Clustering Problem: "Error allocation time array: 1836008193"

[AppalAIR]

Hello,

Our research group is trying to cluster a set of back trajectories (821 in total). A few weeks ago we performed a run and it worked just fine, however now there appears to be some error with the time array. The number listed changes with each attempted run. Also, sometimes the error does not occur but HYSPLIT appears unresponsive to Windows 10, prompting a close program dialog box. Here is some system and configuration specific information:

Hours to cluster: 120
Trajectory skip: 1
Time interval (hrs): 1

Note that the trajectory files contain extended meteorology data, and were generated using the SplitR package in R. We have compared these files with others generated using the HYSPLIT interface and they appear to be the same. The file name is different than those generated using HYSPLIT, however we modified the "base name" field to accommodate this change. We are using the 64-bit version of HYSPLIT.

Computer info:
Windows 10 Education, version 1703 build 15063.1029 (64-bit)
Processor: Intel core i5
RAM: 8 GB

Any insights into the reason for this error and how to address it would be greatly appreciated. Thanks!

Sincerely,

The AppalAIR aerosol research group at Appalachian State University

[barbara.stunder]

Are you using the same set of trajectories as from a few weeks ago? If so, what is different about the run now? If you cannot determine the issue, can you reproduce the problem with a small set of trajectories, and attach the trajectories here? Then I will try to reproduce it.

[AppalAIR]

Hi Barbara,

Thanks for helping us look into this problem. The trajectories are the same as the ones we ran previously, as are the run parameters. We were able to reproduce the same issue using only 40 trajectory files.

endpts.zip

(214.1 KiB) Downloaded 565 times

The HYSPLIT version is the April 2018 release (SVN 951), if that information is useful.

Thanks again!

[barbara.stunder]

Thanks for sending the set of trajectories. I was able to reproduce your error, though I'm not using Windows 10. In looking at one of the tdump (trajbwd...) files that you created using the SplitR package, the last two meteorological variables are written to the next line instead of staying on the line with the latitude, longitude, etc. Please edit those files so the all the meteorological variables are on the same line or make the change in SplitR and re-run.

I also found a bug in our subroutine trjinp.f that reads the endpoints file. When all the possible diagnostic variables are used as you did, the formatting was such that the last one was not read correctly. With your version of the code, that means when you edit or re-run the above, use one less diagnostic variable. The updated code (r954) will be in the next HYSPLIT release.