Hi,
In a CNRS-dependent site there is a interesting database of French chronologies: http://chrono-eco.univ-fcomte.fr/Public ... Tmp/DData/
I hope this is not a redundant information. I have not spent much time on it but from a first look I think all data are stored in three relevant files, .txt format. Hopefully they can be read somehow. They seem to be a list of a sort of decadal format chronologies, but some detail escapes to my interpretation (authors say that it’s a format “convenient” to CNRS researchers). Zero rings are probably indicated by commas, and raw ring values have decimal separator points. I made just one attempt to import one chronology-table to .fh format (by using Cdendro facility) but I had no success.
File http://chrono-eco.univ-fcomte.fr/Public ... CE-Nat.txt (second file in the list): Oak chronologies from 449 BC to 193 AD, ring width values. Maybe someone of you interested in early oak chronologies (although from a bit distant places) could appreciate it. Source: “laboratoire de Chrono-Ecologie de Besançon”. Data are part of a thesis by a researcher named S. Durost, which covers from 546 BC to 193 AD (so not all data are included in the online file).
File http://chrono-eco.univ-fcomte.fr/Public ... CE-AjC.txt (first file in the list): same chronologies, but values corrected by a method (unknown to me) described in Lambert G., Durost S., Cuaz J. 2005 (see extended quote in the same file)
File http://chrono-eco.univ-fcomte.fr/Public ... 6-EBes.txt (third document in the list) stores chronologies from the end of early middle age (I see one starting from IX century) to contemporary times (ring width values?), but I cannot find out the species (various?). Source: several French laboratories.
Others three documents contain information about laboratories that collected data, environmental/context information and other information stuff.
The database was made public by Vincent Bernard, Virginie Chevrier, Claire Doucerain, Olivier Girardclos, Frédéric Guibal, Georges Lambert, Catherine Lavier, Christine Locatelli, Christophe Perrault, Patricia Perrier.
I’m not particularly interested by now for looking at those data, but just for curiosity, if someone of you can “translate” them, or correct my overview, or has already worked with them, please let it know. They seems to be a precious resource.
Ale
CNRS Database
-
- Site Admin
- Posts: 87
- Joined: 21 Dec 2008, 23:24
- Real name: Lars-Ã…ke Larsson
- Location: Saltsjobaden (Stockholm), Sweden
- Contact:
Re: CNRS Database
Thanks Ale!
This is indeed interesting data. I think that it is all oak of some type.
The format is quite new to me and I have to write some sort of input filter for CDendro. Though I would be happy if I knew more about the format!
The data with negative values are probably based on the "Besancon corridor method" (for a reference see the topic at this forum "Questions raised by the Hollstein data") where the final data looks very much as pure ring widhts, though it is offset so that the lower values are negative. The corridor method is yet another type of normalization. The property that the resulting curve looks like a ring width curve, probably means that a low frequency component of the growth is kept in the data used for crossdating.
I am working on an implementation of this within CDendro, though I have to clean up my code a lot because it has been clogged up with too many ad hoc solutions when I have successively added all those other normalization methods.
/Lars-Ake
This is indeed interesting data. I think that it is all oak of some type.
The format is quite new to me and I have to write some sort of input filter for CDendro. Though I would be happy if I knew more about the format!
The data with negative values are probably based on the "Besancon corridor method" (for a reference see the topic at this forum "Questions raised by the Hollstein data") where the final data looks very much as pure ring widhts, though it is offset so that the lower values are negative. The corridor method is yet another type of normalization. The property that the resulting curve looks like a ring width curve, probably means that a low frequency component of the growth is kept in the data used for crossdating.
I am working on an implementation of this within CDendro, though I have to clean up my code a lot because it has been clogged up with too many ad hoc solutions when I have successively added all those other normalization methods.
/Lars-Ake
Re: CNRS Database
Hi Lars-Ake,
Go here: http://www.dendro.bf.uni-lj.si/lab_info.html and search for “F-LCE [Laboratory]” You will find some information (unfortunately not so many, and some of which sound cryptic to me) about data formats and software they use, a list of papers of them, and the contact mail.
Apparently they use Tucson and Tsap format, and their personal format, “Besancon”.
No bibliography listed there seems to be directly related to detailing their standards, but there have to be more information somewhere else. Probably the best and fastest approach would be to write them and ask for concrete information and/or relevant bibliography.
