations, the Caribbean Low Level Jet and tropical cyclone activity, but it is also linked to local drivers like local convection triggered by the topographic configuration and the effect of the Maracaibo Basin Nocturnal Low Level Jet. The analysis indicates that at seasonal scale the relative contribution of the large-scale drivers is more important than the local (basin-wide) ones, due to the synoptic control imposed by the former. Furthermore, meridional CAPE transport at 925 mb is identified as the best potential predictor for lightning activity in the Lake Maracaibo Basin.

It is found that the predictive skill is slightly higher for the minimum lightning season (Jan-Feb) than for the maximum one (Sep-Oct), but that in general the skill is high enough to be useful for decision making processes related to human safety, oil and natural gas exploitation, energy and food security.


 

POSTER ABSTRACTS

An analytical scaling for peak CAPE values in continental midlatitudes

Vince Agard, Kerry Emanuel

MIT

Presenter: Agard, Vince

A simple model for the development of a moist boundary layer in a dry adiabatic environment shows that the peak value of convective available potential energy (CAPE) scales with the difference between surface dry- and wet-bulb temperatures in an idealization of typical continental midlatitude conditions. The canonical setup for continental severe convection is approximated by considering a column of dry adiabatic air which becomes situated over a moist surface. Analytical solutions are mathematically derived in this framework for the evolution of the height, dry static energy, and moist static energy of the resulting surface boundary layer. From these solutions, it is shown that the maximum values of CAPE and convective inhibition (CIN) each scale with the initial wet-bulb depression of air in the surface boundary layer. This result indicates that peak CAPE values will follow the Clausius-Clapeyron relation over a given range of surface temperatures.


 

Predicting the climatology of tornado occurrences in North America with a Bayesian hierarchical modelling framework

Vincent Cheng, George Arhonditsis, David Sills, William Gough, Heather Auld

University of Toronto

Presenter: Cheng, Vincent

Destruction and fatalities from recent tornado outbreaks in North America have raised considerable concerns regarding their climatic and geographic variability. However, regional characterization of tornado activity in relation to large-scale climatic processes remains highly uncertain. Here, we develop a novel Bayesian hierarchical framework for elucidating the spatiotemporal variability of the factors underlying tornado occurrence in North America. We demonstrate that regional variability of tornado activity can be characterized using a hierarchical parameterization of convective available potential energy, storm relative helicity, and vertical wind shear quantities. We show that the spatial variability of tornado occurrence during the warm summer season can be explained by convective available potential energy and storm relative helicity alone, while vertical wind shear is clearly better at capturing the spatial variability of the cool season tornado activity. Our results suggest that the Bayesian hierarchical modelling approach is effective for understanding the regional tornado environment and in forming the basis for establishing tornado prognostic tools in North America.


 

Spatial Redistribution of Tornadoes in a Warming Climate

Samuel Childs

Colorado State University

Presenter: Childs, Samuel

The impacts of climate change on the occurrence, frequency, and intensity of several weather phenomena such as heat waves, droughts, and floods, have been the focus of many studies in recent years, with relatively consistent conclusions. The same cannot be said for climate change impacts on USA tornado statistics, as fewer robust analyses and conflicting results about changing tornado environments preclude any definitive relationships. This study aims to add to the current state of knowledge by assessing possible spatial shifts in tornado distribution over the duration of the modern tornado record. Seasonal analysis is also presented, with a particular focus on cold-season (DJF) tornado events. Using warming temperatures as motivation for driving potential changes in tornado spatial distribution, due to the effects of changing temperatures on meteorological parameters known to promote tornadic activity, two consecutive 30-year periods are defined as a Cold Period I (1954-1983) and Warm Period II (1984-2013). Annual tornado counts and days for the two periods are compared across a gridded domain encompassing (30-50°N, 80-105°W) to search for any changes in the location of relative maxima of tornado occurrence. A noteworthy shift from the south-central Great Plains (KS, OK, TX) toward the Mississippi Valley region (TN, AL, MS) is seen and shown to be statistically significant. Subsequently, results are apportioned by season to potentially discern which times of the year are contributing to the overall spatial shift in tornado activity. Of particular interest are the increasing counts across the Mississippi Valley region during the cold season. Tornadoes occurring during this time of year can pose an enhanced societal risk, with the general public perception that tornadoes are not of major importance or influence during winter months. Recent events such as the deadly late-December 2015 tornadoes across the Mid-South add further support for active cold season periods across the Mississippi Valley region and also raise questions as to any ENSO connection, with December 2015 being one of the warmest Decembers on record in the midst of a very strong El Nino. The need exists for heightened public awareness in order to prevent injury and death in these cold-season situations.


