cluster analysis for 5 year long data

Topics about the trajectory clustering program for HYSPLIT.
Post Reply
tepea
Posts: 5
Joined: August 5th, 2014, 7:02 am
Registered HYSPLIT User: No

cluster analysis for 5 year long data

Post by tepea »

Hi. I'm new in HYSPLIT.

I'm trying to cluster my five year long wind data (.gbl). I clustered my data before with TrajStat. But TrajStat is not doing iteration so it's not giving a specific cluster number. I have to use trial-and-error method in TrajStat and I think it's not completely true.

I'm using this tutorial for understand how can i cluster my data with HYSPLIT

http://www.arl.noaa.gov/documents/works ... torial.pdf

but it's just showing for 1 month. How can i use this cluster analysis with 5 year long data. How can i load my all data in an lumb.
barbara.stunder
Posts: 451
Joined: November 9th, 2012, 4:23 pm
Registered HYSPLIT User: Yes

Re: cluster analysis for 5 year long data

Post by barbara.stunder »

All the trajectories in the file INFILE will be clustered. As you say, the example is for one month. You need to create an INFILE with all your trajectories. Due to possible memory issues, start small, say one trajectory a day for year, increase it to two years, etc. For information about memory, see viewtopic.php?f=12&t=516&p=1781&hilit=c ... mory#p1781
yaqiang
Posts: 55
Joined: October 5th, 2015, 10:20 pm
Registered HYSPLIT User: Yes

Re: cluster analysis for 5 year long data

Post by yaqiang »

TSV figure has been added in clustering function of TrajStat Java plugin from 1.4.3 version. So you can get a specified cluster number with the TSV figure using neweast MeteoInfo Java and TrajStat plugin version.
cluster_cal.png
TVS.png
tepea
Posts: 5
Joined: August 5th, 2014, 7:02 am
Registered HYSPLIT User: No

Re: cluster analysis for 5 year long data

Post by tepea »

barbara.stunder wrote:All the trajectories in the file INFILE will be clustered. As you say, the example is for one month. You need to create an INFILE with all your trajectories. Due to possible memory issues, start small, say one trajectory a day for year, increase it to two years, etc. For information about memory, see viewtopic.php?f=12&t=516&p=1781&hilit=c ... mory#p1781
Thank you for your answer. This is a good solution. :)
tepea
Posts: 5
Joined: August 5th, 2014, 7:02 am
Registered HYSPLIT User: No

Re: cluster analysis for 5 year long data

Post by tepea »

yaqiang wrote:TSV figure has been added in clustering function of TrajStat Java plugin from 1.4.3 version. So you can get a specified cluster number with the TSV figure using neweast MeteoInfo Java and TrajStat plugin version.
cluster_cal.png
TVS.png

Thank you. this was easy :D
Post Reply