Quasi-Linear Convective System Tornado Warnings Prospects for False Alarm Reduction John P. Gagan and Jason S. Schaumann National Weather Service Springfield, MO
Background FY2008-09 tornado warnings issued at a substantial pace There were major events January 2008 – 33 tornadoes May 2009 – “Super Derecho” 18 tornadoes Wide swath of 70-90mph winds In all – 323 tornado warnings issued Going back to 2008-09, WFO SGF issued warnings at a substantial clip. Certainly there were some substantial tornadic events: an incredibly rare January tornado outbreak in 2008 – 33 tornadoes And the “super derecho” of May 2009 – 18 tornadoes and a wide swath of 70-90mph straight-line winds However, a surprising number of tornado warnings were issued. 323 warnings to be exact.
Background Why? Reality required a harsh look in the mirror: Rough transition to Storm Based Warnings? “Hangover Effect” from larger severe events? Unwarned events affecting decisions? Reality required a harsh look in the mirror: Understanding of mesoscale – shaky Understanding of QLCS radar features – shaky Warning decisions – reactive, lacked process Did we have a good pulse on what really mattered? Transition to storm based warnings, major events, unwarned events…these are few “convenient” explanations that could explain the large amount of warnings; but they gloss over the truth. Quote from “super derecho” event review - “This event again illustrated the challenge of issuing tornadoes in a squall line/QLCS environment from both a radar identification and polygon methodology standpoint.” The bottom line: The fundamental understanding of the mesoscale environment was shaky Warning decisions has become based more on reaction rather than process, focusing on what really mattered The time had come to hit the “reset” button.
How To Move Forward? 133 Tornado Warnings FY2009 84 QLCSs 73% POD, 81% FAR QLCS tornadoes – significant challenge Rapid “spin up” Lead time - ~5 min Why not just stop issuing tornado warnings on QLCSs? Problem solved! Right? Focusing in on FY2009, almost 2/3rds of our tornado warnings were for QLCSs. While POD was respectable, FAR was 81%. Without a doubt, QLCSs are a major challenge. They are characterized by rapid spin ups and short lead times. Given the large amount of QLCS tornado warnings, one could logically conclude that the best way to minimize the number of warnings and reduce FAR would be to just simply stop issuing tornado warnings for QLCS events. PROBLEM SOLVED!!! Right????
Buffalo, MO – 2/29/2012
Sunset Hills, MO – 12/31/2010
Sunset Hills, MO – 12/31/2010
A Plan, A Process No waving the white flag – science to the rescue! “Moneyball” approach Physical understanding Focus on variables that matter “Culture eats strategy for breakfast.” Acknowledge the deficiencies, train…train…train. “The approach in what you do results in what you get.” We can do better than wave the “white flag.” Find a way to put the odds in the favor of the radar operator to make sound warning decisions. “Weatherball” Emphasis on physical processes, focusing on variables that matter. While an overwhelming amount of research has been conducted on the topic of environments and physical processes which lead to supercell tornadogenesis, research on environmental settings for QLCS tornadoes has been limited to mainly case studies. Moneyball (2011)
QLCS Radar Interrogation Strategies An A-Z Process for Radar Operators Three Ingredients Method for anticipating mesovortex genesis Warning decision guidance Polygon strategies for mesovortices Radar interrogation strategies and time budget The Karate Kid (1984) Warn Daniel-san! What, when, where, how. Puts the odds in the favor of the radar operator to make sound warning decisions. Stats will shake out Main focus will be on three ingredients method, but we will briefly touch on tornado warning guidance Supports sound warning decisions by systematically honing in on areas where mesovortices and/or tornadoes are favored.
Three Ingredients Method Schaumann and Przybylinski 2012 QLCS mesovortex genesis and strong intensification is favored where the following three criteria are co-located: A portion of the QLCS in which the system cold pool and ambient low-level shear are nearly balanced or slightly shear-dominant Where 0-3 km line-normal bulk shear magnitudes are equal to or greater than 30 knots Where a rear-inflow jet (RIJ) or enhanced outflow causes a surge or bow in the line 2012 Severe Local Storm conference (manuscript and poster) Mesovortex genesis and strong intensification does not necessarily mean a tornado! 17 Nov 2013 Central Indiana Keep in mind that mesovortex genesis and strong intensification does not necessarily mean a tornado!
Shear/Cold Pool Balance Locating Balanced/Slightly Shear Dominant Regions 0.5° SRM 0.5° Z Slightly Shear Dominant UDCZ Inflection Point Balanced Determining ingredient #1 Worst is not always first! Point out inflection point Mesovortices remain tethered to the UDCZ Slightly Cold Pool Dominant 3 Apr 2015 Southern Missouri Locate updraft/downdraft convergence zone (UDCZ) using 0.5° V/SRM products. Compare location of UDCZ to 0.5° reflectivity field.