 

Investigation of Convective Storms over Long Island during the Doppler Radar for Education and Mesoscale Studies (DREAMS) Project

Brian A. Colle, Kelly A. Lombardo (UConn-Avery Pt.), Sara Ganetis, Matthew Sienkiewicz, Michael Colbert (Penn State U.), and Breanna Zavadoff

Stony Brook University

Presenter: Colle, Brian

Understanding the complex interaction of severe storms with the coastal southern New England coast requires high resolution and temporal field data. From 17 June – 8 July 2013 the Doppler Radar for Education and Mesoscale Studies (DREAMS) project was completed on Long Island, NY. DREAMS was a collaborative project between Stony Brook University, the National Weather Service (NWS) located in Upton, NY, and the Center for Severe Weather Research (CSWR) located in Boulder, CO. The goals of the project were: (1) to educate students on local mesoscale weather phenomena while gaining experience utilizing state of art research radar equipment and data (2) to obtain high temporal and spatial resolution data sets of local mesoscale weather phenomena, including severe thunderstorms interacting with the marine environment and sea breezes, and (3) to expose the broader public to the latest atmospheric research including many of the challenges encountered when forecasting the local weather. This field study provided an opportunity to obtain valuable data sets of local convection using high resolution Doppler radar technology (i.e. Doppler on Wheels) together with in situ observations. This poster will highlight the 24 June 2013 convective storm event, in which convective initiation (1645 UTC) occurred along a sea breeze boundary across western Long Island. During the morning hours, the CIN rapidly decreased from 200 J/kg to 5 J/K, while the CAPE increased rapidly from 450 J/kg to 2500 J/kg. Convective initiation occurred along the largest convergence with the sea breeze front over western Long Island. Doppler radar illustrated the pulse nature of these storms in a relatively weak shear environment, which will be highlighted in a series of cross sections. The storms rapidly weakened as they moved eastward over central and eastern Long Island during the early afternoon, where the instability was 500-1000 J/kg and the sea breeze boundary was less defined.


 

A method to automate tornado casualty maps

Tyler Fricker and James B. Elsner

Florida State University

Presenter: Fricker, Tyler

On average, over a thousand people are directly injured or killed by tornadoes in the U.S. each year. Casualty statistics are well documented at the level of individual tornadoes (Brooks and Doswell 2002), but less is known about the spatial variation in these values (Ashley 2007). Here, I explain a method to process the available data on casualties from separate tornado records and automatically generate maps to show tornado casualties (deaths and injuries) spatially. The method is based on the proportional allocation of path area and population in areal units tessellating a chosen region. I begin by applying the method to a latitude/longitude grid at a quarter degree resolution covering the most tornado-prone sector of the country. The method is validated using county-level casualty data from six states within the study area. The intention is to highlight areas that experience a high level of tornado casualties and to identify vulnerable populations from tornado activity.

Work Cited:

Ashley, W., 2007: Spatial and temporal analysis of tornado fatalities in the United States: 1880-2005. Weather and Forecasting, 22, 1214–1228.

Brooks, H. E., and C. A. Doswell, 2002: Deaths in the 3 May 1999 Oklahoma City tornado from a historical perspective. Weather and Forecasting, 17, 354–361.


