Update on NCAR Auto-Nowcaster

1
Update on NCAR Auto-Nowcaster
Juneau, AK
2
The Auto-Nowcaster System

• An expert system which produces short-term (0-1
  hr) forecasts of thunderstorm initiation, growth
  and decay using fuzzy logic.
• Recent developments have allowed human users
  (i.e. forecasters) to input information into the
  system in a relatively quick and simple fashion.
• A test of the updated system in an operational
  environment is taking place at the Dallas/Fort
  Worth WFO.

3
The Auto-Nowcaster System is unique in its
ability to provide nowcasts of storm initiation
by..
Detection and extrapolation of surface
convergence boundaries .
.that trigger
thunderstorm initiation and impact storm
evolution.
4
Example of Auto-Nowcaster Initiation Forecast
1 hour forecast
Verification
5
Flow Chart for the Auto-Nowcaster System
Final Prediction
6
Flow Chart for the Auto-Nowcaster System
Final Prediction
7
Example of fuzzy logic
Predictor Field 1
8
Predictor Field 2
9
Predictor Field 3
Membership Function
Cumulus clouds
1
Likelihood
.8
-.5
.4
.4
Likelihood
-.5
-.5
-1
Cumulus cloud type
10
Likelihood 1
Likelihood 2
Likelihood 3
S
Final combined likelihood of initiation
11

• Environmental conditions (RUC)
• Frontal likelihood
• Layered stability
• CAPE (max between 900 and 700 mb)
• Mean 875 to 725 mb Relative Humidity
• Boundary-layer
• Convergence
• LI (based on METARS)
• Vertical velocity along boundary (maxW)
• Boundary-relative steering flow
• New storm development along boundary
• Clouds
• Clear or Cumulus
• Vertical develop as observed by drop in IR temps

12
60 Minute Initation (rules with satellite data)
Wt 0.10 Range 0 to 0.10
Wt 0.17 Range -0.085 to 0.17
Wt 0.20 Range -0.20 to 0.02
Wt 0.17 Range -0.17 to 0.17
Wt 0.08 Range -0.02 to 0.08
Wt 0.16 Range -0.08 to 0.16
13
60 Minute Initation (rules with satellite
data) Cont.
Wt 0.20 Range -0.2 to 0.2
Wt 0.20 Range 0 to 0.20
Wt 0.20 Range 0 to 0.20
Wt 0.15 Range 0 to 0.15
Wt 0.25 Range 0 to 0.25
14
60 Minute Initation (rules with satellite
data) Cont.
Lake Wt 0.10 Range -0.10 to 0
Sat_Clear Wt 0.40 Range -0.40 to 0
Boundary Collision Wt 0.12 Range 0 to 0.12
Initiation Levels 0.70 gt Init 1 0.90 gt
Init 2 1.20 gt Init 3
15
Predictor Fields
16
Predictor Fields
17
Why do we need a forecaster in the loop??

• Forecasters see the larger picture
• Conceptual Models
• Ignore bad data points
• Understand limitations of NWP and observations

18
(No Transcript)
19
Draw Tool
20
Draw Tool
21
Draw Tool
Entering a convergence boundary in real time is
as simple as this demonstration!
22
Forecaster-tools Boundary Entry
23
Forecaster Entered Polygons
24
Where has the Auto-nowcaster been demonstrated ?

• Weather Forecast Office Washington DC (Sterling,
  VA)
• Sydney Australia Forecast Office
• U. S. Army White Sands Missile Range
• Central U. S. for the FAA

• Process of being transferred to
• Bureau Meteorology Beijing China
• U.S National Weather Service Dallas/Fort Worth
  Weather Forecast Office
• AWIPS

