Diagnosing EF Scale Potential Using Conditional Probabilities Adapted from material and images provided by Bryan Smith, Rich Thompson, Andy Dean, Dr. Patrick.

Slides:



Advertisements
Similar presentations
Radar Climatology of Tornadoes in High Shear, Low CAPE Environments in the Mid-Atlantic and Southeast Jason Davis Matthew Parker North Carolina State University.
Advertisements

The 4 Sep 2011 Tornado in Eastern New York: An Example for Updating Tornado Warning Strategies Brian J. Frugis NOAA/NWS Albany, NY NROW XIII 2-3 November.
A Refresher on Super-Resolution Radar Data Audra Hennecke, Dave Beusterien.
Statistics 100 Lecture Set 6. Re-cap Last day, looked at a variety of plots For categorical variables, most useful plots were bar charts and pie charts.
THE IMPACTS OF THUNDERSTORM GEOMETRY AND WSR-88D BEAM CHARACTERISTICS ON DIAGNOSING SUPERCELL TORNADOES Steve Piltz – WFO Tulsa, OK Don Burgess – CIMSS.
Louisville, KY August 4, 2009 Flash Flood Frank Pereira NOAA/NWS/NCEP/Hydrometeorological Prediction Center.
Analysis of Rare Northeast Flow Events By Joshua Beilman and Stephanie Acito.
Severe Weather Radar Features. Weak Echo Region (WER) Region of low radar reflectivities on inflow side of storm o Near the surface High reflectivities.
 Definition of Spectrum Width (σ v ) › Spectrum width is a measure of the velocity dispersion within a sample volume or a measure of the variability.
SPC Convective Outlook Changes Changes in Category Names/Definitions Bill Sammler Warning Coordination Meteorologist National Weather Service, Wakefield.
A Study on Convective Modes Associated with Tornadoes in Central New York and Northeast Pennsylvania Timothy W. Humphrey 1 Michael Evans 2 1 Department.
6/26/2015 RUC Convective Parameters and Upscale Events in Southern Ontario Mike Leduc Environment Canada.
6/26/2015 Examination of the low level polarimetric radar parameters associated with the Aug southern Ontario tornadic supercells Mike Leduc Sudesh.
Tornado Detection Algorithm (TDA) By: Jeffrey Curtis and Jessica McLaughlin.
The 4 August 2004 Central Pennsylvania Severe Weather Event – Environmental and Topographical Influences on Storm Structure Evolution Joe Villani NOAA/NWS,
Determining Favorable Days for Summertime Severe Convection in the Deep South Chad Entremont NWS Jackson, MS.
NWS Central Region Overview and plans for 2013 Intro to IBW Project 2013.
IBW Project Introduction and Overview. IBW Project Goals of This Training Provide an overview on IBW rationale Provide guidelines on application of IBW.
© Craig Setzer and Al Pietrycha Supercell (mesocyclone) tornadoes: Supercell tornado environments Developed by Jon Davies – Private Meteorologist – Wichita,
Visual Displays of Data and Basic Descriptive Statistics
Chapter 2 Describing Data with Numerical Measurements General Objectives: Graphs are extremely useful for the visual description of a data set. However,
Printed by Investigating Rapid Storm Intensification Mechanisms Including the Role of Storm Mergers in the 22 May 2011 Joplin, MO.
Severe Weather Operations. Severe Weather Staffing (Positions in orange are minimum needed) Severe Weather Coordinator – oversees the operations of the.
National Weather Service Weather Forecast Office – Taunton, MA (BOX)
A Study on the Environments Associated with Significant Tornadoes Occurring Within the Warm Sector versus Those Occurring Along Boundaries Jonathan Garner.
Describing and Displaying Quantitative data. Summarizing continuous data Displaying continuous data Within-subject variability Presentation.
Improving the Forecasting of High Shear, Low CAPE Severe Weather Environments Keith Sherburn and Jason Davis Department of Marine, Earth, and Atmospheric.
TOP CWA Golf Ball Size Hail Study Bill Gargan WFO TOP, KS.
Chad Entremont Daniel Lamb NWS Jackson, MS
Forecast Parameters. CAPE Convective Available Potential Energy – obviously, positive buoyancy is helpful for producing convection –100 mb mixed layer.
The Ingredients Based Tornado Parameter Matt Onderlinde.