Speaking of bibliography, here http://www.persee.fr/web/revues/home/pr ... m_3_3_1339 , on page 289, they quote a “Besançon filter” and his formula. I don’t know if it’s related to the Besançon normalization you already used in Cdendro. I have not read the whole paper, that looks as a general overview of dendrochronology methodology, insisting on GLK test.
Ale
Go here: http://www.dendro.bf.uni-lj.si/lab_info.html and search for “F-LCE [Laboratory]” You will find some information (unfortunately not so many, and some of which sound cryptic to me) about data formats and software they use, a list of papers of them, and the contact mail.
Apparently they use Tucson and Tsap format, and their personal format, “Besancon”.
No bibliography listed there seems to be directly related to detailing their standards, but there have to be more information somewhere else. Probably the best and fastest approach would be to write them and ask for concrete information and/or relevant bibliography.
Speaking of bibliography, here http://www.persee.fr/web/revues/home/pr ... m_3_3_1339 , on page 289, they quote a “Besançon filter” and his formula. I don’t know if it’s related to the Besançon normalization you already used in Cdendro. I have not read the whole paper, that looks as a general overview of dendrochronology methodology, insisting on GLK test.
Ale
-
- Site Admin
- Posts: 87
- Joined: 21 Dec 2008, 23:24
- Real name: Lars-Ã…ke Larsson
- Location: Saltsjobaden (Stockholm), Sweden
- Contact:
Re: CNRS Database
Hi Ale and everybody!
I've now looked into the Besancon files again. I think I can program a read-in mechanism for these files without getting further information.
Of course it would be fine to know more, though many examples are usually the best way to go.
About Becanson normalization:
The filter of your reference on page 289 is unknown to me. Though you should easily be able to set it up within CDendro.
I do not think it is better than the other filters we use. Torbjörn and I have experimented a lot with variants of these "sliding frame" filters.
The best one seems to be the "Proportion of last two years growth" which is a very very LITTLE bit better than the Hollstein filter.
The Besancon Index E algorithm is also a "sliding frame" filter available within the TEST version of CDendro. Though it is not much better than the others!
It is documented at the end of the text within the large text box at "Options for normalization and matching" in CDendro.
Then comes the Besancon corridor method - which tries to remove inconsistensies within a ring width curve and to detrend it as well as possible.
It is using a polynomial for detrending. I am working with the implementation with this now, and yes, I have a contact with Besancon on these matters.
For this work I had to rewrite a lot of CDendro - so now I am trying to remove all errors introduced...
and to "fine tune" the corridor method. There should be an option for the operator to control the order of the polynomials used.
It might be that the corridor method creates data which is useful for climatic analysis - though I am indeed not a specialist on that!
Regards/Lars-Ake
I've now looked into the Besancon files again. I think I can program a read-in mechanism for these files without getting further information.
Of course it would be fine to know more, though many examples are usually the best way to go.
About Becanson normalization:
The filter of your reference on page 289 is unknown to me. Though you should easily be able to set it up within CDendro.
I do not think it is better than the other filters we use. Torbjörn and I have experimented a lot with variants of these "sliding frame" filters.
The best one seems to be the "Proportion of last two years growth" which is a very very LITTLE bit better than the Hollstein filter.
The Besancon Index E algorithm is also a "sliding frame" filter available within the TEST version of CDendro. Though it is not much better than the others!
It is documented at the end of the text within the large text box at "Options for normalization and matching" in CDendro.
Then comes the Besancon corridor method - which tries to remove inconsistensies within a ring width curve and to detrend it as well as possible.
It is using a polynomial for detrending. I am working with the implementation with this now, and yes, I have a contact with Besancon on these matters.
For this work I had to rewrite a lot of CDendro - so now I am trying to remove all errors introduced...
![Very Happy :D](./images/smilies/icon_e_biggrin.gif)
![Very Happy :D](./images/smilies/icon_e_biggrin.gif)
![Very Happy :D](./images/smilies/icon_e_biggrin.gif)
and to "fine tune" the corridor method. There should be an option for the operator to control the order of the polynomials used.
It might be that the corridor method creates data which is useful for climatic analysis - though I am indeed not a specialist on that!
Regards/Lars-Ake
Re: CNRS Database
Hi to all,
A couple of questions about the CNRS database files. Now that Lars Ake has implemented in CDendro (test version) a reading mechanism for this data, I can create without problem a collection from the roman time files.
But if I try to create a collection from the medieval-contemporary chronology file (Hist20050126-EBes.txt), it seems to me as if it doesn’t work properly. The collection is created, with all samples apparently at the right place, but when I want to visualize the mean curve (or plot all curves), there isn’t any ring width curve, just a normalized curve that appears as an almost straight line, with a few strange peaks. In a “test toward rest of collection”, almost all normalization algorithms show a 0 value, while only the “*P2Yrs” shows some (very low) values.
