A Severe Weather Proxy Developed from the NOMADS Real Time Data Base of an Operational Model

1
A Severe Weather Proxy Developed from the NOMADS
Real Time Data Base of an Operational Model
EMC


NCEP/Environmental Modeling Center
Jordan C. Alpert jalpert_at_ncep.noaa.gov Jun
Wang/SAIC jun.wang_at_noaa.gov
21st Conf on IIPS, 18.4, 85 Annual AMS Meeting,
9-13 January 2005, San Diego, California
where the nations climate and weather services
begin
2
Motivation for a Severe Weather Proxy from an
Operational NCEP Numerical Weather Prediction
(NWP) Model
There are difficulties with NWP models predicting
severe weather such as

• Highest resolution Operational NWP model is not
  able to resolve severe weather causing phenomena
  that forecasters need to make accurate
  predictions. This is because of the
• Lack of horizontal and vertical scale resolution
  for observations and models with sufficient
  detail and physical knowledge.
• Severe weather
• Tornado, damaging winds gt50kts or hail gt¾ (SPC
  web page)
• Resulting from severe or organized convection,
• Thunderstorms, super cells, squall lines,
  multi-cell complexes
• An array of indicators is used by forecasters to
  analyze severe weather potential at or near
  observation time and from forecasts
• Lifted Index, Storm Relative Helicity, CAPE,
  Vertical profiles of weather elements such as
  temperature, dew point and wind.

3
A challenge in reducing the severe weather risk
To improve the utilization of model output
signatures of severe weather.
Models represent (simulate) very well dynamic and
physical processes

• Wind convergence patterns,
• Often along frontal boundaries,
• Genesis of complex systems along fronts,
• Vorticity advection and vertical motions,
• Often occurring at scales of motion that are
  barely resolvable by models and may have
  difficulty reproducing the convection details.
• For example
• Qualitatively the vertical extent of convergence
  patterns associated with severe weather appears
  to grow and be advected vertically.
• We can not observe this phenomena directly.
• We suggest that the volume, shape or vertical
  extent of the 3-D convergence patterns from
  forecast models is a signature for sub-grid scale
  convective outbreaks.
• This was noticed by viewing Vis5D 3-D displays
  of model convergence patterns.

4
Interactive Visualization of Animated 3-D Model
Data

• In a previous report (19th Conf on IIPS 15.2,
  AMS) it was shown that convergence pattern
  signatures and other 3-Dimensional displays were
  unique and useful, compared to 2-Dimensional
  ones.
• Display Three-Dimensions (D3D), Szoke, et. al.,
  2001 software available for Advanced Weather
  Interactive Processing System (AWIPS) under test
  for forecasters at some Weather Forecast Offices.
  
• Using Vis5D, visualization 5-dimensional,
  meaning 3 space, time and variable space, is an
  interactive visualization system created at the
  University of Wisconsin (Hibbard, 1994).
• In use by the science, educational and by the
  atmospheric science and NWP model research
  community as well as other display systems.
• Can View renderings of Iso-surfaces of
  convergence patterns.
• The basic visualization capabilities of these
  systems are similar as D3D was created from
  Vis5D.

5
3-dimensional rendering of the 21-hr NCEP
Operational forecast of convergence gt3X10-5 s-1 a
yellow iso-surface from IC of 00Z 17SEP2004
6
A Model Severe Weather Proxy

• Convergence calculated from the motion fields of
  the NCEP Global Forecast System
• Test that the model associates these convective
  areas with severe weather according to vertical
  extent, shape and volume.
• Can the proxy distinguish between events with no
  severe weather?
• Need to define a precise criteria
• Define a convergence amplitude criteria,
• Define a vertical extent as number of vertical
  levels between the surface and 500 mb where
  convergence amplitude criteria is met.
• Qualitatively the convergence patterns grow in
  vertical extent in proportion to increased
  precipitation, winds, decreased surface pressure.
  The NWP model may advect the convergence
  patterns upward in response to unresolvable
  meso-scale phenomena.

7
Verification
A convenient verification for objectively
locating observed severe weather is the NCEP
Severe Storms Prediction Center (SPC) Severe
Weather Watch or Warning Box.

• The Watch means that severe weather has potential
  in an area usually over a few hours,
• The Watch is a manually prepared prediction from
  conventional and non-conventional observations
  such as radar and RAOB soundings.
• The Warning means that severe weather is known to
  have occurred in an area.
• A Watch or Warning consists of a text file
  indicating the location by latitude and
  longitude pairs outlining a polygon area of
  severe weather potential typically 25,000 square
  miles.
• The often used model forecasts of Best Lifted
  Index, where negative values less than 4 are
  considered an indicator of potential severe
  thunderstorms, will also be verified with watch
  box locations as a comparison.
• Watch and Warning box areas will be used as the
  verification area. Forecasted patterns meeting a
  convergence criteria will be considered a success
  (Hit) if it is forecast within the Watch or
  Warning box.

8
Convergence pattern (X10-5 s-1) at 1000mb for
17SEP04 21-h forecast and the SPC Watch box (in
black, middle Atlantic location) valid at the
same time and over the next 3 hours. The
verification area of interest is shown in red.
9
Accessing the Data Using NOMADS servers
The NOMADS real time server was used to obtain
the wind fields needed to calculate the
convergence as well as the lifted index.

