 |
Running The SOO/STRC WS Eta |
1. Environmental variables you should know and love
Once you have gone through the process of installing the WS Eta on your workstation it is almost time to run the model and generate some forecasts, but first you will need to learn some environmental variables:
$WS_ETA Top level of the WS Eta distribution. Typically /usr1/worketa. $ETA_RUN worketa/run_time - Where you run all your simulations. Typically /usr1/worketa/run_time. $ETA_SCRIPTS Location of all the C-Shell scripts in the SOO/STRC distribution. Typically /usr1/worketa/bin/scripts. $ETA_PREP Location of all the grib files waiting for processing into initial and boundary conditions. Typically /usr1/worketa/data/eta_prep. $ETA_BIN Location of all the ws_eta binaries. Typically /usr1/worketa/bin/{OS} $ETA_LOGS Location of all the output log files. Typically /usr1/worketa/logs $ETA_DOCS Location of all the WS Eta documentation files. Typically /usr1/worketa/docs $ETA_TMP Location of all the temporary files. Typically /tmp
2. The run-time scripts
You should first know that all model runs are to be executed from the $WS_ETA/run_time directory, which is pointed to by the $ETA_RUN environmental variable. So again, all model runs begin in $WS_ETA/run_time:
% cd $WS_ETA/run_time
In $WS_ETA/run_time you will see 4 C-shell scripts, eta_download.csh, eta_run.csh, eta_convert.csh, and eta_autorun.csh. These 4 files are known as the "run-time" scripts and are actually links to files in the $ETA_SCRIPTS directory. These will probably be the only 4 scripts you need to edit to run the model successfully. Each of these scripts is documented with the configurable parameters described. Here is a brief summary of each file:
eta_download.csh
Script to download and process the grib files off the NCEP, OOS, Regional, or SOO/STRC ftp servers. Also processes the data for local case studies.
eta_run.csh
Script that defines your model domain and run configuration. Also generates the topography file, boundary and initial condition files and then starts the actual integration of the forecast.
eta_convert.csh
Script processes the raw WS Eta model output into GRIB and BUFR formats. It also takes the final grib and bufr files and converts them into a variety of formats for viewing with software such as NAWIPS and BUFKIT. It will also ftp the data files to other machines if you desire further processing such as in AWIPS.
eta_autorun.csh
Primarily intended for real-time forecasts, the eta_autorun.csh file executes each of the above 3 scripts in sequence. There are a number of useful user-configurable features available for executing real-time model runs. If you plan on running the WS ETA in real-time then the eta_autorun.csh should be used.
In addition to the run-time scripts, there is another useful script located in your $ETA_SCRIPTS directory called eta_clean. The eta_clean script is run to purge the run-time directories of unwanted files. Here is a summary of the arguments passed to eta_clean and what they do:
eta_clean 0
Clean out just the WS Eta log files in $ETA_LOGS
eta_clean 1
Return the run_time directory post-run state. Just like it looked after you ran the model but before you ran eta_convert.csh.
eta_clean 2
Return the run_time directory pre-run state. Just like it looked after you ran the eta_download.csh script but before you ran eta_run.csh. Retain all the files in the $ETA_PREP directory.
eta_clean 3
Return the run_time directory pre-download state. Just like it looked after you ran the eta_download.csh script but before you ran eta_run.csh. This argument is especially useful to run immediately prior to running a real-time simulation.
eta_clean 4
Passing this argument will return the WS Eta to the state just after you unpacked the tar file. It will also overwrite any changes you have made to the run-time scripts. You now have the power. Use it wisely.
3. Making a forecast
There are 3 steps in making a SOO/STRC WS Eta model forecast:
a) run eta_download.csh
b) run eta_run.csh
c) run eta_convert.cshThe above sequence may be accomplished for real-time simulations with the eta_autorun.csh script; however, you will still need to configure each of the above files.
Step 1: run eta_download.csh
You will have to run eta_download.csh regardless of whether you want to make a real-time or case study simulation. If you are running a case study then you will have to place all the grib files serving as your initial and boundary conditions in $ETA_PREP. There are couple of configurable parameters at the beginning of this file. Gif images of the areal coverage for each grib file grid used to initialize the WS Eta can be found in $ETA_DOCS.
The following guidance is also provided by passing "-help" to eta_download.csh, i.e,
% eta_download.csh -help
This script will download requested files from the NCEP, OOS, Regional, or SOO/STRC ftp servers and convert them into a format suitable for processing by the workstation ETA preprocessor.
The script takes the following arguments:
% eta_download.csh GRIB CYCLE INIT FINAL FREQ SITE YYMMDD
Where:
GRIB identifies the model grid to download for the initial and boundary conditions. The choices include:
Note: Gif images depicting each of these domains is available in $WS_ETA/docs
104 ETA Model on a 91km domain 212 ETA Model on a 40km domain 221 ETA Model on a 32km domain TILE32 ETA Model 32km Tiled data TILE12 ETA Model 32km Tiled data AVN 1 degree Global AVN Model data set RNL 2.5 degree Reanalysis data WSE Grib files from a previous Ws Eta Model run CYCLE is the initialization time of the ETA or AVN model run, usually 00, 06, 12, or 18 (UTC). The exception would be if you are running a nested simulation in which case you may begin the nest at any time.
INIT is the first forecast hour from the selected ETA or AVN model run to download. Note that this time does not have to necessarily be a 00 hour forecast. You can initialize a workstation ETA model run from a 06, 12, or 24 hour forecast.
FINAL is the final forecast time to download from the selected NCEP model run. This time will determine the maximum length of a workstation ETA forecast.
FREQ is the frequency of the ETA or AVN data to download between the INIT and FINAL hours. These data are to be used as boundary conditions for the workstation ETA forecast. It is strongly suggested that 3 hourly data be used since this is the frequency of data on the servers. If you are running a nested simulation then this value should be 01.