SPC Ensemble Applications: Current Status, Recent Developments, and Future Plans David Bright Storm Prediction Center Science Support Branch Norman, OK.
Using Ensemble Probability Forecasts And High Resolution Models To Identify Severe Weather Threats Josh Korotky NOAA/NWS, Pittsburgh, PA and Richard H.
March 14, 2001 Bow Echo in Southeast Texas – A Mid-Altitude Radial Convergence Case Study Paul Lewis II.
Updated Radar-Based Techniques for Tornado Warning Guidance in the Northeastern United States Brian J. Frugis & Thomas A. Wasula NOAA/NWS Albany, New York.
Chapter 5: Boxplots  Objective: To find the five-number summaries of data and create and analyze boxplots CHS Statistics.
© 2008 The MITRE Corporation. All rights reserved. F065-B Risk Management for TFM through Incremental Decision Making NBAA 61 ST Annual Meeting &
Quasi-Linear Convective System Tornado Warnings
An Examination of “Parallel” and “Transition” Severe Weather/Flash Flood Events Kyle J. Pallozzi and Lance F. Bosart Department of Atmospheric and Environmental.
Spatial Verification Methods for Ensemble Forecasts of Low-Level Rotation in Supercells Patrick S. Skinner 1, Louis J. Wicker 1, Dustan M. Wheatley 1,2,
Summer Tornadoes – NWA 2015 Statistical Severe Convective Risk Assessment Model (SSCRAM) (Hart & Cohen, 2015) SPC Mesoanalysis Data Every hour from
BP-31 Updated Radar-Based Techniques for Tornado Warning Guidance in the Northeastern United States Brian J. Frugis and Thomas A. Wasula NOAA/National.
Tornado Warning Skill as a Function of Environment National Weather Service Sub-Regional Workshop Binghamton, New York September 23, 2015 Yvette Richardson.
Statistical Severe Convective Risk Assessment Model (SSCRAM) SPC Mesoanalysis Data every hour from (Bothwell et al. 2002) + CG NLDN Lightning.
A Review of the March 28, 2007 Tornado Event Teresa Keck NWS North Platte, Nebraska Courtesy of Mike Hollingshead.
Applied Meteorology Unit 1 High Resolution Analysis Products to Support Severe Weather and Cloud-to-Ground Lightning Threat Assessments over Florida 31.
A Case Study of Two Left-Moving Mesoanticyclonic Supercells on 24 April 2006 Chris Bowman National Weather Service – Wichita, KS.
2. Basic Characteristics and Forecast The 500-hPa pattern for this event featured a deep low centered over Idaho. A composite analysis of past tornado.
A Rare Severe Weather and Tornado Event in Central New York and Northeast Pennsylvania: July 8, 2014 Presented by Mike Evans 1.
Kenneth R. Cook James Caruso Mickey McGuire National Weather Service, Wichita, KS.
The 1 November 2004 tornadic QLCS event over southwest Illinois Ron W. Przybylinski Science and Operations Officer National Weather Service – St. Louis.
Tornadoes – forecasting, dynamics and genesis Mteor 417 – Iowa State University – Week 12 Bill Gallus.
Travis Smith U. Of Oklahoma & National Severe Storms Laboratory Severe Convection and Climate Workshop 14 Mar 2013 The Multi-Year Reanalysis of Remotely.
Ray Wolf and Alex Gibbs NOAA/NWS Davenport, Iowa
Comparison of Spectrum Width, Normalized Rotation, and Correlation Coefficient in Tornadic QLCS and Supercell Circulations Brian Greene University of Oklahoma.
WSR - 88D Characteristics of Significant Tornadoes in New York and New England Lance Franck University of Massachusetts Hayden Frank NOAA/NWS/Weather.
Measures of dispersion
Paper Review Jennie Bukowski ATS APR-2017
Zdr/Kdp Behavior in Potentially Tornadic Storms
Michael L. Jurewicz, Sr. and Christopher Gitro
Hiding under a freeway overpass will protect me from a tornado.
Bow Echo Workshop March 2017
Case Jan Squall line moves through early afternoon...leaving stable conditions, stratus and low LCLs over region through early night hours.
Section Ii: statistics
Zdr/Kdp Behavior in Potentially Tornadic Storms
Brian J. Frugis NOAA/NWS Albany, New York NROW XIX – 7 November 2018
Differences Between High Shear / Low CAPE Environments in the Northeast US Favoring Straight-Line Damaging Winds vs Tornadoes Michael E. Main, Ross A.
Advanced Algebra Unit 1 Vocabulary
Tukey Box Plots Review.
Supercell tornado environments
Presentation transcript:

Diagnosing EF Scale Potential Using Conditional Probabilities Adapted from material and images provided by Bryan Smith, Rich Thompson, Andy Dean, Dr. Patrick Marsh (affiliations SPC)

Impact-Based Warnings “Explore an evolution of the existing NWS warning system to facilitate improved public response and decision making in the most life-threatening weather events.” Intended Outcomes: reframe the warning problem and warning message in terms of societal needs In NWS CR in last 5 years… over 3,000 tornadoes have occurred. 87% of those tornadoes were EF0-1 resulting in 3% of tornado fatalities (all from EF1). 13% of those tornadoes were EF2-5 resulting in 97% of all tornado fatalities increase fidelity of warnings (distinguishing situational urgency by better emphasizing potentially HIGH IMPACT events) incrementally improve warning system (can be done within existing structure) conduct an initial “proof of concept” (small steps) 2

Overview of Smith, et. al. study (2012, 2014) Manual radar analysis –Convective mode assigned using full volumetric WSR-88D archived level II data at beginning of each tornado event –Low-level rotational velocity at 0.5° tilt analyzed during life span of tornado (starting one volume scan prior) and peak value recorded Near-storm environment –Estimated using archived SPC mesoanalysis data Development of conditional tornado probabilities –Box/Whisker diagrams developed that normalize dataset, distinguish between convective modes, and distinguish between radar range from the target –Initial development of raw probabilities are range and mode independent –Raw probabilities alone are not enough for decision-makers –Normalized probabilities are derived as best fits for operational application –Forecaster expertise continues to play a key role in conceptual application

Key Definitions –Low-Level Rotational Velocity (Vrot) – taken at the 0.5 degree slice independent of radar range. For example, dataset encompasses miles from the radar, or 100 – 10,000 feet Above Radar Level (ARL). Peak values recorded not necessarily gate-to-gate. –Convective Mode - determined subjectively via examination of radar signatures –Raw Conditional Probability of Tornado Intensity – probability derived from complete, unfiltered, dataset of Maximum Vrot vs. Tornado Intensity –Normalized Conditional Probability of Tornado Intensity – probability derived from dataset after filtering outliers and normalizing data distribution across EF scale.

January 2009 – May 2013 Tornado segment data filtered by max EF-scale on hourly 40 km horizontal grid Tornado events < 10,000 ft above radar level (1–101 mi range) Total number of tornadoes sampled = 4378

EF-scale Note that the raw dataset is dominated by population of EF0-1 tornadoes (almost 5X more than EF2-5) Data includes all convective modes and 0.5 degree samples at all ranges to 101 miles

EF0 EF3 EF2 EF1 EF4+ Shaded zones indicate most probable EF scale outcome - conditional on tornado occurrence Probabilities are based on raw, unfiltered data for the entire sample * Probabilities are derived by accounting for each tornado (and assigned EF scale intensity) in a 10 kt Vrot bin (e.g kts). The derived probability for each bin is assigned to the mid point of the bin (e.g. 55kts). The total sample size is 4378 tornadoes.

EF2-5 EF0-1 EF2-3 EF4-5 Same dataset, only within IBW Framework (Base Tier Warnings EF0-1 vs. Enhanced Tier Warnings EF2-5). Threshold where EF2-5 tornado becomes the most probable outcome is Vrot > 60 kts. This is very useful information. However, operationally, the use of raw probability does not tell the whole story.

EF-scale This is because the distribution across EF scale is non-normal and is weighted toward the EF0-1 population

Standard Box and Whisker plots. Whisker tips represent 10 th and 90 th percentiles, while boxes are bounded by 1 st and 3 rd quartiles, and dash in the middle is the median value or 2 nd quartile. Note that ~80% of the EF2 population, and ~40% of the EF3 population, fall below the raw conditional 60 kt threshold for EF2+ tornadoes. We need to capture more of these events. Instead, we can normalize the dataset by equally weighting each EF-scale bin (as in the diagrams above) and filter the sample “outliers” outside the tips of the whiskers.

EF2-5 EF0-1 EF2-3 EF4+ Using the normalized and filtered data set, we can derive a set of “Normalized” probabilities for conditional tornado intensity. The “normalized” threshold where a EF2- 5 tornado is the most probable outcome is 45 kts (conditional on tornado occurrence).

