SR SSD 2001-06

Technical Attachment

Providing Local Modeling Support for Costa Rica

Shawn Bennett, SOO
WFO Brownsville, Texas

This report will summarize the results of a collaborative project involving international support provided by WFO Brownsville. In June, 2000 I traveled to Costa Rica as part of a WMO-sponsored effort to participate in WAFS-STAR4 refresher training hosted and coordinated by the Regional Committee for Hydrological Resources (CRRH) and the National Meteorological Institute (IMN) in Costa Rica. During my visit, IMN's forecast operations manager, Mr. Werner Stolz, asked me if I knew of a high-resolution model that could be used to assist forecasters in Costa Rica and Central America. They had been attempting to use MM5 and wanted to try something else. I suggested using the newly released NCEP workstation ETA (WSETA). That model had been running at WFO Brownsville in south Texas since August, 1999, and we had attained some experience and expertise. The idea was discussed with Eladio Zarate, director of IMN. As part of what we characterize as Phase 1 of our project, I offered to attempt some experimental runs of the WSETA model at the WFO, but focusing on a domain over Central America centered on Costa Rica. If successful, Phase 2 of our project would seek to implement the WSETA model at IMN.

By August, 2000 several runs of the WSETA model were completed over a large scale Central American domain and over a smaller domain centered on Costa Rica. The model runs were done on the WFO Brownsville HP workstation (HPC360) at 10 km resolution with model forecast grids generated every 3 hr out to 36 hours. I found that given the computer resources available, the smaller Costa Rica domain was feasible in an operational setting. The 36 hr forecast run was completed in time to provide at least 24 hr of useable forecast information. WSETA model runs over the larger Central American domain took too long to complete to provide useful output. In order to complete model runs for all of Central America, the domain would have to be split into at least two parts.

The WSETA generates output in the GEMPAK grid format. IMN made available a server and I uploaded the model grid files for viewing and evaluation by IMN meteorologists. IMN had hoped to view the WSETA forecast grid output using the GEMPAK software included in their WAFS-STAR4 system, but they were unsuccessful because of unresolvable incompatibility problems. To work around this problem, some of the basic output gridded fields were converted to GIFs which were uploaded to the IMN server. This is a rather cumbersome solution but workable on a temporary basis.

In late summer of 2000, WFO Brownsville received a new Dell dual-processor 800 MHZ Pentium machine running the LINUX operating system. This machine is now being configured to run the WSETA and view its output with the newly released AWIPS D2D for LINUX. It is expected IMN will acquire a similar LINUX-based machine to run the WSETA and view its output. Once the computer is acquired, I plan to work with IMN to install the WSETA and AWIPS D2D software for LINUX. In Phase 3 of this project IMN, as a leader in the region, would like to distribute the WSETA output to the other offices in the region, but a means will be required for doing so. Most likely this will consist of a combination of efforts, including the Web and distribution of the model grid files via WAFS-STAR4.

In summary, Phase 1 of this international collaboration has been completed. The WSETA will run successfully over the region. NCEP has helped by moving the southern border of the model grid domain farther to the south, thus eliminating some of the model boundary value issues. Phase 2 will begin once a suitable computer is acquired by IMN. During Phase 2 I will assist IMN in setting up and installing the WSETA model and AWIPS D2D for LINUX software. Work on Phase 3 can begin once the WSETA is up and running in Costa Rica.