Transcript Consortium Meeting 2/16/2006 - Washington State University

Update on the Regional
Modeling System
NASA Roses Meeting
April 13, 2007
Supported by the Northwest Modeling
Consortium…the regional modeling effort centered at
the UW is
• Running the MM5 at 36, 12, and 4 km resolution
• Running the new WRF model at 36, 12 km and 4
km resolution
• Running TWO high resolution regional ensemble
systems to provide probabilistic forecasts and data
assimilation
• Gathering all local weather observations from
dozens of networks. Plus quality control.
• Running a wide range of weather applications
dealing with air quality, hydrology, transportation
weather and fire weather.
36 km
12 km
4 km
A Few of the Major Efforts
and Improvements this Year
Evaluation of WRF
• WRF: the Weather Research and Forecasting
Model, is the replacement for the MM5 and
should be the national mesoscale model used by
both the operational and research community.
• We are now running it at 36, 12 and 4 km grid
spacing, and recently replaced our previous test
version with the newest version with major
physics improvements and nudging on the outer
domain.
Evaluation
• The new real-time version of WRF will be
evaluated this spring and for major
historical cases.
• If equal or superior, the modeling
consortium will probably ok the shift from
MM5 to WRF.
• Evaluations of old version show differences
but not superiority.
New UW WRF
Example of Old MM5 vs WRF
Verifications
March 1
Convergence
Zone Event
The UW Quality Control System
• A major task continues to be the gathering
of all real-time observations of the region
into one place
• Right now we acquire over 60 networks in
real time for displaying on our web site,
verification, and many other uses
• Quality Control is essential for such a
heterogeneous network of networks.
The UW Quality Control and
Warning System
• We have developed an advanced QC system
suitable for an area of complex terrain (Jeff
Baars)
• Have also created an automated QC display
system that one can check on the web and
which can automatically tell the manager of
a network when their data is suspect (David
Carey)
Direction QC
Satellite and Other Data Assets
• We are acquiring a substantial number of
satellite and other observational assets.
• These include cloud and water vapor winds,
satellite radiances, microwave and
scattermeter data.
• Also ACARS data.
Regional Data Assimilation: The
Next Major Challenge
• Until this point, the high resolution regional
prediction effort has been based on cold starts
using grids from other model centers for IC and
BCs.
• Key area for improvement is to begin our own
assimilation of observation assets to determine
whether we can improve initialization and
forecasts over the Pacific, and to better initialize
mesoscale structures over land.
• The EnKF work, spearheaded by Greg Hakim,
offers a potent approach for such assimilation.
Local Data Assimilation
• A major new effort has begun to assimilate all
local observations to create a physically consistent
three dimensional picture of the regional
atmosphere.
• Needed for many reasons…including better shortterm forecasts and for air quality studies which
demand descriptions of the 3D state of the
atmosphere.
• The current UW approach makes use of a 90member ensemble system--Ensemble Kalman
Filter (EnKF)--probably the best approach
possible for using the forecast model to use local
observations
EnKF Data Assimilation
• Ryan Torn and Greg Hakim developed an initial
EnKF system at 45 km grid spacing using 90
members.
• Has shown great promise.
• This has led to a joint project between
Hakim/Mass groups to replace the old EnKF
system with a truly mesoscale (36/12 km)
version…which is now running (Brian Ancell
lead)
Local Data Assimilation
• The system produces 90 different analyses that can
be combined to produce the best guess at what is
there and tell us the uncertainty in the analyses.
• These analyses can be integrated forward in time
to give us probabilistic predictions of the future
• We now have it running at 36 and 12 km
resolution…
Regional Data Assimilation
• The system is easily capable of dealing with
many traditional (ASOS, sounding) and
non-traditional (ACARS, cloud and water
vapor track winds, radar winds) data types.
• Can be extended to radiances and other
remote sensing types if we acquire the
necessary forward models.
• Will be the basis for the NASA Roses work
The END
Map Selection Interface
• We have a new map
interface for getting
soundings, time
height cross
sections and
meteograms at ANY
location.
• Done by Phil
Regulski