SOO/STRC Workstation Eta User's Guide

I. SOO/STRC Workstation Eta, What is it?

The SOO Science and Training Resource Center Workstation Eta package is a complete, full-physics, NWP system, nearly identical to that run operationally at the National Center for Environmental Prediction (NCEP). However, the installation, configuration, and running of the SOO/STRC version have been greatly simplified from the NCEP release to encourage its use by forecast offices and universities. The SOO/STRC WS Eta is easy to run on most UNIX and LINUX workstations; it should possible for those with limited modeling experience to have the model running in less than 1 hour. In addition, the WS Eta code is very efficient, which should be an advantage for real-time forecasting purposes.

II. Why Develop and Support a NWP Package for the WFOs?

The Science Operations Officer/Science and Training Resource Center (SOO/STRC) version of the workstation (WS) Eta modeling package was developed to promote the local use of numerical weather prediction models in the Weather Forecast Offices (WFOs) and to achieve the following three goals set by the SOO Science and Training Resource Coordinator (SOO STRC):

1. To improve the knowledge and use of NWP models and issues at the local level

2. To advance the forecasting process through an improved understanding of mesoscale processes and the use of non-traditional diagnostic tools

3. To increase participation within the WFOs in developing and executing NWP studies to examine local forecast problems

Running the SOO/STRC WS Eta locally at the WFOs will serve to provide 1) numerical weather prediction (NWP) guidance to National Weather Service (NWS) and River Forecast Center (RFC) forecasters at temporal and spatial scales not available from operational data sources, 2) a powerful tool for studying local forecast problems and historically significant weather events, 3) an alternative to the configuration and physics of operational systems, 4) a means to develop and test new diagnostic forecast techniques, and 5) a method of training forecasters on NWP-related issues.

IV. STRC Comments about the WS Eta Package

The SOO/STRC Workstation Eta is nearly identical to NCEP's WS Eta and operational Eta Models. The primary difference is its easy-of-use that comes from the run-time scripts and organization of the source code and various data files. Model source code remains the same as the NCEP release except for the inclusion on the Kain-Fritsch cumulus scheme and the addition of output fields. No compiler is necessary since the SOO/STRC maintains binary distributions for linux and HPUX.

The SOO/STRC Workstation Eta package takes previous model output in GRIB format to serve as the initial and lateral boundary conditions for your model run. Most of these data sets are available in real-time from the NCEP or OOS data servers. Historical cases can be run from data available on the SOO/STRC data archive or the Reanalysis project.

The WS Eta includes an eta/sigma coordinate option, a non-hydrostatic option, and a choice of Kain-Fritsch or Betts-Miller-Janic cumulus schemes. It also has nesting capability. The code if VERY efficient, running 25-33% faster than a similarly configured version of MM5 on 1 CPU. The model is also very scalable with up to 90% performance gain going from 1 to 2 CPUs. If you are looking for a machine to run the model on I strongly encourage you to obtain a multi-CPU workstation. Lots of performance gain for the money. The model has been ported to HPUX and LINUX operating systems. I suspect it will run under any OS but you will have to make some changes to the Makefiles if you compile on a different system. Binaries are provided for HPUX and LINUX.

Running the model for either real-time forecasts or local case studies requires that you configure and run 3 different run-time scripts, eta_download.csh, eta_run.csh, and eta_convert.csh, in that order. The names are self explanatory. All the variables you need to configure are documented in these scripts. There is a 4th script that can be used for real-time forecasts that runs the 3 run-time scripts and tells you if you were successful.

A utility, etamap, has been provided for you to get a better idea as to the computational workstation ETA model domain. It is strongly recommended that you use this program to establish the areal coverage of your model runs. The program is based on the GEMPAK program, GPMAP, but takes the the parameters used to define your model domain.

