Transcript cleveland_nwa

Updated Thunderstorm Probability
Forecast Guidance in Support of
Gridded Model Output Statistics
NWA Annual Meeting, Cleveland, OH
October 16, 2006
E-mail: [email protected]
Current State of Ensemble MOS



MOS equations are applied to all 15
members of the GFS ensemble and the
operational run to create 16 text
messages
• Local applications like getENS
generate summary values
The current gridded MOS is an example
of a lagged ensemble.
• Temperature, dew point, max and
min temperature grids are created
from two runs, weighted 2:1
Future ensemble MOS requirements are
vague
Image from PSU web site
Enhanced-Resolution MOS
Support for the operational forecasters


Two Approaches
“True” Gridded MOS
• Observations and forecasts valid on fine grid
• Uses remotely-sensed predictand data
 cloud-to-ground lightning data
“MOS at any point”
• Emphasis on high-density surface networks
• Uses high-resolution geophysical data
Remotely-Sensed Lightning Data



NLDN cloud-to-ground lightning observations define a
thunderstorm event
• Observations are the predictand
Monthly lightning climatologies were developed from ten years
of data for 3-, 6-, 12-, and 24-h time periods, to be used as
predictors
Observed lightning data are used to
verify the thunderstorm guidance
Creating the Gridded Lightning
Datasets from Observations
Strikes are summed over the appropriate time period and assigned to
the center of the grid boxes
××
××
××
×
×
×
×
××
××
×
= thunderstorm
= no thunderstorm
Probabilities for Different Grid Resolutions
40-km 21-24h period
10% contour interval
20-km 21-24h period
10% contour interval
Probabilities for Different Time Periods
As the resolution increases in time and space, the magnitude of the
probabilities will decrease as the likelihood of an event at an exact time
and point in space approaches zero
MOS Thunderstorm Probability Guidance
Given that:
• The area covered by the forecast should be at least as large as the
event
• Thunderstorm forecast probabilities traditionally represent an area
within 25 miles of a point, and
• We don’t want the probabilities to become so small they lose
information and credibility…
The gridded MOS thunderstorm probability is defined as the
probability of one or more cloud-to-ground lightning strikes in
a 40-km square grid box during a given time period.
So:
• The guidance is interpolated to the 5-km NDFD grid using bilinear
interpolation
• It represents the likelihood of a cloud-to-ground lightning strike
40-km around each 5-km grid box, and
• Stations have the value of the 40-km grid box they are sitting in,
using nearest neighbor assignments, to fill in the text messages
Thunderstorm Equation Development


Equations developed from 5 years of GFS model output
All 40-km “stations” were combined into one region
• increased sample size and eliminated boundaries



Separate equations were developed for 3 seasons
• Spring: March 16 – June 30
• Summer: July 1 – October 15
• Cool: October 16 – March 15
3-h time periods through day 3; 6- and 12-h time periods through
day 7.
Important Predictors
• Stability Indices
• Convective precipitation amount
• Product of vertical velocity * relative humidity
• K-index * thunderstorm relative frequency
• Thunderstorm relative frequencies
Gridded MOS Thunderstorm Products




NDGD web-based
graphics
Distributed in GRIB2
format
3-,6-, 12-h forecast
periods
Available in AWIPS
with OB7.1 release, as
part of the gridded
MOS system
“Enhanced-Resolution” MOS


“True” gridded MOS
• Uses remotely-sensed data, i.e. the probability of a thunderstorm
“MOS at any point”
• Emphasis on high-density surface networks; Co-Op, buoy, mesonet,
etc
• Uses high-resolution geophysical data
• Analyze forecasts on high-resolution grid
• Traditional MOS downscales the model output to specific sites
• Examples of weather elements






temperature
dew point temperature
max/min temperature
wind speed and direction
relative humidity
probability of precipitation (6- and 12-h periods)
Gridded MOS Objective
Generate guidance sufficient for forecast initialization at WFOs
Forecaster Feedback is Invaluable
Gridded MOS wind speed problem
•Strange dots
Bad “if test” in a wind postprocessing subroutine created
calm winds, when the
direction was near 0.
•Too many calm mountain winds
A control file setting caused
excessive calm winds over
the mountains
20060921 1200 UTC
21 hour forecast
Forecaster Feedback is Invaluable
Thank you
The subroutine and control files have been fixed and implemented
20060921 1200 UTC
21 hour forecast
20060921 1200 UTC
21 hour forecast
Summary




Gridded MOS thunderstorm probabilities are valid over a
40-km area, for each 5-km NDFD grid box
Gridded MOS is a work in progress
• more elements are coming
• improvements to current algorithms and equations
expected
• increased resolution from 5-km to 2.5-km
User feedback is critical to define requirements and
improve the products
Alaska and other OCONUS areas are in the early stages of
development
Link to documentation and products
http://www.weather.gov/mdl/synop/gmos.html