 

Extreme Rainfall and Severe Weather on the Periphery of Continental Anticyclones

Thomas J. Galarneau

University of Arizona

Presenter: Galarneau, Thomas

Warm season continental anticyclones can be associated with extreme heat waves and drought. On the periphery of these anticyclones, subsynoptic-scale cyclonic troughs and cutoff circulations can provide corridors of enhanced vertical wind shear, instability, and forcing for ascent. Extreme rainfall and severe weather can occur in conjunction with these cyclonic disturbances. Cyclonic disturbances on the periphery of continental anticyclones were first recognized by Namias (1936) in his study of coherent moisture plumes and regions of isentropic ascent that moved around the periphery of synoptic-scale anticyclones over the central U.S. More recent (in the last 20 years) heat-wave-producing anticyclones that also had extreme rain and severe weather events on their periphery include the July 1995 event over the central U.S., the February 2004 event over Australia, and the August 2010 event over eastern Europe and west-central Asia. These types of events (i) pose a significant forecast challenge as several “threats” – extreme heat, flooding rains, and severe thunderstorms – accompany a single synoptic-scale weather system, and (ii) are important in the subseasonal to interannual variability of rainfall and severe weather.

The aim of this presentation is to document the structure and evolution heat-wave-producing continental anticyclones over the U.S. in July 1995, Australia in February 2004, and eastern Europe and west-central Asia in August 2010. Subsynoptic-scale disturbances were located on the periphery of these three anticyclones, but produced different sensible weather and societal impacts. These disturbances produced severe organized mesoscale convective systems during the July 1995 event and historic rainfall and flooding for the August 2010 event. In contrast, the disturbances associated with the February 2004 event over Australia produced low-to-moderate rainfall with relatively minor societal impact. In addition to comparing these three events with wide-ranging impacts, we will examine a preliminary analysis of the subseasonal to interannual variability in the position and structure of continental anticyclones and their peripheral subsynoptic-scale cyclonic disturbances over the U.S., including possible linkages between this variability and large-scale teleconnection indices.


Climate Change and Hazardous Convective Weather in the United States: Insights from High-Resolution Dynamical downscaling

Kimberly Hoogewind, Jeff Trapp, Mike Baldwin

Purdue University

Presenter: Hoogewind, Kimberly

Hazardous convective weather (HCW)— large hail, strong surface wind gusts, and tornadoes—present a significant risk to life and property every year in the United States. Amidst continued warming of global mean temperature, there is uncertainty regarding the response of these extreme events in a future climate. Several studies in recent years have shown that global climate model (GCM) projections of favorable severe weather environments increase through the 21st century. However, this approach is undoubtedly limited by the assumption that the convective-scale phenomena will be realized within these environments; the resolution of GCMs remain much too coarse to adequately represent the scales at which severe weather phenomena occur, including processes that may lead to the initiation of convective storms. To better address this question, high-resolution (4-km), dynamically downscaled WRF simulations have been performed using initial and boundary conditions from a high-performing GCM for historical (1971-2000) and future (2071-2100) periods. This work will discuss potential impacts upon the frequency, intensity, and spatiotemporal distributions of HCW under an aggressive climate scenario. Further discussion will be devoted to the linkage between GCM environments and ensuing events in the downscaled simulations.


 

Can we predict seasonal changes in high impact weather in the United States?

Eunsil Jung and Ben P. Kirtman

University of Miami

Presenter: Jung, Eunsil

High-impact weather events, such as tornadoes, threaten lives and cost billions of dollars in property damage throughout the US every year. The frequency of tornado events has increased in recent years, and it is expected to continue to rise in the warming climate scenarios, suggesting that any predictive capability is of great social benefit. While predicting the individual tornado events is only possible a few hours in advance, the large-scale background atmospheric conditions that influence the likelihood of tornado events may be more likely predictable on longer time scales. Here we use the NCAR Community Climate System Model version 4.0 (CCSM4) forecasts and North American Regional Reanalysis (NARR) for the period of 1982-2011 to examine whether we can predict the seasonal changes in the likelihood of a tornado event in the US. El Niño– Southern Oscillation (ENSO) is known to influence tornado activity in the US during early spring, this study, however, highlights that the influence of ENSO on US tornado activity is weak during May-July (MJJ). Instead, warm water in the Gulf of Mexico is a potential predictor of forecasting US tornado activity during MJJ. Considering our current ability to predict SST in the Gulf of Mexico, compared with the difficulty of predicting the seasonal outlook of tornado activity in the US, the findings provide evidence for the seasonal prediction of high-impact weather in the US.


