[R-sig-Geo] How to estimate sig2 attribute for functions linker and kernelbb from the adehabitatHR package

[Breitbach, Nils]

Hi Andy,

in my study we followed the birds for the whole day from sunrise (ca. 5.15 h) until sunset (ca. 21.45 h) and recorded every change in position (if we were fast enough). So I could have a temporal resolution of one minute if want to. Of course we have gaps and the individuals have not been followed every day, but for one burst the data are indeed highly serial autocorrelated. That is why I was thinking about using the BB-Kernels to have a look on the differences to the more traditional UD-Kernels.

I know that GPS-telemetry is making progress but there are still a lot of animals out there that can not carry these heavy loads. Furthermore GPS-telemetry is maybe not always the best method; for example if one wants to additionally record a certain behavioural variables like feeding or mating along with the telemetry data - that is something a satellite cannot do. So I think it is worth to discuss the usage of new and promising methods also for such kind of data. Unfortunately I am not a mathematician and I am busy enough being a Biologist with all its sub-disciplines and meanwhile half a geographer for the necessary GIS knowledge. So I feel not able to step deep enough into the formulas to judge about some of the parameters or the best method for obtaining them. Biologists should be the all-in-one human Swiss army knife ;-)

So I thought we might start a discussion in the forum to talk about using the BB-Kernels for various kinds of data?!
Every kind of input or opinion is welcome ...

Cheers,

Nils


________________________________________
Von: Andrew Duff [andrewaduff at gmail.com]
Gesendet: Donnerstag, 24. Februar 2011 16:00
Bis: Breitbach, Nils
Betreff: RE: [R-sig-Geo] How to estimate sig2 attribute for functions linker and kernelbb from the adehabitatHR package

Hi Nils,

The Brownian bridge movement model is best suited for data with high serial
autocorrelation (locations taken within a short time period).  I am
currently applying it to GPS radiocollar hits from golden eagles recorded
every 1 hour.  I don't think that the method is best suited for VHF
telemetry, you would want to have an equal time period between fixes or you
are going to really balloon the uncertainty in your image with respect to
sig1.  I hope this helps,

Andy

-----Original Message-----
From: Breitbach, Nils [mailto:breitbach at uni-mainz.de]
Sent: Thursday, February 24, 2011 1:30 AM
To: Andrew Duff
Subject: AW: [R-sig-Geo] How to estimate sig2 attribute for functions linker
and kernelbb from the adehabitatHR package

Thank you Andy!

I just realised that I forgot to mention that I used terrestrial radio
telemetry - well it was a late-night posting. The two publications that you
referred to just describing the calculation of sig2 for radio collars. In my
case it might be not so easy or moreover I am not sure if it would make
sense at all. Of course I could place my radio tags in different habitat
types over the whole study area and try to locate them from differnt
positions and from different distances. But is locating a tag not much more
dependent on the experience of the person that is taking the relocations and
less dependent of the accuracy of the tag itself? Or at least I would think
so; especially in a project where many people with different levels of
experience are involved. However, what I do have, are comments from the
field, wheather the observer has seen the birds, heard the birds or just
located the bird by signal strength (indicated by a range fromm "++" to
"--"). We also noted an estimated radius of uncertainty based on personal
evaluation of the quality of each fix (e. g. I am sure, that the bird was
within a radius of 20 m arround the fixed coordinate).

What do you think? Could I use one of my measures or should I track my tags
again from various positions and distances in different types of landscape?
How many trials per landscape type and distance would be sufficient? Or is
sig2 only useful for GPS or ARGOS studies?

Cheers,

Nils


________________________________________
Von: Andrew Duff [andrewaduff at gmail.com]
Gesendet: Donnerstag, 24. Februar 2011 02:55
Bis: Breitbach, Nils
Betreff: RE: [R-sig-Geo] How to estimate sig2 attribute for functions linker
and kernelbb from the adehabitatHR package

Sig2 is the second smoothing parameter for the brownian bridge method which
is related to the imprecision of the relocations, supposed known.  These
values could come from previously published GPS collar error studies or from
some GPS collar testing and error evaluation.

I would also suggest reading
http://www.cnr.uidaho.edu/population_ecology/Horne,%20Garton,%20Krone%20&%20
Lewis%202007%20Ecology%20-%20BrownianBridge.pdf and
http://www.esajournals.org/doi/abs/10.1890/08-2034.1?journalCode=ecap

Are you referring to sig1 instead of sig2?

Andy


-----Original Message-----
From: r-sig-geo-bounces at r-project.org
[mailto:r-sig-geo-bounces at r-project.org] On Behalf Of Breitbach, Nils
Sent: Wednesday, February 23, 2011 3:10 PM
To: R-sig-Geo at r-project.org
Subject: [R-sig-Geo] How to estimate sig2 attribute for functions linker and
kernelbb from the adehabitatHR package

Dear Community,

I just thought that it would be a good idea to have a look on the kernelbb
function from the adehabitatHR package. What really puzzles me is how to
estimate/calculate the sig2 attribute needed for the functions liker() and
kernelbb(). Unfortunately the help for the functions refers to a PhD-thesis
(Maillard, 1996, p. 63) that is written in French and I have made the
mistake to choose Latin in school. Has anyone of you a solution to calculate
a sensible value?

Do I need to calculate sig2 then for each individual and/or burst separately
or just pooled over all individuals and bursts?

Cheers,

Nils
_______________________________________________
R-sig-Geo mailing list
R-sig-Geo at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-geo