SITE is an flag indicating whether this is a REAL-TIME (R, N, or O) CASE STUDY (S or C), or BENCHMARK (B) run.
R: If R is selected, then the data will be obtained from a REGIONAL ftp server if available.
N: If N is selected, then the data will be obtained from the NCEP ftp server.
O: If O is selected, then the data will be obtained from the OOS ftp server. Note that only a select few data sets are available from the OOS site.
S: If S is selected, then the data will be obtained from the SOO/STRC ftp server. The server maintains a 7-day running archive of operational Eta model runs for case studies. Note: You will need access to the SOO/STRC server.
C: C is the case study option. If C is used, then the necessary grib and surface files are assumed to already reside in the $ETA_PREP directory. You can use this option with the archived ETA model data available from the SOO/STRC ftp site and the SOO/STRC coordinator. Note that the SST and snow cover data must also reside in the $ETA_PREP directory. All NESTED simulations are run as case study events. You must include YYMMDD for a C runs.
B: This option will process a pre selected data set located in the until directory for testing and benchmarking the workstation Eta. If B is passed, all other arguments will be ignored and 24 hours of the 32km tiled grid will be processed.
YYMMDD is the 2-digit year, month, and day of the data you wish to process for your WS Eta run. NOTE that this argument is OPTIONAL. Failure to include it will cause the script to use the current date. This option is use when running a case study or downloading earlier data from the SOO/STRC archive.
EXAMPLES
% eta_download.csh AVN 06 06 36 03 N
will download and process the first 30 hours of the current days AVN forecast from the most recent 06 UTC model run beginning with the 6 hour and ending with the 36 hour forecast at 3 hr intervals. Files will be downloaded from the NCEP ft0p server.
% eta_download.csh TILE32 18 00 36 03 S 020603
will download and process the first 36 hours of Eta forecasts on 32km tiles from the 03 June 2002 18 UTC run beginning with 00 hour and ending with the 36 hour forecast at 3 hr intervals. NOTE: You must have access to the SOO/STRC ftp server.
% eta_download.csh RNL 12 00 24 06 C 501124
the C in the above example indicates case study and that the data already reside in the data/eta_prep directory. In this example the data are from the reanalysis data set. The user want to initialize a 24 hour run at 12 UTC 24 Nov 1950. In most cases the frequency of the Reanalysis data are only 6-hourly. NOTE that due to the spatial resolution it is advised that you start you simulation well before the period of interest. In addition, to minimize the impact of the 6-hourly boundary conditions you should make your domain as large as possible. And have a nice day.
% eta_download.csh WSE 00 06 18 01 C 501124
will processes 12 hours of a previous Ws Eta model run beginning with the 06 hour and ending with the 24 (18 + 6) hour forecast at 01 hour intervals. For this event you would be running a nested simulation. NOTE: You must include YYMMDD for all case study and nested runs.
Or for the benchmark case:
% eta_download.csh B
will process all data needed for the WS Eta benchmark case. The data are actually included in your SOO/STRC distribution and you are encouraged to run the benchmark case for your first forecast. Please see benchmarking your system for more information.Step 2: run eta_run.csh
As stated above , the eta_run.csh script that defines your model domain and run configuration, generates the topography file, boundary and initial condition files and starts the actual integration of the forecast. Most everything you need to configure your model run, whether it's domain location, grid spacing, model physics, model dynamics, or forecast length, it's in eta_run.csh. The file contains full documentation of each of the parameters along with suggested settings. You are strongly encouraged to read the file before running for the first time, but that has never convinced anybody before.
The eta_run script is typically run without arguments:
% eta_run.csh
However, there are 2 arguments that can be passed to eta_run.csh:
% eta_run.csh B
Will run the benchmark run. Again, read benchmarking your system for more information.
% eta_run.csh N
Will execute the nested run using the nest_ binaries in the $ETA_BIN directory and the nest_ variables found in eta_run.csh. You should have already run and processed the coarse mesh grid from an earlier WS Eta model forecast to serve as initial boundary conditions. It is imperative that you output your large grid simulation hourly to provide the best temporal resolution for your fine mesh run.
Step 3: run eta_convert.csh
The eta_convert.csh processes the raw WS Eta model output into GRIB and BUFR format. It also takes the grib and bufr files and converts them into a variety of formats for viewing with software such as NAWIPS and BUFKIT. It will also ftp the data files to other machines such as for processing in AWIPS. Everything you need to know to successfully post-process your WS Eta model run is documented in this script.
The eta_run script is typically run without arguments:
% eta_convert.csh
If eta_convert.csh is run without arguments then the post processing will be done as defined by your configuration of the various parameters in the file. The exception to this almost-rule is if you have set the variable"SAVE_FOR_NEST" to YES, in which case calling eta_convert.csh without arguments will cause the script to completely ignore your post processing requests and simply create grib files for use with the nested run. The grib files will then be move to the $ETA_PREP directory.
Following successful completion of the nested run call eta_convert.csh with the "N" argument to post process the nested simulation:
% eta_convert.csh N
To run and post process a nested simulation:
1) Run coarse mesh simulation. Dump output every hour.
2) Set SAVE_FOR_NEST in eta_convert.csh to YES
3) Run eta_convert.csh
% eta_convert.csh
4) Run eta_download.csh to processes the coarse mesh grib files now located in $ETA_PREP
% eta_download.csh WSE <CYCLE> <INIT> <FINAL> 01 C <YYMMDD>
5) Run the eta_run.csh script to run the nested simulation. Pass the "N" argument.
% eta_run.csh N
6) Run eta_convert.csh
% eta_convert.csh N
7) Enjoy the model output