EMC/HPC/MPC/NCO/CPC Synergy Meeting Highlights 04/28/03


This meeting was led by Peter Manousos and was attended by the following: Eric Rogers, Hua-Lu Pan, John Ward, Jim Hoke, Larry Burroughs, Zoltan Toth, Kevin McCarthy, Pete Caplan, Dave Caldwell, Steve Jascourt, Keith Brill, Bill Bua, Ed Danaher, and Joe Sienkiewicz.


1. IBM SP and CCS

Fire Weather (8km NMM) runs are currently running on the CCS. John Ward reported that by May 12th the CCS will be ready to be declared operational if the current 30 day test goes well. At that time the GFS output will be ready about 10 minutes earlier compared to its current availability. The GFDL will then be upgraded in vertical resolution from 18 to 42 levels and run out to 120 hours. John also reported that there will be some changes in the manual QC of observational data on May 1st. The routine daily QC support provided by Data Management Branch will be determined. The SDMs will still continue a level of manual QC sufficient to prevent egregiously erroneous data from impacting the forecast models. For special projects, like Winter Storm & Hurricane Recon, the SDMs will QC as much data as possible.



2. Notes from EMC

a. Global Modeling Group: Hua-Lu Pan reported that the long wave radiation package is being tested in parallel. This package has no discernable impact in the troposphere , but tends to make the stratosphere cooler as compared to the current version of the GFS. This may have an impact on the assimilation of satellite radiances in the GFS analysis. Also, half degree GFS output will be available out to 36 hours per requests of AWC. However, due to bandwidth limitations, this will only be available to the NCEP Service Centers. While a variety of minor modifications are being or will be worked on, no major model changes with anticipated major forecast impact are anticipated until 2005.


b. Mesoscale Modeling Group: Eric Rogers reported that spring implementation of the EtaX has been moved to a yet-undetermined date perhaps as late as June 1. Subsequent test runs on the Eta centering around the physics packages will not occur until fall.


c. Global Ensembles: Zoltan Toth reported that the 4x daily 10 member configuration, along with the long-planned resolution increase to T126 out to 180 hours (presently T126 to 84 hours) will be implemented during the summer. Zoltan also reported that performance, production, post processing, and verification issues of a joint Canadian Meteorological Center/NCEP ensemble package will be discussed with the CMC at a conference later this week. A major goal is to develop products that remove model biases of mean and variance.


d. Short Range Ensembles: Eric Rogers reported that an additional SREF system will still be tested using an ensemble of different physics packages this spring.


e. Marine Modeling and Analysis Branch (MMAB): Larry Burroughs reported that the

NWW3 wave model output is set to be run once the GFS starts being produced on the CCS 4/day out to 168 hours. Further, the Eastern Pacific Hurricane wave model will be implemented when the updated version of the GFDL Hurricane Model is implemented (pending operational declaration of the CCS). Testing with AWIPS bandwidth will begin and some data may be available with the install of AWIPS version OB2 (slated for some time near August). Finally, visibility fields are slated to be implemented in early summer of 2003.



3. Input to EMC.

The Global Modeling Group received a question on the occurrence of grid scale feedback in the GFS. Hua-Lu acknowledged that the problem and reiterated that the GFS and many other operational models are not able to properly handle this kind of mesoscale process. It is Hua-Lu’s impression that this would not improve until the model’s horizontal resolution is on the order of 10km or less, so that mesoscale flows associated with MCSs can be resolved. At coarser resolution, the model still feels the forcing but instead responds at too large a scale. Further, tuning this aspect of the GFS may negatively impact the model’s performance handling tropical systems. Stephen Jascourt then noted the following:

         All models produce precip “bombs” occasionally

         Mass flux convective parameterizations, such (as in the AVN and also the KF parameterization), are particularly susceptible, but have other advantages

         The impact of the precip “bomb” on low development and precip at later times downstream depends on model resolution compared to the Rossby radius of deformation: small grid spacing as in the 12-km Eta model results in small disturbances that disperse away with little lasting influence; coarser grid spacing as in the GFS result in larger-scale disturbances that may last longer and have greater forecast impact except not as much in the tropics due to the longer Rossby radius. Also, since the forecast impact is related to the impact of the PV anomaly created by the precip “bomb”, the impact will be stronger when there is a baroclinic zone to interact with the PV anomaly and the PV anomaly is wide enough to result in parcel displacements a good distance across the baroclinic zone (e.g. up the isentropes - warm advection). The PV anomaly width, again, is a function of the radius of the precip “bomb”, which depends on model resolution.

         Improving resolution may not decrease the occurrence of precip “bombs”, but it will decrease the areal extent of the problem and their adverse impact on the forecast downstream at later times


4. Next Meeting Proposed Monday June 2nd, 2003 at noon in room 209.