 

Importance-ranking of Climate Variables for Prediction of Damaging Straight-line Winds

Ryan Lagerquist, Amy McGovern, Travis Smith, Valliappa Lakshmanan, Michael Richman

Univ of Oklahoma / CIMMS

Presenter: Lagerquist, Ryan

In the past decade there has been explosive growth in the use of machine learning (ML) to predict thunderstorm hazards such as hail, aircraft turbulence, and tornadoes. However, relatively few researchers have focused on the threat from severe non-tornadic (or “straight-line”) winds, which occur much more frequently and are capable of causing great damage. We have developed machine-learning models that predict the occurrence of severe straight-line winds at lead times up to one hour. Our main objective is to have these models adopted by operational forecasters.

Three types of data are used for this project: archived radar grids from the Multi-year Reanalysis for Remotely Sensed Storms (MYRORSS); proxy soundings from the North American Regional Reanalysis (NARR); and surface wind observations from the Meteorological Assimilation Data Ingest System (MADIS), Oklahoma Mesonet, and one-minute METARs. The domain and time period are the continental United States from 2004-11 (excluding 2009). There are 306 days used for training, validation, and testing (all those with at least 100 unfiltered severe wind reports from the Storm Prediction Center).

First, storm cells are identified and tracked using w2segmotionll and w2besttrack, both of which are algorithms in the Warning Decision Support System with Integrated Information (WDSS-II). A “storm cell” is one thunderstorm at one time step; a “storm track” is the trajectory followed by a thunderstorm through time. Next, surface wind observations must be attributed to storm cells. Each wind observation is linked to the nearest storm cell, or none if there is no storm cell within 10 km. Then four types of features are calculated for each storm cell: statistics describing the shape of the bounding polygon; statistics for radar fields inside the bounding polygon; basic storm information (e.g., area, speed, direction of motion); and sounding indices. Sounding indices are calculated from interpolated NARR data by the SHARPpy software. Overall, 565 features are calculated and used as ML predictors.

Due to the large number of features, our preferred ML algorithms are random forests and gradient-boosted regression trees. The dependent variable is the 90th -percentile wind produced by each storm cell with a given lead time (from 15-60 minutes) and buffer distance (from 0-10 km). We have achieved good results for classification (cutoffs of 30 and 50 kt).

Several methods are used to rank feature importance in the best models, including J-measure ranking, sequential forward/backward selection, principal-component analysis, and permutation selection. We will present the most important features and link these, where possible, to climate change. For example, all methods rank low- to mid-level lapse rates among the most important predictors of straight-line winds. Straight-line winds are positively correlated with low- to mid-level lapse rates, which most climate models agree will decrease with global warming. This relationship alone suggest that the straight-line wind threat may decrease with global warming.

Although our main objective is operational forecasting, feature-ranking allows physical relationships to be inferred from ML models. We hope that knowledge presented here will be used to further the state of both thunderstorm and climate science.


 

High Resolution Simulations of an Extreme Precipitation Event over Long Island on 13 August 2014

Nicholas Leonardo and Brian A. Colle

Stony Brook University

Presenter: Leonardo, Nicholas

On 13 August 2014, the heavily-populated suburbs of Central Long Island were caught off guard by a historical flood. Within a 4 hour time period, almost 12” (~305 mm) of rainfall accumulated in an intense band over Suffolk County, with Islip MacArthur Airport receiving a record-breaking 24-h total of 13.57” (345 mm). While heavy rain was forecasted for the Northeast the day before, the amplitude of this event was severely underpredicted by all of the operational models. Extreme urban rainfall events like this are expected to increase in frequency in relation to climate change. Hence, it is important to understand the mechanisms behind their development and factors affecting their predictability.

This study seeks to analyze the mesoscale evolution of this flooding event, and the key mechanisms behind it. Another objective is to determine whether a mesoscale model run at high resolution can realistically reproduce the development and intensity of the rainfall, and to explore some of the sources of model uncertainty limiting its predictability. The Weather Research and Forecasting (WRF v3.5.1) model was used to simulate this event down to 1-km grid spacing for an 18-h prediction starting at 0000 UTC 13 August 2014. Those WRF simulations initialized using the 0.5-degree Global Forecast System (GFS) analysis and 6-hourly forecast grids produced the most realistic predictions for this event. An ensemble of different model physics were also tested, but the control (best) run utilized the WSM6 microphysics, RRTM longwave radiation, Dudhia shortwave radiation, MYNN2.5 PBL, and KF cumulus schemes (for grid spacing > 5 km).

