Make a collection of samples in CDendro

Last update: Sept 9 2013.

From the two previous lessons we know that NMBS01.POS and NMBS05.POS are two samples which match well when their first years (0) lay over each other.

We will now add these two files together into a collection and then make a "mean-value-sample" out of it.

Then we will match yet another of the files from Bergvik towards that "mean-value-sample", and then add it to the collection.

1. Open NMBS05.pos and select it as the reference

2. Open NMBS01.pos and run a correlation analysis to make sure that the files match together.

Note: You may select several sample files at the same time: Hold down the control key and click on those files you want to open. Then click open.

Click on "Collections" on the menu bar and then click "Create new empty collection". Then give it the temporary name "NewDemo" and click OK.

The new collection will automatically be selected as the "target collection", i.e. when samples are added to a collection they are added to this "target" collection. Now you have an empty basket to collect your samples!

Click on the top of NMBS01.pos to make it the uppermost (not hidden) window.

You can also select it from "Window" in the menu bar as shown in the picture above. If you prefer to work with maximized windows, this is one alternative for switching between windows! The other alternative is to use the WindowsList which lists the windows within a separate frame, as shown above. (For more info see the section "The WindowsList" under the heading "Basics".)

Click on the button "Add to target collection"

A new window for setting an offset will pop up - just click OK to accept zero as the offset.

To see what actually happened, click on the top of the collection window, so you can see your sample in that basket!

Now select the other sample NMBS05.pos and then add also that sample to the collection so you get two samples in the collection! Also this one should have zero offset as we know that the samples are of the same age and they fit together at their beginnings.

Close both windows NMBS05 and NMBS01: First select a window by clicking on the top of it, then click the close button at the top right corner.

Note: You might instead use the option "Close that window" on the WindowsList to close any window in the list.

Now we will calculate a mean value of these two samples. - Click on the button "Create Mean value sample"

A new "sample" has been created. It consists of mean values of ring width data and mean values of normalized data created from our two samples from Bergvik. - We might name this a very simple reference curve!

The diagram shows that "reference curve" as a group of red curves plotted from mean values of the normalized data of the samples included in the collection - and as a group of green curves plotted from the mean values of the ring width data of the samples.

The red and green curves in the middle of the two tripple-curve-groups are the real mean value curves.
The outer (enclosing) curves are based on standard deviation values - not of much interest when mean values are based on only two curves, but of much more interest when there are many samples in a collection.

The red "1" and "2" digits along the top of the curve diagram, represent the number of samples (trees) involved in the respective time segments of the curve. These numbers are sometimes called "the sample depth" or "the replication".

Click on the Workbench tab!

The report shows how many ring width values there are behind the mean value for each year.
There are two years for the first 84 years. Between 85 and 102 there is only one value for each year because there were more ring coordinates in NMBS05.pos than in NMBS01.pos
These numbers represent the same information as the red digits printed along the top of the curve diagram, as described above.

You should now add a third sample to the collection, and then calculate new mean values.

Click on the button "Select as reference", so you can match the next sample towards this mean-value-sample.

Then open the next sample file NMBS08.pos!

Click the button "Make corr. analysis" and inspect the curve matching!

The red numbers (like "0.75") along the bottom of the diagram represent correlation coefficient values for 20 years long segments (blocks) lagged every 10th year. That block length and block distance can be set with the "Blocklength" and "Blockdistance" frames in the lower left corner of the picture above.

The grey bar above these numbers represents "the trend" along the curve. When both the ring width curves of nmbs08 and the reference turn the same way (up or down) then the trend is the same and the gray bar plotted gray. The trend bar can be turned on or off through the "Plotting window settings".

Click the Workbench tab!
Note that this sample matches best at relative year 3 in the reference, created out of the collection.

Click on the button "Add to target collection"!

You are prompted for the offset to use when adding this third sample to the collection. See there is a 3 in the offset field! Then click OK and close the NMBS08.pos window.
In the Offset column you have the relative years covered by data from the corresponding member file.

If you want to delete a member of this collection, select it by clicking on it and then hit the Delete selected members button! (If you want to delete several members you can select them as normal in Windows (e.g. ctrl-Click to select) and then delete them by "Delete selected members".)

You can attach a comment to a member: Select it by right-clicking on it to open an edit window! Here you can also change the offset value. - You can even change the identity of a member!

Save the collection file. Use the name NMBS1_5_8.fil (Click on "Collections" on the menu bar, then "Save current collection..." and "as .fil"). Then close it.

Close also any other open file.

Note: A collection can be saved either as a collection file (extension .fil) or as a decadal file (extension .rwl (Tucson) or .fh (Heidelberg formatted)).
The collection file (.fil) contains only filenames, comments and offsets - NO ring width data!
When you save a collection in a .rwl decadal file, you get a file with all ring width data written in the same format as files in the ITRDB. (International tree ring data bank in Arizona). Though the files published in the ITRDB do normally not contain comments. You have two commands in the Collection-menu for writing in .rwl Tucson decadal format: "... as Tucson .rwl file WITH comments" or "... as Tucson .rwl file WITHOUT comments but with .txt comment file"
More on the decadal format in a subsequent lesson.

Note: When you save a "mean-value-sample" created with the button "Create mean value sample" in a collection you save it either as normalized mean value data (.d12 format) with standard deviations and ring widths with standard deviations or as only ring width mean values in one of three possible formats - .wid, .rwl or .fh

There are four different ways to get a "mean-value-curve" with standard deviations on the screen:
1. Open a normalized (.d12) file with a mean value curve created as mentioned above.
2. Open a collection "Open collection (.fil)" with all its members and click on the button "Create mean value sample"
3. Use the command Collections/Create new collection from decadal file (.rwl), click the "Create mean value sample".
4. Use the command Collections/Create reference curve from big decadal file (.rwl or .fh).

Note: In early versions of CDendro ".dec" was used as the file extension for decadal files. When the ".rwl" extension was introduced in CDendro, the old .dec-extension was kept to make it possible to open old files with the .dec extension.



See also the section on Multi-radii collections, i.e. when several radii are taken from the same trunk.
Lars-Åke Larsson. Last update: Sept 9 2013.


Copyright © 2013, Cybis Elektronik & Data AB, www.cybis.se