25
Auto-Nowcaster at Ft. Worth WFO
26
Auto-Nowcaster at Ft. Worth WFO05 April 2005
On this day, dryline was moving in from the west.
Below is an example of one of the Area Weather
Updates that they put out on that day AREA
WEATHER UPDATE NATIONAL WEATHER SERVICE FORT
WORTH TX 310 PM CDT SUN APR 10 2005 gt WARNING
DECISION UPDATE FOR NORTH TEXAS MESOANALYSIS
PROGRAMS SHOW 1000-1500 J/KG CAPE ALONG AND JUST
AHEAD OF DRYLINE. THUS...CU/DEVELOPING STORMS
ALONG/E OF DRYLINE SHOULD CONTINUE TO INTENSIFY.
STORM INITIATION TOOL ALSO SUGGESTS
HIGH POSSIBILITY 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.
27
Ft. Worth WFO 05 April 2005
Forecast 2003 Z Radar Image 2003 Z
Forecast 2003 Z Radar Image 2108 Z
28
Ft. Worth WFO 05 April 2005
Forecast 2044 Z Radar Image 2044 Z
Forecast 2044 Z Radar Image 2142 Z
29
Ft. Worth WFO 05 April 2005
30
Ft. Worth WFO 25 April 2005

• On this day there was a dryline moving in from
  the west.
• Storms developed a couple of hours earlier than
  originally thought.
• The ANC system captured the initiation of this
  event with about a 40-50 minute lead time.
• The storms went severe within 30 minutes of
  initiation (about 50 miles west of DFW Metroplex)
  with quarter sized hail, strong winds and several
  mesocyclones.

31
Ft. Worth WFO 25 April 2005
Forecast 1831 Z Radar Image 1831 Z
Forecast 1831 Z Radar Image 1929 Z
32
Ft. Worth WFO 25 April 2005
Forecast 1842 Z Radar Image 1842 Z
Forecast 1842 Z Radar Image 1941 Z
33
Ft. Worth WFO 25 April 2005
Forecast 1854 Z Radar Image 1854 Z
Forecast 1854 Z Radar Image 1952 Z
34
Ft. Worth WFO 25 April 2005
Forecast 1905 Z Radar Image 1905 Z
Forecast 1905 Z Radar Image 2003 Z
35
Ft. Worth WFO 25 April 2005
Forecast 1917 Z Radar Image 1917 Z
Forecast 1917 Z Radar Image 2020 Z
36
Ft. Worth WFO 25 April 2005
Forecast 1929 Z Radar Image 1934 Z
Forecast 1929 Z Radar Image 2030 Z
37
Ft. Worth WFO 25 April 2005
Forecast 1941 Z Radar Image 1940 Z
Forecast 1941 Z Radar Image 2042 Z
38
Ft. Worth WFO 25 April 2005
Forecast 1952 Z Radar Image 1951 Z
Forecast 1952 Z Radar Image 2054 Z
39
Ft. Worth WFO 25 April 2005
Forecast 2003 Z Radar Image 2003 Z
Forecast 2003 Z Radar Image 2107 Z
40
Ft. Worth WFO 25 April 2005
Forecast 2020 Z Radar Image 2020 Z
Forecast 2020 Z Radar Image 2119 Z
41
Ft. Worth WFO 25 April 2005
Forecast 2030 Z Radar Image 2030 Z
Forecast 2030 Z Radar Image 2129 Z
42
Ft. Worth WFO 25 April 2005
Forecast 2042 Z Radar Image 2042 Z
Forecast 2042 Z Radar Image 2142 Z
43
Ft. Worth WFO 25 April 2005

• This was one of the first events after some
  modifications had been made to the system based
  on forecaster feedback (i.e. a CIN predictor
  field was added to the forecast logic).
• Had this information not been included, the ANC
  system would have produced several hours of false
  alarms prior to storms initiating.
• Several forecasters consulted the system during
  this event and it was used to help with spotter
  activation.

44
Our Impressions of the Ft. Worth WFO Deployment

• Users have been very accommodating of new
  technology.
• Users have done a really good job of entering
  boundaries.
• Use the system most often early on in the
  convective events during initiation/development
  phase.
• Received a lot of very useful feedback from the
  users.
• Have incorporated some of the forecasters ideas
  which we feel have led to improved forecasts and
  are in the process of implementing others.