This realistic WRF member produced a narrow band of rainfall amounts in excess of 300 mm very close to the correct location. We will highlight some of the low-level forcing and vertical circulations associated with the rainfall in this successful run and compare them with observations. Both radar and surface observations, as well as 1-km WRF simulations, illustrate a weak meso-low near Long Island that enhanced the low-level convergence, upward motion, and rain rates over this region. The mean storm motion was parallel to the orientation of the front, resulting in cells “training” over the same location for a few hours. Sensitivity runs illustrate the importance of latent heating in the enhancement of the mesoscale trough boundary associated with the rainfall maximum. There was a weak coastal front during this event, but removing Long Island did not have much impact on the precipitation totals. Many WRF members using different initial conditions and physics underpredict the precipitation by a factor of two. We will show that the size of the nested domain has a significant impact on the evolution of the precipitation and the subsequent evolution of the intense precipitation band. A larger explicit precipitation domain (3-km domain) results in more spurious model convection forming to the south along the warm front during the first 6 hours of the simulation. This in turn perturbs (weakens) the low-level jet transporting moisture and instability to the location where cells initiate and grow over Long Island.


The Role of Initial Cloud Condensation Nuclei Concentration in Hail Using the WRF Idealized Simulation

Xiaofei Li,Qinghong Zhang and Huiwen Xue

Peking University

Presenter: Li, Xiaofei

The effects of the initial cloud condensation nuclei (CCN) concentrations (from 100 mg-1 to 3000 mg-1) on hail have been investigated in an idealized supercell experiment of WRF model with NSSL 2-moment microphysics scheme. The initial CCN concentration has obvious non-monotonic effects on the mixing ratio, number concentrations and radius of hail both in cloud and at the surface with a threshold CCN concentration between 300 mg-1 and 500 mg-1. The increasing CCN concentration is conductive (suppressive) to the amounts of surface hail precipitation below (above) the threshold. The reason for the non-monotonic effects is from both the thermodynamics and microphysics. Below the threshold, the mixing ratios of cloud droplets and ice crystals increase dramatically with CCN, resulting in more latent heat released from vapor condensation and intensified updrafts in cloud. The riming process, which is the primary process for hail production, increases dramatically. Above the threshold, the mixing ratio of cloud droplets and ice crystals increase continuously but the maximum updraft is weakened because of the reduced latent heating, which is related to the reduced riming rate in the storm core area; the even smaller ice crystals reduce the formation of hail and the even smaller cloud and less rain water weaken the riming efficiency so that graupel and hail also decrease with CCN, which is unfavorable for hail growth.


 

Hailstorms over Switzerland in a warmer climate: a surrogated climate change experiment

Andrey Martynov, Olivia Martius, Luca Nisi

University of Bern

Presenter: Martynov, Andrey

The WRF model was used for simulating the hailstorms over Switzerland in summer periods of 2012-2015, with 2-km horizontal resolution and using the ECMWF analysis for boundary forcing. The climate conditions, expected towards the end of the XXI century (RCP85 scenario) were simulated by introducing biases of air temperature and of soil temperature/water content, based on CMIP5 simulations. The relative air humidity was left constant.

Based on the comparison with the unbiased present day simulation, changes in spatial and temporal distributions of precipitation patterns and of hailstorms, in characteristics of thunderstorm clouds and in parameters of ground-reaching hailstones were studied. Considerable increase of mean and maximal precipitation rate has been remarked. More hail days, more frequent hailstorms and larger hailstones were produced in the climate change simulation.