Standard Box and Whisker plots. Whisker tips represent 10 th and 90 th percentiles, while boxes are bounded by 1 st and 3 rd quartiles, and dash in the middle is the median value or 2 nd quartile. For the Filtered Population… 45 kt threshold captures 100% of EF3+ tornadoes and ~75% of the EF2 population. However, it also captures ~7% of the EF0 population and ~23% of the EF1 population.

EF-scale In both the scatter plots and box/whisker diagrams there is significant overlap of EF1 and EF2 tornadoes between our normalized 45 kt and raw 60 kt decision thresholds. Because of this, a clean threshold is unattainable. This is where forecaster expertise becomes most important in the warning decision process. Now applied to the non-filtered data set: The 45 kt threshold captures over 92% of EF3+ tornadoes and nearly 67% of the EF2 population. And also captures ~10% of the EF0 population and ~33% of EF1 tornadoes.

Operational Application 1)Use your situational awareness of the mesoscale and near-storm environments. 2)Use your understanding of convective modes. 3)Use your understanding of the character of the low level circulation. 4)Use your understanding of raw and normalized probabilities of conditional tornado intensity. Character of Low Level Circulation Consideration of Convective Mode Use raw and normalized probabilities of conditional tornado intensity Understand mesoscale and near- storm environment Diagnosing/Anticipating the Range of Possibilities 4 Diagnose and Anticipate Most Probable Category of Tornado Intensity (EF0-1 vs EF2-5)

Operational Forecasting Application 1)Use your situational awareness of the mesoscale and near-storm environments. a)Examine CAPE/Shear relationships for environments favorable for supercell development. b)Examine SPC mesoscale analysis for environments favorable for larger tornadoes (e.g. Sig TOR Parameter – STP). c)Be aware of low level boundaries conducive for rapid tilting and/or stretching of local vorticity maxima…. and LCL heights for estimates of cloud base. Diagnosing/Anticipating the Range of Possibilities Neighborhood max value (dark bounded B/W plot) vs. grid value (gray shaded B/W plot) Neighborhood value STP = within 185 km radius, Grid value STP= within 40km x 40 km grid space STP vs. EF-scale

Operational Application 2)Use your understanding of convective mode a)RM Supercells are most likely to produce tornadoes that require enhanced tags. b)QLCS storms that produce significant tornadoes appear to do so with lower Vrot thresholds than RM Supercells. (possibly due to enhanced forward motion vector contributions on right flanks of low level circulations). c)Circulations in disorganized convection are unlikely to produce significant tornadoes that need enhanced tornado tags. Diagnosing/Anticipating the Range of Possibilities

Operational Application 3) Use your understanding of the character of the low level circulation. a)Anticipate how convergent low level circulations will behave given the near-storm environment. b)Be cognizant of radar range from the target. For close-in storms be sure to sample as close to the cloud base as possible for storms that are not yet tornadic. Use the 0.9 slice if necessary. c)Study uses both broad Vrot maxima and Gate-to- Gate Vrot maxima, depending on which is strongest for a given case. Gate-to-Gate Vrot maxima should operationally command more weight and a lower Vrot threshold for EF2+ events. Diagnosing/Anticipating the Range of Possibilities 1933Z 0.5 slice 1933Z 4.0 slice Example of a 0.5 degree convergent rotation below a broad 4.0 degree rotating mesocyclone. Prominent BWER evident in the lower right. This storm is intensifying and will soon produce a tight GTG low level circulation and eventually an EF4 tornado.

Operational Application 4) Use your understanding of raw and normalized probabilities of conditional tornado intensity. a)Keep in mind these are conditional probabilities, but also remember that lead time is important and use as many tools as possible to help anticipate tornado occurrence and potential intensity. You do not have to wait for a report of a tornado before issuing a “CONSIDERABLE DAMAGE THREAT” tag. b)Once a decision is made that a tornado is likely, use the Vrot threshold of 45 knots as the initial point where you should start seriously thinking about a “CONSIDERABLE DAMAGE THREAT” tag. c)Use the Vrot threshold of 60 knots as the point where you should definitely issue a “CONSIDERABLE DAMAGE THREAT” tag. Diagnosing/Anticipating the Range of Possibilities d) For warning decisions between these conditional thresholds, forecaster judgment should be exercise based on your knowledge of 1) near-storm environment, 2) convective mode, 3) character and evolution of the low level circulation.