The txt file has a lot of negative values. Are those values the result of some filter used by the lab? Are they the cause of my (supposed) problem? How can I deal with this file?
An also: what does it means the option “Enable Besancon txt file lookup” in the “setting” menu? I selected it but I noticed nothing.
Thanks in advance.
Regards,
Ale
A couple of questions about the CNRS database files. Now that Lars Ake has implemented in CDendro (test version) a reading mechanism for this data, I can create without problem a collection from the roman time files.
But if I try to create a collection from the medieval-contemporary chronology file (Hist20050126-EBes.txt), it seems to me as if it doesn’t work properly. The collection is created, with all samples apparently at the right place, but when I want to visualize the mean curve (or plot all curves), there isn’t any ring width curve, just a normalized curve that appears as an almost straight line, with a few strange peaks. In a “test toward rest of collection”, almost all normalization algorithms show a 0 value, while only the “*P2Yrs” shows some (very low) values.
The txt file has a lot of negative values. Are those values the result of some filter used by the lab? Are they the cause of my (supposed) problem? How can I deal with this file?
An also: what does it means the option “Enable Besancon txt file lookup” in the “setting” menu? I selected it but I noticed nothing.
Thanks in advance.
Regards,
Ale
-
- Site Admin
- Posts: 87
- Joined: 21 Dec 2008, 23:24
- Real name: Lars-Ã…ke Larsson
- Location: Saltsjobaden (Stockholm), Sweden
- Contact:
Re: CNRS Database
Please note that this reading command is not able to handle "corridor normalized Besancon data" with a mean value of zero and both positive and negative "ring width values"!ale wrote:... Now that Lars Ake has implemented in CDendro (test version) a reading mechanism for [Besancon data], I can create without problem a collection from the roman time files.
But if I try to create a collection from the medieval-contemporary chronology file (Hist20050126-EBes.txt), it seems to me as if it doesn’t work properly. The collection is created, with all samples apparently at the right place, but when I want to visualize the mean curve (or plot all curves), there isn’t any ring width curve, just a normalized curve that appears as an almost straight line, with a few strange peaks. In a “test toward rest of collection”, almost all normalization algorithms show a 0 value, while only the “*P2Yrs” shows some (very low) values.
The txt file has a lot of negative values. Are those values the result of some filter used by the lab? Are they the cause of my (supposed) problem? How can I deal with this file?
... what does it means the option “Enable Besancon txt file lookup” in the “setting” menu? I selected it but I noticed nothing.
Ale
I am not really happy with the current implementation of "read besancon data" in CDendro.
When running into "corridor normalized Besancon data" it would be best with a warning and
some sort of transformation where that data is moved up to positive data but done in such a way
that the smallest value is not too small, see http://www.cybis.se/forfun/dendro/foolb ... /index.htm
To make the .txt files show up in the Open files dialogue box, you have to Enable Besancon .txt file lookup in the Settings menu.
The reason to have this settings, is to give you a chance to avoid having all .txt files on your computer listed as being Besancon .txt files
when you do not have any such file on your machine.
Another problem with the current implementation is the handling of member identities. There should best be
an option while reading to convert into a Heidelberg file format where member identites longer than 7 characters are allowed.
By the way, thanks for writing your questions on this forum! /Lars-Åke
Re: CNRS Database
Hi again,
Just for curiosity... there is online a paper by the people of Besançon laboratory:
http://halshs.archives-ouvertes.fr/docs ... CE2005.pdf
It’s principally focused on dendroclimatology-related aspects, but pages 244-249 are about the corridor method (already explained by Lars Ake when he implemented the reading mechanism for Besançon files: http://www.cybis.se/forfun/dendro/corridor/) and the construction of the two French chronologies, the protohistoric/roman one and the historic one.
Best regards,
Ale
Just for curiosity... there is online a paper by the people of Besançon laboratory:
http://halshs.archives-ouvertes.fr/docs ... CE2005.pdf
It’s principally focused on dendroclimatology-related aspects, but pages 244-249 are about the corridor method (already explained by Lars Ake when he implemented the reading mechanism for Besançon files: http://www.cybis.se/forfun/dendro/corridor/) and the construction of the two French chronologies, the protohistoric/roman one and the historic one.
Best regards,
Ale
Who is online
Users browsing this forum: No registered users and 64 guests