V. What's new for version 3.1 of the SOO/STRC WS Eta package

a. Version 3.1 is nearly identical to current incarnation of operational Eta run on NCEP's IBM

b. Nesting capability is now available thanks to Paul Wolyn of PUB who requested it.

c. 14km high resolution lake surface temperatures are now available over the Great Lakes - Great for lake-effect studies. See a the difference between the 14km high resolution and 0.5 degree fields here. Note that the 14km data are responsible for the cooling over Lake Michigan and warming to the east.

d. Improved 30 second Land-Sea mask replaces the 2 minute data set over North America

e. NCEP Reanalysis grid support for simulating your favorite historical events going back to 1949. Recreate that 6 " snowstorm that seemed like 10' when you were 4 years old.

f. Support for 12-km AWIPS 218 tiles on the NCEP ftp server. Great for real-time modeling!

g. Full resolution BUFR files for ANY location you want. Just add the station information to a file and the sounding data are yours!

h. Additional output fields including such favorites as cloud base vertical mass flux and fire potential.

i. Vertical resolution of the output grib files has been increased to 25 mb.

j. The SOO/STRC WS Eta version 3.1 run-time scripts have been completely rewritten to accommodate the new data sets, file names, and directory structures used on the remote data servers. All that remains from previous versions are the data set request configuration and a few conjunctions. The scripts are more "user-proofed" to avoid the anxiety that often accompanies complete failure.

k. There is a new fail-over option (PREVIOUS) that will initialize a real-time model run from a previous 6-hour forecast should the current cycle data not be available. For example, if the 00 hour forecast from the 12 UTC cycle run is not available from the NCEP ftp server by a specified time, the 18 UTC cycle run will be used for initial and boundary conditions beginning with the 06 hour forecast (that would still make your 00 hour forecast at 00 UTC).

l. More love notes! If the send mail option is configured in the eta_autorun.csh script you will now receive a wider variety of email messages telling you that your real-time model run failed. Oh the Joy!

m. Data sets used to initialize your runs now include the 12 and 32km tiles, 32km 221, 40km 212, and 90 km 104 grid from operational Eta model runs, the AVN 0n the 1 degree 2 grid, 2.5 degree reanalysis data set going back to 1949, and previous WS Eta model runs.

n. Raw model output can be processed into BUFR or GRIB-formatted files. The BUFR data can be further processed for viewing in NSHARP or BUFKIT while the GRIB files can be converted into GEMPAK format for display in NAWIPS, viewed with GrADS, or pushed into AWIPS and IFPS. It's all in there.

o. The Ws Eta can not over-state profits since the intellectual proceeds achieved by running a local area model are infinite.

XIV. Scheduling Cronjobs for Real-time Model Forecasts

The eta_autorun.csh script make the task of setting up and running a real-time forecasting system much easier. The script has a number of fail-over options to make sure the model run is completed and the forecast makes it to your forecasters. I encourage you to review and configure the parameters in the eta_autorun.csh file.

Here is an example of a crontab file entry that I use for my 4xday WS Eta model runs. Note that the times are stated in Mountain Daylight Time. Also note that prior to each run I execute the eta_clean script to clean up the run-time directory.

# Realtime WS Eta model entry for the 06 UTC cycle run
38 02 * * * /usr1/worketa/bin/scripts/eta_clean 3 >> /tmp/wseta_realtime06.log 2>&1
40 02 * * * /usr1/worketa/run_time/eta_autorun.csh 06 >> /tmp/wseta_realtime06.log 2>&1

# Realtime WS Eta model entry for the 12 UTC cycle run
38 08 * * * /usr1/worketa/bin/scripts/eta_clean 3 >> /tmp/wseta_realtime12.log 2>&1
4 0 08 * * * /usr1/worketa/run_time/eta_autorun.csh 12 >> /tmp/wseta_realtime12.log 2>&1

# Realtime WS Eta model entry for the 18 UTC cycle run
38 14 * * * /usr1/worketa/bin/scripts/eta_clean 3 >> /tmp/wseta_realtime18.log 2>&1
40 14 * * * /usr1/worketa/run_time/eta_autorun.csh 18 >> /tmp/wseta_realtime18.log 2>&1

# Realtime WS Eta model entry for the 00 UTC cycle run
38 20 * * * /usr1/worketa/bin/scripts/eta_clean 3 >> /tmp/wseta_realtime00.log 2>&1
40 20 * * * /usr1/worketa/run_time/eta_autorun.csh 00 >> /tmp/wseta_realtime00.log 2>&1