running multiple trajectory in time and space

[shipra]

Hello,

I wanted to run Windows version of hysplit for WRF output. I wanted to release multiple trajectory in time and space.

For e.g. There are around 100,000 grid points from which the back trajectories are to be released. Each grid point has different x,y,z values. I want to release six hour back trajectory after every 1 hr from each grid point. How can I achieve this?

i tried using the methodology described in tutorial but it is limited to 3 initial point and also same z values.

Please suggest.

I have access to LINUX system here but I have no idea how to do this in LINUX/UNIX. The module is preinstalled in the system. Is there any tutorial which can guide me to do this in UNIX.

Any help is really appreciated.

Sincerely,
Shipra

[alicec]

You probably want to do this with some sort of scripting.
You can find example scripts in examples/scripts/ folder. There are example batch scripts for windows and shell scripts for linux/unix.
There are also bash and batch scripts for most of the tutorial in the Tutorial folder.
Please see the tutorial section 2.4 Batch file scripting for more information.

You can specify the number of starting locations you want in the GUI (trajectory / setup run) and then put the number in the starting locations.
Then you can enter the latitude, longitude and altitude of the starting points by clicking 'setup starting locations'.

HYSPLIT reads a file called CONTROL and a file called SETUP.CFG in the working directory in which it is run.
The CONTROL and SETUP.CFG file are created by the gui or scripts and then HYSPLIT is called by the gui or scripts.
The CONTROl and SETUP.CFG files are simple text files.
In the control file, the number of starting locations is specified in the second line.
Subsequent lines give the latitude, longitude and altitude of each starting location.

To create multiple trajectories in time follow the tutorial 4.11 multiple trajectories in time.
The variable NSTR must be set in the SETUP.CFG file. (This is done in the GUI by going to Advanced/ configuration setup / trajectory) menu 3.

Although we provide example shell and batch scripts, you can script in whatever language you are most comfortable with.
The main idea is to write a CONTROL and SETUP.CFG file and then call HYSPLIT.

I hope this helps.

When you say that you tried to use the methodology described in the tutorial "but it is limited to 3 points and also the same values" could you specify which section of the tutorial you were following? Are you using the GUI or did you try some scripts?


Cheers,
Alice

[shipra]

Dear Alice,

Thank you so much for your help and reply. I have found the CONTROL, SETUP.CFG files, batch and shell scripts in the sub directories you've mentioned.
Let me re-write what you've explained so that I don't have any confusion.

1. Set the number of starting locations in CONTROL file followed by the latitude, longitude and height (Is there any upper limit for number of locations?)
2. Put the SETUP.CFG in the working directory and edit NSTR variable and run model.

This can be done from GUI also right? Or should I go for shell scripting? Please suggest me which would be better.

I have another query. I was trying to convert the wrfout file by calling the arw2arl executable in /exec directory in UNIX and I am getting an error.
NetCDF: Variable not found
ZNT
Variable not found: ZNT
Some defined variables not found in file: wrfout_d01_2000-01-24_12:00:00
Edit WRFDATA.CFG and try again ...

I found some threads on this forum and realized that this error is due to the fact the ZNT is not written in wrfout file by default.
But, instead of re-running the wrf model I have edited the WRFDATA.CFG and removed the ZNT parameter from the list of 3D parameters. The executable seems to work now. But I don't know if it is a correct thing to do. Please let me know what implications it will have on the trajectory model if I remove the ZNT variable.

I am really thankful to you for your kind help and suggestions.

Sincerely,
Shipra

[alicec]

It is hard to recommend which way is better.

I would recommend going through the entire tutorial on trajectory with the GUI and do the following:

-look at the CONTROL and SETUP.CFG file generated for each example.
-look at the bash and or batch scripts for each example and try to run them.
-look at the output file for each example and think about how you want your output to be structured - how are you going to analyze your data?

The following may be particularly helpful to you:

Section 4.11 Multiple trajectories in time.

Section 4.12 Multiple trajectories in space may be helpful. Look at the output file and note that all the trajectories generated
are in the same ascii output file.

Section 5.2 Cluster analysis. Note that this basically creates a loop which calls HYSPLIT multiple times for each trajectory. Each trajectory gets its own output file.

To generate your 100,000+ trajectories you will probably want to use some combination of the above. Maybe you don't want each trajectory in its own output file, but if you try to get them all into the same output file you may end up with a very large ascii file.
How you want to break them up will depend on how you plan to analyze them.

Cheers,
Alice

[alicec]

Also it is ok not to have the ZNT field for the trajectories. It is not used for the trajectory calculations.
Cheers,
Alice