 

A Climatology of Synoptic-Scale Atmospheric Conditions Associated with January-April Tornado Outbreaks

Ashton Robinson Cook

NOAA NWS Storm Prediction Center

Presenter: Robinson Cook, Ashton

A 61-year climatology (1950-2010) of synoptic-scale atmospheric conditions associated with January-April tornado outbreaks was created to 1) identify dominant patterns associated with tornado outbreaks and 2) assess potential influences of larger-scale climate system components (i.e. ENSO) on tornado outbreaks. This climatology was created via the use of S-mode Principal Component Analysis to identify groups of outbreaks with similar spatial anomalies in geopotential height (300 hPa and 850 hPa), lifted index, and precipitable water from the NCEP/NCAR Reanalysis Dataset. These data were collected on tornado outbreak days, or days of 6 or more tornadoes in a 24-hour period across the continental U.S.

Results from this analysis suggest that geopotential height, static instability, and moisture patterns each influence the location of tornadoes in outbreaks. Recurring geopotential height patterns for tornado outbreaks are identified across all four months of the period of study, although a few patterns appear to be specific to a particular month of outbreak occurrence. Furthermore, larger and more impactful outbreaks tend to occur in geopotential height patterns characterized by a geopotential height trough centered across the southwestern U.S. and central Great Plains, while weaker tornado outbreaks tend to occur when geopotential height troughs are centered across the eastern U.S. Future work will include the application of these results to climate models to assess the U.S. tornado climatology in varied climate scenarios.


Recent changes in the contiguity of extreme precipitation over the US

Danielle Touma, Noah S. Diffenbaugh

Stanford University

Presenter: Touma, Danielle

The characteristics of an extreme precipitation event, including the type, intensity, duration and spatial extent, can govern the hydrologic response of a watershed. In some cases, extreme precipitation can cause catastrophic floods resulting in humanitarian, economic and agricultural disasters. In this study, we examine the spatial extent of daily extreme precipitation in the US by defining its “footprint”: a contiguous area of rainfall exceeding a certain threshold (e.g., 90th percentile). Using the ORNL DAAC Daymet 1-km gridded dataset, we find that both large and small footprint sizes have significantly increased during the winter and spring months in the Midwest and significantly decreased in the summer and autumn months in the Northwest and Southwest. There are few significant changes in the total number of footprints in the eastern US, with the exception of some significant increases in June rainfall in the Northeast. Though the total number of footprints may be unchanging in the Southeast, there are significant changes in the distribution of the size of these footprints. Over all months in the year, there is increasing contiguity of extreme rainfall events, with significant decreases in 1 km2 footprints and increases in 10-1000 km2 footprints. Similarly, the Midwest also shows significant decreases in 1 km2 footprints and increases in 10-1000 km2 footprints throughout the year. Conversely, the Northwest shows significant decreases in medium sized footprints and increases in 1 km2 ones. To understand the roles of large scale atmospheric conditions contributing to these changes, we compare the NCEP NARR ¼ -degree geopotential heights, CAPE, precipitable water and winds associated with the footprint distributions for the earlier and later periods. In particular, we assess the increases in atmospheric conditions favorable to the increasing or decreasing contiguity of extreme precipitation in the recent past. By disentangling the effects of changes in various atmospheric variables on the footprint size, this study will further our understanding of the anthropogenic fingerprint in observed changes of extreme precipitation characteristics.


The Effects of Antecedent Soil Moisture Anomalies on Tornado Activity in the United States

Ryann Wakefield, Esther Mullens, Derek Rosendahl, Harold Brooks

Rutgers University

Presenter: Wakefield, Ryann

Recently, there has been increased interest in the ability to forecast severe weather events on a seasonal scale. Being able to forecast events such as the April 2011 tornado outbreak could have beneficial impacts such as increased preparedness at the Federal and local level as well as greater public awareness. To better forecast tornadoes on a seasonal scale, we must first look at the underlying factors that influence the inter-annual variability of both tornado locations and intensity. Soil moisture has been shown, on regional scales, to have an effect on moisture within the boundary layer, and therefore, it has the potential to impact deep convection. Previous studies have examined relationships between factors related to soil moisture such as precipitation and evapotranspiration and their effects on tornado climatology at the local and regional scale. This study examined the relationship between antecedent soil moisture and tornado activity in five regions within the United States east of the Rocky Mountains, using two approaches. The first approach used fall and winter soil moisture anomalies as a predictor for spring tornado activity. The second looked at the six months preceding each month in the year. Statistically significant correlations between tornado days and soil moisture anomalies were found in the Northern Plains, the Southeast and Oklahoma, indicating that regionally, soil moisture may have an effect on seasonal tornado activity. In addition, we also assessed the reliability of our modeled soil moisture dataset by comparing it to the high-resolution Oklahoma Mesonet observational network. It was found that the Climate Prediction Center’s soil moisture dataset was significantly correlated at a 95% confidence level or greater, in each grid box, where there was Mesonet data available for comparison. This suggests that the soil moisture dataset used for our analysis models reality well.