Re-Capping: Operational Application 1)Use your situational awareness of the mesoscale and near-storm environments. 2)Use your understanding of convective modes. 3)Use your understanding of the character of the low level circulation. 4)Use your understanding of raw and normalized probabilities of conditional tornado intensity. Character of Low Level Circulation Consideration of Convective Mode Use raw and normalized probabilities of conditional tornado intensity Understand mesoscale and near- storm environment Diagnosing/Anticipating the Range of Possibilities 4 Diagnose and Anticipate Most Probable Category of Tornado Intensity (EF0-1 vs EF2-5)

Recently published work Wea. Forecasting (2012) –Demonstrated a relationship between environment, convective mode, mesocyclone strength, and tornado damage intensity EJSSM (2013) –Displayed spatial distributions of supercell-related parameters Wea. Forecasting (2013) –Tornado warning performance (POD and lead-time) related to convective mode and supercell-related parameters

The following slides show some snapshots of recent significant tornadoes. (slides courtesy of Rich Thompson, SPC).

V rot = 84.5 kt Max grid STP = 13.1 Outlook = SLGT 5% Watch = TOR EF3 damage (17JUL2011)

V rot = 40.8 Max grid STP = 1.9 Outlook = SLGT 5% Watch = SVR EF2 damage (20JUN2010)

V rot = 46.7 Max grid STP = 3.6 Outlook = SLGT 10% Watch = TOR EF2 damage (5JUN2009)

V rot = 81.6 Max grid STP = 4.6 Outlook = MDT 10% SIG Watch = TOR EF4 damage (19MAY2013)

V rot = 69.0 Max grid STP = 5.0 Outlook = SLGT 10% Watch = TOR EF2 damage (22MAR2011)

V rot = 51.5 Max grid STP = 1.2 Outlook = SLGT < 2% Watch = SVR EF3 damage (15MAR2012)

V rot = 73.3 Max grid STP = 6.6 Outlook = MDT 5% Watch = SVR EF4 damage (26JUN2010)

V rot = 76.8 Max grid STP = 0.4 Outlook = SLGT 5% Watch = TOR EF3 damage (27JUL2010)

V rot = 69.5 Max grid STP = 7.8 Outlook = MDT 15% SIG Watch = PDS TOR EF3 damage (10APR2011)

V rot = 88.9 Max grid STP = 10.9 Outlook = HIGH 30% SIG Watch = PDS TOR EF3 damage (15APR2012)

V rot = Max grid STP = 6.1 Outlook = HIGH 30% SIG Watch = PDS TOR EF3 damage (2MAR2012)

V rot = 54.4 Max grid STP = 5.8 Outlook = SLGT 5% Watch = TOR EF3 damage (12AUG2011)

V rot = 81.6 Max grid STP = 5.5 Outlook = HIGH 30% SIG Watch = PDS TOR EF4 damage (2MAR2012)

V rot = 84.8 Max grid STP = 8.8 Outlook = MDT 10% SIG Watch = TOR EF4 damage (17JUN2010)

V rot = 70.7 Max grid STP = 7.0 Outlook = MDT 15% SIG Watch = PDS TOR EF2 damage (11APR2011)

V rot = 47.6 Max grid STP = 4.5 Outlook = SLGT 10% SIG Watch = TOR EF4 damage (17JUN2010)

V rot = 62.7 Max grid STP = 14.2 Outlook = MDT 10% Watch = TOR EF3 damage (10APR2011)

V rot = 69.5 Max grid STP = 3.9 Outlook = SLGT 5% Watch = TOR EF4 damage (29FEB2012)

V rot = 65.6 Max grid STP = 3.6 Outlook = SLGT 2% Watch = SVR EF2 damage (27APR2012)

V rot = 99.0 Max grid STP = 7.8 Outlook = MDT 10% SIG Watch = TOR EF5 damage (22MAY2011)

V rot = 84.0 Max grid STP = 9.6 Outlook = SLGT 5% Watch = TOR EF3 damage (28MAY2013)

V rot = 94.7 Max grid STP = 0.8 Outlook = SLGT 2% Watch = SVR EF2 damage (23JUN2012)

V rot = 93.8 Max grid STP = 6.0 Outlook = MDT 10% SIG Watch = TOR EF3 damage (28MAY2013)

V rot = 73.9 Max grid STP = 6.0 Outlook = MDT 10% SIG Watch = PDS TOR EF3 damage (20JUN2011)