• Using the NOMADS OPEN-DAP/DODS server to download
  the GFS model run history, including unpacking
  the model GRIB files and organizing
  projection/location information.
• This is done using constrained queries to the
  server of the form http//nomad2.ncep.noaa.gov90
  90/dods/gfs/archive/gfs20040917/gfs/_00z.ascii?vgr
  d10m77110140235300
• Replacing vgrd10m with vgrd would include the
  vertical wind profile.
• 77 indicates the seventh forecast time (3-hr
  intervals) or 21z.
• 110140235300 are the grid points used in
  the eastern US domain chosen to encompass
  inclusively an area where severe weather often
  occurs. The region under consideration starting
  with 110 is 100 degrees north of the south pole
  and 235 represents 235 degrees east of 0
  meridian for a 30 by 65, over 1200 grid point
  area.

10
Results of the Automated Convergence Verification
Program
The verification program considers the distance
from forecasted locations to the nearest point
within the watch box

• Many large distance errors in the location of the
  severe weather box indicate a high false alarm
  rate.
• When the convergence criteria for severe weather
  is met, but the verifying distance exceeds 1000
  km, then the result does not apply (N/A) as there
  may be synoptic weather from another system. This
  is also done for the lifted index indicator.
• Six criteria C1-C6 are chosen for the convergence
  calculation for location and vertical extent.
• When the convergence criteria is met the data is
  distributed into 6 accuracy categories Hits,
  lt100 km, 100-200 km, 200-300 km gt300 km and N/A,
  as described above.

11
The convergence is calculated at vertical levels
from the surface (10m wind), 1000, 925, 850, 700,
500 mb. Thus, if the convergence criteria is 6
vertical levels (e.g., C1), then all the vertical
levels must exceed the convergence amplitude
criteria.

• Criteria Convergence Number of Vertical
• Amplitude (10-5s-1) levels required to meet
• the criteria for severe weather
• C1 gt 3.0 6 (lt400 mb)
• C2 gt 3.0 5 (lt500 mb)
• C3 gt 5.0 6 (lt400 mb)
• C4 gt 5.0 5 (lt500 mb)
• C5 gt10.0 6 (lt400 mb)
• C6 gt10.0 5 (lt500 mb)
• For lifted index tests
• C7 lt -3
• C8 lt -4
• C9 lt -5

12
Excerpt showing convergence criteria and distance
form the watch box in km. 0 km means the
convergence criteria points were within the watch
box (Hits).
13
Total number of hits or distance from the watch
box for convergence exceeding the criteria for
all watch boxes. Same for Lifted index for C7
and C8 criteria. A six week period containing 47
watch boxes was used to gather the SPC Watch (no
Warnings were issued in this period) box
locations.
Criteria (km) C1 C2 C3 C4 C5 C6 C7 C8 Hits
0 27 0 25 0 13 17 7 lt100 4 0 3 0 2 0
3 3 100-200 5 0 3 0 1 0
9 13 200-300 14 0 11 0 7 0 6
7 gt300 13 12 11 6 13 6 6 12 N/A 11
8 19 16 24 28 6 5
14
Summary of Results

• The criteria C2, C4 were superior with the least
  distance errors compared to lifted index.
• Requiring the vertical extent of the convergence
  criteria to extend past 500 mb C1, C3, C5, does
  not improve the accuracy.
• We found the Best Lifted Index to have a larger
  number of false alarms and distance errors larger
  than 200 km.
• A six week period containing 47 watch boxes was
  used to gather the SPC Watch (no Warnings were
  issued in this period) box locations.
• Model initial conditions were set as the nearest
  such that 24-h forecasts to the valid times
  covered by the Watch box could be made.
• NOMADS OPEN-DAP/DODS queries can be used to make
  client (user) calculations.

15
Conclusion

• Model predicted convergence patterns have
  signatures that indicate the potential of severe
  weather using the SPC Watch and Warning Box
  product as verification.
• The volume or shape as measured by vertical
  extent and amplitude of the convergence pattern
  was used to test the accuracy of predicting the
  potential for severe weather.
• Convergence amplitude and vertical extent appear
  to be competitive with other stability measures
  like Best Lifted Index.
• Forecasters viewing the 3-Dimensional convergence
  structures of model forecasted convergence
  patterns could potentially improve their severe
  weather predictions.
• NOMADS OPEN-DAP/DODS servers can enhance user
  access to NOAA model database.
• Quantitative verification statistics will be
  improved to include false alarm rates and threat
  scores when more samples are used.

16
Real Time Vis5D files from NCEP Model Forecasts

• http//www.emc.ncep.noaa.gov/gmb/wd23ja/vis5d.html
  
• http//www.emc.ncep.noaa.gov/gmb/wd23ja/vis5d/avn3
  h
• for the GFS over the CONUS
• Western and Northern hemisphere add /avn
• South America add /avnsa
• Africa add /gfsaf
• Improved vertical resolution over the CONUS add
  gfsv
• Eta 212 grid http//www.emc.ncep.noaa.gov/gmb/w
  d23ja/vis5d/eta212
• Global Ensembles stacked in the vertical
  dimension
• http//www.emc.ncep.noaa.gov/gmb/wd23ja/vis5d/ens
  
• RSM over Hawaii at 10 km http//www.emc.ncep.noaa
  .gov/gmb/wd23ja/vis5d/rsmh10