Statistical Predictions of Seasonal Hail/Tornado Activity

Hui Wang, Kirstin Harnos, Arun Kumar

NOAA

Presenter: Wang, Hui

A method for seasonal predictions of March–June hail/tornado activity over the U.S. was developed with the singular value decomposition (SVD) and using January sea surface temperature (SST) as a predictor. The SVD analysis objectively selects distinctive lagged relationships between springtime hail/tornado activity and different modes of winter SST, including global SST warming trend, El Nino/La Nina, and the PDO-like pattern. The forecast model was cross-validated over the 1955–2014 period, indicating a certain degree of predictability of seasonal hail and tornado activity over the central and eastern U.S. Potential for applying this method for real-time prediction will be discussed.


 

Examining U.S. Tornado Vulnerability Using the NWS Damage Assessment Toolkit

Holly M. Widen and James B. Elsner

Florida State University

Presenter: Widen, Holly

Understanding tornado vulnerability is critical in making efforts to decrease the threat to life and property in communities across the country, especially as our population expands. Advancements in technology have not only improved tornado forecasts and warnings, but also the collection of tornado damage survey data. The National Weather Service (NWS) Damage Assessment Toolkit (DAT) contains the most extensive GIS-based damage survey data available to the public but has yet to be utilized as a means to assess tornado vulnerability on a regional scale. This research examines physical and social factors that contribute to the assessment of tornado vulnerability in the United States. Correlation and regression analyses are performed to examine relationships between tornado casualties and the physical and social factors. This study is important for future research on tornadoes, as the DAT will continue to expand and may later be used for climatological studies. Thus, this work is relevant to forecasters and climatologists. The results highlight specific areas of high physical and social vulnerability as well as significant factors contributing to casualties in past events. These findings are important to insurance and disaster management agencies for preparedness and mitigation.


 

Dual-Doppler and polarimetric observations of two tornadic supercells in central Oklahoma on 19 May 2013

Zachary B. Wienhoff, H. B. Bluestein, J. Houser, J. C. Snyder, A. Shapiro, and C. K. Potvin

University of Oklahoma

Presenter: Wienhoff, Zachary B.

On 19 May 2013, two supercells moved through central Oklahoma and produced seven tornadoes, including one of EF-3 and one of EF-4 intensity. High temporal and spatial resolution observations were collected in both strong tornadoes by a mobile, rapid-scan, X-band, polarimetric radar (RaXPol). In conjunction with the nearby WSR-88D Twin Lakes radar, dual- Doppler analyses were synthesized to produce the three-dimensional wind field and vertical vorticity throughout the tornadoes’ lifecycles. In order to account for significant differences in the temporal resolutions of the two radars, a reflectivity-tracking scheme was employed to interpolate linearly (in a Lagrangian sense) between two radar volumes through a variational algorithm. These analyses are unique because they are one of the first, to the authors’ knowledge, to make use of data from both an operational, S-band radar and a mobile, rapid-scan, X-band radar.

This work will focus on the methodology used to combine both RaXPol and KTLX datasets, including discussion regarding the strengths and weaknesses of this technique. Analyses will be presented on the both the Edmond-Carney and Norman-Shawnee tornadoes, examining the characteristics of the wind field associated with the intensification, maturity, and demise of the tornado. The 3-D wind fields of the tornado and mesocyclone will be examined to see how they evolved in time, and especially how vertical vorticity and vertical velocity changed as the tornado intensified. Differential reflectivity (ZDR) from the WSR-88D will be compared to the vertical velocity field estimated from dual-Doppler analyses to examine how the characteristics of a ZDR column varied with respect to changes in the updraft intensity. Results from both cases will be compared, and some thoughts on future improvement to this technique will be presented.


