1 / 84

Gridded Warnings The next generation of NWS warning product?

Gridded Warnings The next generation of NWS warning product?. Contributions: David Andra Harold Brooks Don Burgess Charles Kerr Jim LaDue Mike Magsig Kiel Ortega Kevin Scharfenberg Travis Smith Greg Stumpf Lou Wicker. HWT/EWP Participants: Brad Grant Les Lemon Dan Miller

gareth
Télécharger la présentation

Gridded Warnings The next generation of NWS warning product?

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Gridded WarningsThe next generation of NWS warning product? Contributions: David Andra Harold Brooks Don Burgess Charles Kerr Jim LaDue Mike Magsig Kiel Ortega Kevin Scharfenberg Travis Smith Greg Stumpf Lou Wicker HWT/EWP Participants: Brad Grant Les Lemon Dan Miller Steve Nelson Patrick Burke Liz Quoetone Dave Sharp Steve Zubrick

  2. Introduction • This idea and project are nascent. • This work must consider many intersecting disciplines: • Meteorology • Technology • Sociology • Human Factors

  3. Recap from 1st Wkshp • Warnings (forecasts) are for areas • Reports (events) are points • Probability of Detection (POD): • points within polygons / total points • False Alarm Ratio (FAR): • polygons without points / total polygons • Measures of skill (e.g., CSI) cannot be calculated • These definitions are inconsistent • No measure of correct null forecasts • Warning area sizes are inconsistent • Verification really affects areas, not points

  4. Storm versus County • NWS implementing “Storm-based Warnings” for FY08. • Characterizes the area to be threatened with (supposedly) no regard to geopolitical boundaries • But there are still issues

  5. Area-Based WarningShortcomings • Storm-based warnings are still area forecasts verified by point events. • The polygons are just “differently shaped” • The threat information for the polygon remains “monotonic” • Each location inside polygon is under exact same threat for the exact same time period • Each location inside (outside) polygon is given 100% (0%) certainty of event

  6. Vision • Warnings and Reports are GRIDDED • Consistency between forecasts and events • Consistency with other forecast products • Allows for growth (added detail) • Can be aggregated to simpler formats

  7. OB8.1 WarnGen Limitations • Allow toggling individual counties off/on • Automatically clip small portions of counties within an area threshold setting • Not provide for situations when threats straddle adjacent CWAs • Not match the edges of adjacent warnings where there was manual or automated county clipping • Not provide a easily-identifiable preview of the actual polygon until the after the warning was transmitted

  8. Polygon Slivers This storm produced a damaging tornado that went nearly down the unwarned “sliver” that the arrow points to!

  9. Was this really intended?

  10. Downstream Warnings • Very little overlap between adjacent warnings • Many times, a new polygon is issued only when storm is just about to exit the previous polygon. • This gives inconsistent lead time to all points within each polygon, with some points getting much less and sometimes zero lead time.

  11. A t=0

  12. A t=0

  13. No Warning for User B A B t=0

  14. No Warning for User B A B t=5

  15. No Warning for User B A B t=10

  16. No Warning for User B A B t=15

  17. No Warning for User B A B t=20

  18. No Warning for User B A B t=25

  19. No Warning for User B A B t=30

  20. No Warning for User B A B t=35

  21. No Warning for User B A B t=40

  22. No Warning for User B A B t=45

  23. < 5 min lead time for User B A B t=50

  24. < 5 min lead time for User B A B t=55

  25. Area-Based WarningShortcomings • WarnGen has a motion estimator (“Drag Me To Storm”): • But, once any vertex in the swath polygon is changed, the storm motion and swath become decoupled • e.g., if swath is lengthened or shortened, it no longer matches the storm speed and warning time. • Forecaster has no control over initial swath area provided by WarnGen (point and line motion only) • Forecaster has no control over motion uncertainty values (speed and direction) • Therefore, the forecaster can only guess the result of these variables when adjusting the swath polygon

  26. Hypothesis • Instead of the forecaster guessing at the swath… • …we propose that much more robust warning information can be derived if the forecaster instead determines • the initial threat area at time=0, along with • the motion vector, and adds • motion uncertainty information

  27. Drivers for growth • This information used to integrate a more accurate swath • Continuously advecting threat areas • Meaningful guidance on time of arrival and time of departure • Allows for the addition of forecaster uncertainty information (e.g., probabilities)

  28. Deriving Swath User defines: Initial Threat Area at time=0 time = 0

  29. Method User: Drag Me To Storm Use cursor to reposition threat area time = -20

  30. Method User: Drag Me To Storm Use cursor to reposition threat area time = -20

  31. Method Motion Vector Calculated time = 0, 15, 30, 45, 60

  32. Method Threat Area Advected

  33. Method User Defines: Motion Uncertainty ± 5 m s-1 ±15°

  34. Method Speed uncertainty stretches threat along motion vector ± 5 m s-1 ±15°

  35. Method Direction uncertainty stretches threat perpendicular to motion vector ± 5 m s-1 ±15°

  36. Method Integrated threat area across warning time provides swath

  37. Swath properties • Greater (lesser) motion uncertainty leads to wider (more narrow) swath • Larger (smaller) initial threat area leads to wider (more narrow) swath • Neither of these factors can be controlled in WarnGen. • Swaths are truly independent of geopolitical boundaries.

  38. Advecting threat • As threat area advects downstream, the area affects different grid points at different times. • For each grid point in the warning, the intersection of the threat area over time provides information about • time of arrival • time of departure

  39. Advecting Threat Area x A B t=0 TOA TOD NOTE: No warning (yet) For User B 1 Warn? 0 A B Time 

  40. Advecting Threat Area x A B t=5 TOA TOD TOA TOD NOTE: > 40 min lead time For User B 1 Warn? 0 A B Time 

  41. Advecting Threat Area x A B t=10 TOA TOD TOA TOD 1 Warn? 0 A B Time 

  42. Advecting Threat Area x A B t=15 TOA TOD TOA TOD 1 Warn? 0 A B Time 

  43. Advecting Threat Area x A B t=20 TOA TOD TOA TOD 1 Warn? 0 A B Time 

  44. Advecting Threat Area x A B t=25 TOA TOD TOA TOD 1 Warn? 0 A B Time 

  45. Advecting Threat Area x A B t=30 TOA TOD TOA TOD 1 Warn? 0 A B Time 

  46. Advecting Threat Area x A B t=35 TOA TOD TOA TOD NOTE: Point A already In “all clear” 1 Warn? 0 A B Time 

  47. Advecting Threat Area x A B t=40 TOA TOD TOA TOD NOTE: Point A already In “all clear” 1 Warn? 0 A B Time 

  48. Advecting Threat Area x A B t=35 TOA TOD TOA TOD NOTE: Point A already In “all clear” 1 Warn? 0 A B Time 

  49. Advecting Threat Area x A B t=50 TOA TOD TOA TOD NOTE: Point A already In “all clear” 1 Warn? 0 A B Time 

  50. Threat Type • Each severe weather type can be depicted on separate grids • Hail • Wind • Tornado • Lightning • Threat subtypes could also be depicted • Hail size • Wind speed • Lightning Density

More Related