Since
the NBM went to matching METARS, there have been some undesirable
side effects like frozen AWOS 6 hour Max/Min groups and bad Tds
causing unrealistic forecasts from NBM.
Here
is an example from KMZZ in Indiana, where the AWOS MaxT MinT 6 hour
group has been stuck at 87F High and -12F Low for a while. This
site has been blacklisted by the local WFO using the process via RTMA/URMA.
The
NBM v4.0 forecast 104F today at that location as a result of the
bias correction from the erroneous observation, with a low tonight
of about 10F. Until the owner of the AWOS has this issue repaired,
NBM v4.0 will continue to bias-correct to the bogus 99F diurnal range.
Recently,
Scott Scallion has worked on a method (in collaboration with EMCs
RTMA/URMA group) to check the URMA blacklisted observations and
remove them from use in NBM METAR matching. If an observation is
blacklisted in URMA, in v4.1 it will instead use the URMA grid point
(as it does for all none METAR grid points). If the site is
removed from the URMA blacklist, the METAR (or some of the RAW
sites) will be used again instead of the URMA value at the grid point.
This
change has recently been added to the real time NBM v4.1 parallel
runs, and you can see that KMZZ forecasts are starting to drift
toward more realistic levels. Highs are already 3-8F cooler and
lows 15-23F warmer in just a couple of weeks of having this running
in the real time tag. Obviously in these cases it will take at
least a couple of months for the NBM forecast values to become
realistic again, but this does allow us to directly use the
URMA blacklist to make sure NBM v4.1 stops using METARs and RAWs
sites that are selected by local WFOs as having unrepresentative or
erroneous values.
JPC
Jeff Craven
Chief,
Statistical
Modeling
Division (SMD)
Vice
Lead,
Innovation,
Science &
Technology
(Evolve PMO)
National
Weather
Service, W/STI-12
Meteorological
Development
Laboratory (MDL)
Room
10410, SSMC2
Silver
Spring, MD 20910
(816)
506-9783 cell/text
(301)
427-9475 office
@jpcstorm
