error in making clusters using gdas meteorological files

General questions and postings pertaining to the use of HYSPLIT regardless of the platform. For platform specific questions, use the HYSPLIT Platform forums.
Post Reply
sonalkumari
Posts: 6
Joined: August 2nd, 2017, 3:34 am
Registered HYSPLIT User: No

error in making clusters using gdas meteorological files

Post by sonalkumari » August 2nd, 2017, 3:46 am

Dear sir
I am using HYSPLIT4 (January 2017 release svn:854) to make clusters for 72 hours for 35 days.
I am using GDAS meteorolgical files as I am doing cluster analysis for two nearby sites. On running the standard cluster analysis it is showing an error-clus-read should not be there, fewer endpoints than needed but the endpoint files are 124. But on doing cluster analysis for same duration using Reanalysis meteorological files cluster is obtain. Can you please help me why this error is coming and how I can deal with it.


With regards
Sonal Kumari

barbara.stunder
Posts: 273
Joined: November 9th, 2012, 4:23 pm
Registered HYSPLIT User: Yes

Re: error in making clusters using gdas meteorological files

Post by barbara.stunder » August 8th, 2017, 11:31 am

Assuming the trajectories ran fine, it does not matter for the cluster program whether the trajectories were created using the GDAS or Reanlaysis meteorology. The only difference in the cluster inputs is the "Endpoints folder" because you would be pointing to a different set of trajectories.

I suggest re-running the Reanalysis clusters so all the other inputs are fine, then change the Endpoints folder to the GDAS folder, and re-run the clusters.

If it fails, look at the trajectory endpoints files to see if there is any substantial difference, such as they terminate before 72 hours or there are multiple trajectory start heights in one file. If they seem fine, please quote or paste an image of the particular error.

sonalkumari
Posts: 6
Joined: August 2nd, 2017, 3:34 am
Registered HYSPLIT User: No

Re: error in making clusters using gdas meteorological files

Post by sonalkumari » August 9th, 2017, 9:31 am

Dear Barbara,
As per your suggestion I have done the reanalysis of clusters using gdas data and have chosen the endpoints from GDAS folder, but after making infile it starts deleting every endpoints and at the end no end point was left in the file.
While if I proceeded without making any folder for GDAS endpoints, and have chosen only the endpoint files generated for this analysis from working directory it showed a particular error which is attached.
Attachments
sonalhysplit.pdf
(24.56 KiB) Downloaded 77 times

sonalkumari
Posts: 6
Joined: August 2nd, 2017, 3:34 am
Registered HYSPLIT User: No

Re: error in making clusters using gdas meteorological files

Post by sonalkumari » August 9th, 2017, 10:06 am

Dear Barbara,
On seeing the gdas endpoint files it was found that some of the files have terminated before 72 hours while the endpoint file for the same day ran for 72 hr when the analysis was done by reanalysis data.

sonalkumari
Posts: 6
Joined: August 2nd, 2017, 3:34 am
Registered HYSPLIT User: No

Re: error in making clusters using gdas meteorological files

Post by sonalkumari » August 9th, 2017, 10:32 am

Dear Barbara,
When I compared the endpoint files generated by GDAS and reanalysis data I found that some of the endpoint files in GDAS ended before 72 hr while the same file in reanalysis ran for 72 hr.

barbara.stunder
Posts: 273
Joined: November 9th, 2012, 4:23 pm
Registered HYSPLIT User: Yes

Re: error in making clusters using gdas meteorological files

Post by barbara.stunder » August 17th, 2017, 9:30 am

Look at your INFILE and make sure it has the endpoint files you are clustering.

Look at the CMESSAGE file for where it lists the files being read and lists files that will not be clustered. In particular:
"WARNING: traj not long enough; traj not clustered <filename>"
See if you can tell which file is causing the error. If you can remove it from the INFILE and successfully run the clusters, that may be a work-around.

If you say you want to cluster for 72 hours, then the trajectories that are less than 72 hours long will be skipped. These will be mentioned in the CMESSAGE file and listed in the file TNOCLUS. The file TCLUS is a list of the files that will be clustered.

From the error you got, it seems it is not correctly skipping a "short" trajectory file. Can you zip your endpoints files or a smaller subset that still gets the same error and put them here or email to me at barbara.stunder@noaa.gov so I can try to reproduce your error.

Zainab
Posts: 4
Joined: October 15th, 2018, 2:08 am
Registered HYSPLIT User: No

Re: error in making clusters using gdas meteorological files

Post by Zainab » November 24th, 2018, 12:17 pm

Hello Barbara,

I have tried using reanalysis data for 2017 and 2018. I am getting the same error message:clusmean should not be here, this trajectory has fewer endpoints. I read the cmessage, found the files causing the error and deleted the same from Infile though they were numerous. Everytime I ran the clusters, there were more trajectories appearing in the cmessage . I kept on deleting from the INFILE. At last, there were none in the cmessage , but the error message was retained.
I tried clustering back trajectories for 72 hour back trajectories for Jan-Apr 2017 and Feb-April 2018. Please check if there is some problem with the data. I also checked with gdas1 data for 15th Feb- 31st March, 2017 and got the same error message. Please resolve these issue.

Regards,
Zainab.

barbara.stunder
Posts: 273
Joined: November 9th, 2012, 4:23 pm
Registered HYSPLIT User: Yes

Re: error in making clusters using gdas meteorological files

Post by barbara.stunder » November 26th, 2018, 11:41 am

I ran 72-h back trajectories beginning 00z from January 4 - April 30, 2017 and February 4 - April 30, 2018 using the Reanalysis data. There is no problem with the Reanalysis data. The clusters for both of these ran fine. There are no trajectories before the 4th of the month because 72-h back trajectories would have 'gone into' the previous month, which weren't in the trajectory CONTROL file. When "Hours to cluster" is set to 72, only trajectories at least 72-h long are used. The trajectory endpoints are output every hour, which is the default. In the Cluster GUI, I used the default "1" as "Time interval" and "1" as "Trajectory skip". Please run these sets of trajectories beginning at 00z and see if it works.

Zainab
Posts: 4
Joined: October 15th, 2018, 2:08 am
Registered HYSPLIT User: No

Re: error in making clusters using gdas meteorological files

Post by Zainab » December 5th, 2018, 4:08 am

Thankyou Barbara. It ran fine.

Post Reply