Multi-radii cross correlation

You may post any question related to CDendro in this meeting.
Post Reply
Colin Bergeron
Posts: 6
Joined: 26 Sep 2009, 11:08
Real name: ColinB
Contact:

Multi-radii cross correlation

Post by Colin Bergeron »

I am building a chronology from living trees based on a big number of trees (>100 per species). I do have 4 radii from each of them. So when i do a cross correlation to see which are best match, radii from the same trees always match up even if the sum by stem button is checked. It would be lot of work to create a collection for each trees and then add them all in the same collection to work with independent trees. So, it is a bit anoying for the decision of which tree I should use to start the collection. In that case, I use test toward the rest of collection but it is still hard to trust that because then I do test against the radii with missing or double rings. Maybe all these problems are because me being newbie... Any advices?
[b][i]Colin[/i][/b]
Lars-Ake
Site Admin
Posts: 87
Joined: 21 Dec 2008, 23:24
Real name: Lars-Ã…ke Larsson
Location: Saltsjobaden (Stockholm), Sweden
Contact:

Re: Multi-radii cross correlation

Post by Lars-Ake »

Colin Bergeron wrote:I am building a chronology from living trees based on a big number of trees (>100 per species). I do have 4 radii from each of them. So when i do a cross correlation to see which are best match, radii from the same trees always match up even if the sum by stem button is checked. It would be lot of work to create a collection for each trees and then add them all in the same collection to work with independent trees. So, it is a bit anoying for the decision of which tree I should use to start the collection. In that case, I use test toward the rest of collection but it is still hard to trust that because then I do test against the radii with missing or double rings. Maybe all these problems are because me being newbie... Any advices?
I have considered this problem for a while now... you are absolutely right, this is a problem.
And it is related also to other commands, e.g. "Add best members to target collection".
But since nobody has complained about this before, I have not considered it a serious matter.... :)

A proper solution to this needs quite a lot of programming and testing, so I cannot promise a quick solution.

The way I have used is to uncheck all but one member of each stem and then take the decisions from that.
Easiest way to do that is to uncheck all members (a button command) and then check one member of each stem.
Quite clumsy... - and it does not solve the problem when several .pos-files make up one radius.

EDIT BY Lars-Åke: -------
To have only the first member of any stem checked, do like this:
See that "Sum by stem" is checked.
Use the button "Check all". Then the menu command Collections/Uncheck handling.../Uncheck members marked as SameStem(x)
Note: Currently this only works when the members are sorted by the first column, i.e. by identity. I will fix this.
END EDIT-----------------

To actually use all your radii, you have to create a mean value out of each stem - as you propose!

When I work with such a problem, I actually put all my .pos-files of one stem into a new empty collection -
use "Add to this collection".
Then I crossdate these radii, i.e. I adjust the offsets if necessary and check that no rings have been missed or duplicated.
Often I use the Plot all members button to see all the radii curves over each other.
When the crossdating is OK I save the collection as a .fil file, i.e. with references to my .pos files.
That way it is easy to update any measurement, save the .pos-file, and already have it referenced from the .fil-collection file.

Now I can also easily create a mean value file for this stem and save it as a .wid file.

If this is done successively during the measurement work it is not much extra work to create those mean value files -
one for each stem-collection.

Finally I can create a new empty collection and add in all the mean-value .wid files in one single operation.

This is a solution that you can use today. But indeed, what you ask for would be much better...!
Colin Bergeron
Posts: 6
Joined: 26 Sep 2009, 11:08
Real name: ColinB
Contact:

Re: Multi-radii cross correlation

Post by Colin Bergeron »

Thanks for the advices. It ends up being lot of work but I guess it is worth for a thesis...
[b][i]Colin[/i][/b]
Post Reply

Who is online

Users browsing this forum: No registered users and 24 guests