Trends in hail and severe storms in Tibetan Plateau during 1980 to 2013

Tian Zou, Qinghong Zhang

Peking University

Presenter: Zou, Tian

The long-term trend of hail in Tibetan Plateau (TP) is essential for us to understand the impact of global climate change on local weather extremes. Based on the surface observations over TP during the warm season (from June to September) from 1980 to 2013, it is found that the occurrence of severe storm is decreasing, with 0.82 fewer hail days and 5.8 fewer storm days per decade. Meanwhile, it is statistically significant that the percentage of hail days versus storm days has reduced by 2.6% per decade. Features of several atmospheric fields in Naqu, one of the highest frequency areas of hail, which are favorable to the development of severe storm during the same period have been analyzed. They are freezing level height (FLH), vertical wind shear (VWS), precipitable water (PW) and modified K index (MKI). The results show that the decreasing of severe storms is relevant to the missing of days with MKI larger than 14.5K and PW larger than 1.87cm, which are the first quartile of MKI and PW, respectively. And the number of hail days drops as a consequence of significantly higher FLH and weaker VWS.


Impacts of the Gulf of Mexico on Severe Thunderstorm Activity

Maria Timmer, Reed Timmer, John Allen

Columbia University

Presenter: Timmer, Maria

The Gulf of Mexico (GOM) is the primary source for boundary layer moisture over the severe thunderstorm prone Great Plains of the United States. However, its correlation to the El Niño Southern Oscillation (ENSO) and other climate factors poses challenges for attributing the influence of the GOM on these events. In an effort to isolate and detect the effects of the GOM on severe thunderstorm occurrence, ENSO-neutral years and particular GOM climate zones were considered in this study. Predominantly, the warmer (cooler) the GOM sea-surface temperatures (SSTs), the more (less) hail and tornado events occur both during the spring (March-May) and summer (June-August) across areas east of the Rockies. These results imply that GOM regions may provide predictability of seasonal severe weather independently of ENSO. Correspondingly, this suggests that ONI and GOM should both be considered in creation of seasonal severe weather forecasts, as both offer value as predictands since SSTs are slow to change from month to month.


The Tornado Climatology of Australia 1795-2014

John Allen, Edwina Allen

Columbia University

Presenter: Allen, John

A new climatology for the occurrence of tornadoes in Australia has been developed for the period 1795 to 2014, the second largest single country record. However, extensive media coverage in 2013 raised the question ‘Was 2013 a record tornado year?’ Like many places outside of the United States, the historical records for tornadoes are poorly documented. Existing data from the Australian Bureau of Meteorology National Severe Storms Archive also suffer from observer-driven spatial limitations, and biases related to institutional policy of event documentation. Recently, extensive library archives of scanned newspapers, and digitization of the original severe thunderstorm reports material have become available for Australia that can offer insight into historical events and extend the existing climatology.

Keyword optimization was used to identify tornadoes from the scanned data while reflecting changes to terms used in the historical vernacular. Additional metadata relating to intensity, time of occurrence, path characteristics, injuries, fatalities and damage were inferred from newspaper accounts. Further, tornadoes from the existing Severe Storms Archive were cross-validated and additional metadata determined for inclusion in the new climatology. Based on documentary evidence, tornadoes were rated via the Fujita scale using three categorizations to reflect uncertainty in historical strength determination (Weak F0-F1, Strong F2-F3 and Violent F4-F5). The quality of record for each identified event was categorized into three levels (Possible, Likely or Definite) based on the reliability of observations, as well as documentation of characteristics indicating the presence of a tornadic event.

The climatology in context of a recent observed year (2013) will be presented, highlighting that the annual frequency of tornadoes in Australia ranges between 30 and 80 observed tornado events per year but likely underestimates the total frequency given underreporting due to population density. Numerous tornado outbreak cases have also been identified throughout the length of the record. To further illustrate the risk posed for Australia by tornadoes, cases from 2013 encompassing the broad spectrum of tornado formative environments will be discussed. These results reveal that Australia is subject to tornadoes from most environmental sources on a relatively frequent basis, and this should play a greater role in the forecasting and warning process.