All,
This is a final notice of a major change in some of the physical process
parameterization in the NCEP Medium Range Forecast (MRF) model. The forecast
suites that will be affected are the AVN, the MRF, and the global ensemble
suites. The planned date of implementation is 15 May 2001 (1200 GMT). We have
made extensive internal evaluation of the changes as well as NCEP-wide
evaluation. The attached Notice of Intent to Change includes Web sites where
daily forecast maps and verification statistics can be found. A good place to
read about the changes is the draft Technical Procedure Bulletin at :
http://sgi62.wwb.noaa.gov:8080/tpb97/CLW/html/clw.html
The major changes are the following:
1. We are adding a prognostic cloud water/ice process parameterization into
the model. We currently use the model relative humidity to estimate the cloud
amount. The new procedure explicitly models the condensation and precipitation
processes. In addition, the cloud optical depth and emissivity used in the
radiation calculation is now computed from the cloud water/ice which should
make the cloud radiation feedback process more realistically modelled. The
detainment of water/ice from the cumulus clouds is also included in the
parameterization. This is an important source of cirrus in the tropics.
2. We are including a cumulus momentum mixing algorithm in the cumulus
parameterization scheme. The AVN/MRF model has been plagued with false alarm
storms in the tropics for several years and the addition of this process
greatly reduces the occurrence of false alarm storms. A crude form of the
parameterization of the pressure effect on cloud parcels is included in this
version.
3. We are modifying the cumulus cloud parameterization scheme to simulate the
effect of an ensemble of clouds. This is accomplished by randomly selecting
cloud tops from a range of possible heights at each time step. The current
procedure led to excess drying of the mid and upper troposphere and the new
procedure improves the water vapor climate of the model forecasts.
The major results of the new package are:
1. Significant improvement in skill for forecasts of atmospheric circulation in
the medium range (3-7 days) and the extended range (7-10 days). Some bias
toward less-than-observed wind speeds in mid and upper levels in the
extratropics and low levels in the tropics.
2. Significant reduction of the false alarm predictions of the tropical
storms. Analysis improvements have led to a reduction of false alarm storms in
the model initial states. The 3-5 day AVN/MRF forecasts continue to generate
many false alarm storms presently. In our extended period of parallel runs and
retrospective runs of the model, we noticed very few false alarm storms in the
forecast. The real storms are still predicted by the model but with weaker
intensity than the current model predictions.
3. Improvement of tropical circulation forecast skills. The changes in
convection and cloud water/ice algorithms have led to a reduction of noises in
the forecasts in tropics. This has led to improved skill scores.
4. The lower atmosphere is colder in general. The long-standing warm bias
near the surface in the West has been replaced with a cool bias. We are
working to reduce the biases in a follow-up implementation.
5. The wind speeds aloft in the forecasts have small bias in the first two
days and develop a negative bias with time.
6. The precipitation forecasts in the U.S. show a slightly larger wet bias
than the operational for small amounts (< 0.1"), but the overall totals are
similar.
Please feel free to contact me if you have any question about this set
of changes.
Hua-Lu Pan
Chief, Global Modeling Branch, EMC, NCEP