Three Ingredients Method Application 0-3 km Bulk Shear/MLCAPE SPC Meso Page 0.5° SRM UDCZ 0.5° Z Favored Region for Mesovortex Genesis 0-3 km line-normal bulk shear magnitude ≥ 30 knots Balanced or slightly shear dominant RIJ or enhanced outflow causing surge or bow in the line
Statistical Performance of Three Ingredients Method 2013 Hollings Study – Stanford et al. Correctly predicted 79% of mesovortices Falsely predicted a mesovortex 23% of the time Out of 67 mesovortices identified, the following two ingredients were always present: Line was balanced or slightly shear dominant (Ingredient #1) Local surge or bow in the line (Ingredient #3) 67 mesovortices and 64 non-mesovortex surges Without ingredients 1 and 3, a mesovort is highly unlikely (probably don’t need a TOR) Cold Season Warm Season Hollings Study Case Domain
Building on the Three Ingredients Tornado Warning Guidance Radar signatures and mesoscale parameters which lead to increased tornado potential Combination of past research and local studies Further tilts odds in favor of radar operators
Warning Decision Guidance Scenarios for Considering a Tornado Warning UDCZ Enhanced Surge Inflection Point RIN FIN EF-1 Using the three ingredients method as a foundation, we have identified additional radar features and mesoscale parameters which indicate an increased likelihood for tornadoes EF-1 tornado near Vanzant, MO Don’t want to wait until the last two scenarios if at all possible Poster session this afternoon and again on Wednesday 0.5° Z 0.5° SRM While one of the eight scenarios being met is often worthy of a Tornado Warning, confidence should further increase if more than one scenario and/or “nudgers” are present.
Three Ingredients Methodology Before and After FY09 to FY11 FY12 to FY15 QLCS Warnings 107 63 QLCS Tornado Events 47 49 Fully Warned 29 28 Partially 4 4 Unwarned 14 17 Verified 21 26 Not Verified 86 37 POD 68% 64% FAR 81% 59% Warnings reduced by 37% - almost identical number of tornado events Unverified warnings reduced by 57% FAR reduced 22%
FY09 to FY11 compared to FY12 to FY15 QLCS tornado warnings reduced by 37%. Unverified warnings reduced by 57%. FAR drop of 22% POD drop of 3.5% Bottom line- the 3 ingredients methodology is having a positive affect on QLCS tornado warning accuracy. This remains a work in progress – the development of the scenarios and nudgers presented earlier will further refine the approach to QLCS tornado warnings.
May 2009 Case Warning Execution Prior to the Three Ingredients Method 0-3 km Bulk Shear/VGP SPC Meso Page Line – slightly outflow dominate, rearward leaning 0-3km bulk shear was ~30kt, however the line normal component was poor – parallel to the line Applying the 3 ingredients method to a line such as this today, would yielded 0 tornado warnings (11 were issued for this case alone).
Success Story Revisiting the Southern MO Line Segment 0-3 km Bulk Shear/MLCAPE SPC Meso Page EF-1 This example from southern MO is an example of a success story. Three ingredients method was met – along with multiple scenarios and nudgers End result: Quality warning decision, well executed polygon 12 minute lead time for Vanzant (EF-1) 13 minute lead for Pomona (EF-2)
References and Resources John.Gagan@noaa.gov Jason.Schaumann@noaa.gov Schaumann, J. S., and R. W. Przybylinski, 2012: Operational Application of 0-3 km Bulk Shear Vectors in Assessing QLCS Mesovortex and Tornado Potential. Preprints, 26th Conf. on Severe Local Storms, Nashville, TN, AMS, P9.10. 2013 Hollings Study by Stanford et al. (Available on WFO SGF website under Local Office Research) SGF Google Site (SGF QLCS Guidelines) Poster Sessions Monday 3:00 to 4:30 PM – A Three Ingredients Approach to Anticipating Mesovortex Genesis (Schaumann et al.) Wednesday 2:30 to 4:00 PM – Application of the Three Ingredients Method to QLCS Tornado Warning Decisions at NWS Bismarck (Mathews and Schaumann) John.Gagan@noaa.gov Jason.Schaumann@noaa.gov
0-3 km Line-Normal Bulk Shear Threshold Ingredient #2 Mean 33.5 Mean 25.9 Lower Quartile 30.2 Shows statistical significance of 30 knot threshold Shear lower than 30 knots accounted for all of the “misses” P-value for line-normal 0-3 km bulk shear magnitudes was significantly lower than 1%.
Mesovortex Tornadoes Key Findings Average lead time between surge genesis and first tornado report was 18 minutes Orientation of vectors to UDCZ appears to play a key role Using a surge genesis as a possible precursor to draft a polygon can save valuable minutes Similar magnitudes show that 0-3 km bulk shear vectors were nearly perpendicular to UDCZ for tornadic mesovortices Lower Quartile 33.5 Lower Quartile 31.0
Teaser FY09 to FY11 FY12 to FY15 QLCS Warnings 107 Tornado Events 47 Fully Warned 29 Partially 4 Unwarned 14 Verified 21 Not Verified 86 POD 68% FAR 81% Teaser
The Effect of the Three Ingredients Methodology FY 2009 to FY 2011 FY 2012 to FY 2015 QLCS Warnings – 107 Verified – 21 Not Verified – 86 Tornado Events – 47 Fully Warned – 29 Partially – 4 Unwarned – 14 POD – 68% FAR – 81% QLCS Warnings – 63 Verified – 26 Not Verified – 37 Tornado Events – 49 Fully Warned – 28 Partially – 4 Unwarned – 17 POD – 64% FAR – 59% Option 1