Post on 05-Jan-2016
The Man-In-The-Loop (MITL) Nowcast Demonstration: Forecaster Input into a Thunderstorm Nowcasting System
R. Roberts, T. Saxen, C. Mueller, E. Nelson, A. Crook, H. Cai, J. Pinto, J. Wilson, D. Megenhardt, D. Albo, S. Dettling, N. Oien,; National Center for Atmospheric Research, Boulder, CO
S. Fano, W. Bunting; National Weather Service Forecast Office, Ft. Worth, TX
K. Johnston, S. Smith, M. Ba; National Weather Service, Silver Spring, MD
T. Amis; NWS Central Weather Service Unit, Ft. Worth, TX
8 September 2005Toulouse
•To determine the role of the forecaster in nowcasting
•To demonstrate the added value of forecaster input into automated aviation-related thunderstorm nowcast products
MITL Demonstration
NWS Forecaster NCARAuto-nowcaster System
National Weather Service Forecast Office Ft. Worth, Texas
End User Aviation Community Public
March 2005 – March 2007
Extrapolates, grows, dissipates and initiates storms.
1 hour forecast Verification
Storm Initiationnowcasts
Extrapolated nowcasts
NCAR Auto-Nowcaster (ANC) System
How does it do this?
Combined Likelihood Field for Storm Initiation
Blue Regions - Little chance of storm development
Green Regions - Moderate likelihood for thunderstorms
Red Regions - Areas of forecast storm initiation
60 min Storm Initiation Likelihood Field
• Environmental conditions (Numerical Model)
– Frontal forcing (θe gradients, vorticity, convergence)
– CAPE/CIN Layered instability– CAPE (max between 900 and 700 mb)– Relative humidity (average from 875-to-
725 mb)• Boundary-layer
– Magnitude of convergence and vertical shear
– Colliding boundaries– Vertical velocity along boundary – Boundary-relative steering flow– New storm development along boundary
• Clouds– Clear sky or cumulus clouds– Cloud growth observed with cloud top
cooling rate
MITL Demonstration
NWS Forecaster? NCARAuto-nowcaster System
National Weather Service Forecast Office Ft. Worth, Texas
End User Aviation Community Public
March 2005 – March 2007
FAA RCWF Domain June 12, 2003
A single piece of observational data alone is not sufficient to automatically detect surface convergence boundaries on all scales, from a cold front down to gust fronts and lake breezes. Forecasters can fill this role.
60 min
0 min
Entering convergence features locations requires only a few minutes of a forecaster’s time…
However, not all boundaries are created equal…..
Thunderstorm Initiation Likelihood Field
So have added a new tool: Human-drawn polygon tool to increase or decrease forecast interest in an area
Tool was used at the WFO to bump up interest in this area
Thunderstorm Initiation Likelihood Field
Ft. Worth WFO 25 April 2005
Forecast: 1831 ZRadar Image: 1831 Z
Ft. Worth WFO 25 April 2005
Forecast: 1842 ZRadar Image: 1842 Z
Ft. Worth WFO 25 April 2005
Forecast: 1854 ZRadar Image: 1854 Z
Ft. Worth WFO 25 April 2005
Forecast: 1905 ZRadar Image: 1905 Z
Ft. Worth WFO 25 April 2005
Forecast: 1831 ZRadar Image: 1831 Z
Forecast: 1831 ZRadar Image: 1929 Z
Ft. Worth WFO 25 April 2005
Forecast: 1842 ZRadar Image: 1842 Z
Forecast: 1842 ZRadar Image: 1941 Z
Ft. Worth WFO 25 April 2005
Forecast: 1854 ZRadar Image: 1854 Z
Forecast: 1854 ZRadar Image: 1952 Z
Ft. Worth WFO 25 April 2005
Forecast: 1905 ZRadar Image: 1905 Z
Forecast: 1905 ZRadar Image: 2003 Z
Ft. Worth WFO 25 April 2005
Forecast: 1917 ZRadar Image: 1917 Z
Forecast: 1917 ZRadar Image: 2020 Z
Ft. Worth WFO 25 April 2005
Forecast: 1929 ZRadar Image: 1934 Z
Forecast: 1929 ZRadar Image: 2030 Z
Ft. Worth WFO 25 April 2005
Forecast: 1941 ZRadar Image: 1940 Z
Forecast: 1941 ZRadar Image: 2042 Z
Ft. Worth WFO 25 April 2005
Forecast: 1952 ZRadar Image: 1951 Z
Forecast: 1952 ZRadar Image: 2054 Z
Ft. Worth WFO 25 April 2005
Forecast: 2003 ZRadar Image: 2003 Z
Forecast: 2003 ZRadar Image: 2107 Z
Ft. Worth WFO 25 April 2005
Forecast: 2020 ZRadar Image: 2020 Z
Forecast: 2020 ZRadar Image: 2119 Z
Ft. Worth WFO 25 April 2005
Forecast: 2030 ZRadar Image: 2030 Z
Forecast: 2030 ZRadar Image: 2129 Z
60 min Nowcast Field
End End End
0
5
10
15
20
25
Poor Reasonable Good Very Good
Forecasters’ Qualitative Assessment of Nowcasts
For 43 events
Forecast Quality: GoodNo initiation was forecasted on boundary 1 and none occurred. Boundary 3 is showing initiation just west of Young and Stephens Counties.Forecast Quality: PoorSlow response on Boundary 6. Cells beginning to initiate in Kaufman County but no initiation shown by Auto-nowcaster.
AREA WEATHER UPDATE NATIONAL WEATHER SERVICE FORT WORTH TX 310 PM CDT SUN APR 10 2005 > WARNING DECISION UPDATE FOR NORTH TEXAS
MESOANALYSIS PROGRAMS SHOW 1000-1500 J/KG CAPE ALONG AND JUST AHEADOF DRYLINE. THUS...CU/DEVELOPING STORMS ALONG/E OF DRYLINE SHOULDCONTINUE TO INTENSIFY. STORM INITIATION TOOL ALSO SUGGESTS HIGHPOSSIBILITY OF DEVELOPMENT FARTHER SW...OVER CORYELL/LAMPASAS COUNTY AREA. DEEP-LAYER SHEAR MORE THAN SUFFICIENT FOR ORGANIZED STORMS AND AT LEAST MID-LEVEL MESOS. AS STORMS EVOLVE INTO THE EVENING...A MORE LINEAR MODE IS EXPECTED.
Summary• Initial results of the forecaster-computer mix are very encouraging
• Based on forecaster feedback modifications have been made to the ANC system in real-time and more will be added this winter
• Quantitative and qualitative evaluations of performance will be done at the end of the first year
-includes working closely with B. Brown’s group to use their new object-oriented verification techniques
• Continue training forecasters on the ANC system and working closely with the CWSU to refine the nowcast products for the aviation community
Is there a role for the forecaster in nowcasting?