Page 1 of 1
Cluster Problem
Posted: March 4th, 2019, 10:39 am
by roscag
I’m trying to cluster back trajectories (120-h, 500 m agl, Traj skip=1) in Hysplit4 but the program assigns more than 90 % of the calculated BT to cluster #0 (WARNING:” traj not long enough; traj not clustered” in the CMESSAGE file). I have tried running the cluster analysis several times and I have tried using a few different sets of trajectory files
Thanks for the attention, hope someone helps me
Re: Cluster Problem
Posted: March 4th, 2019, 12:17 pm
by barbara.stunder
Please check your trajectories to verify they are at least 120-h long.
Does it work with that same set of trajectories for a shorter duration such as 24-h?
Re: Cluster Problem
Posted: March 5th, 2019, 5:13 am
by roscag
yes, trajectories are 120-h long.
It doesn't work also with shorter trajectories
Re: Cluster Problem
Posted: March 5th, 2019, 8:34 am
by barbara.stunder
Please send a subset of the trajectories with a case that fails so I can reproduce the issue.
Re: Cluster Problem
Posted: March 7th, 2019, 9:34 am
by roscag
Attached a subset of the trajectories used
thanks
Re: Cluster Problem
Posted: March 7th, 2019, 10:00 am
by barbara.stunder
I don't see the attachment. Please try again.
Re: Cluster Problem
Posted: March 8th, 2019, 5:32 am
by roscag
Please, see a new attached
thanks
Re: Cluster Problem
Posted: March 8th, 2019, 8:13 am
by barbara.stunder
I just opened several files and none of them are for 120-hours. For example
.fdump14060612 - 8 hours
.fdump14060412 - 9 hours
.fdump14050100 - 0 hours
.fdump14060218 - 3 hours
Hence if you are clustering for 120 hours, none of these I looked at would be used.