[AniMov] kernelUD error

Mathieu Basille basille at ase-research.org
Sun Mar 6 18:49:56 CET 2011


What about adding a little noise to the relocations of the motionless 
animals? Have a look at ?jitter.

Mathieu.


Le 06/03/2011 02:54, Hazel, Julia a écrit :
> Dear Clément and Paolo,
>
> Thank you very much for quick and helpful replies, giving me plenty to think about and work on.
>
> At this stage UDs serve as a helpful tool for exploring my data. In this I can simply assign NA for rare instances of 'animal that did not move' during an arbitrary short period.
>
> In reality the animal was not stationary but during some periods it moved less than my tracking system could record. From a biological perspective it would be nice to assign a 'pseudo-kernel' or maybe just guesstimate a small radius... cannot foresee a rigorous way to estimate this. Also recognise that my 'locations' derived from a passive acoustic array will not support strict interpretation of UD probabilities... but getting beyond the scope of this email!
>
> Many thanks again,
> Julia
>
> ________________________________________
> From: Clément Calenge [clement.calenge at oncfs.gouv.fr]
> Sent: Saturday, 5 March 2011 20:38
> To: Animal Movement
> Cc: Hazel, Julia
> Subject: Re: [AniMov] kernelUD error
>
> Dear Julia,
>
>> Many thanks to those who create and maintain AniMov software!
>> After a steep learing curve with adehabitat I'm finding it very useful.
>> However with kernels for short periods, I sometimes get an error.
>> It seems this is caused by a set of locations that don't change.
>>
>> ### adehabitatHR_0.1  R 2.12.1 Windows XP
>> #demo: error with animal at same location during period of interest
>> X<- rep(699889, 10)
>> Y<- rep(3161559, 10)
>> Name<- rep("Brock", 10)
>> test<- data.frame (Name, X, Y)
>> coordinates(test)<- c("X", "Y")
>> ud<- kernelUD(test)
>> #Error in points2grid(points, tolerance, round, fuzz.tol) :
>> #  cannot determine cell size from constant coordinate 1
>
> Yes, this error is expected: the UD is a continuous probability density
> function estimated for each pixel of a grid. The function uses the X and
> Y range of the relocations to determine the extent of the grid. When
> this range is equal to zero, the function cannot identify this range.
>   From a *purely technical* point of view, this can be fixed by first
> building a grid with the desired extent, and then use it as the
> parameter grid in the function kernelUD. However, you will have another
> problem when the function will have to calculate the smoothing
> parameter: whatever the method you choose, it depends on the variability
> of the X and Y coordinates. So that the only *technical* way to have a
> result from kernelUD in such a case is also to specify the smoothing
> parameter "manually". For example:
>
> ## Prepares the data
> X<- rep(699889, 10)
> Y<- rep(3161559, 10)
> Name<- rep("Brock", 10)
> test<- data.frame (Name, X, Y)
> coordinates(test)<- c("X", "Y")
>
> ## Prepares the grid
> fakeda<- data.frame(X = c(699889-200, 699889+200), Y=c(3161559-200,
> 3161559+200))
> coordinates(fakeda)<- c("X","Y")
> grid<- ascgen(fakeda, nrcol=100)
>
> ## estimate the UD with h=100
> kud<- kernelUD(test, h=100, grid=grid)
>
>
> Now, you have to think about the scientific issue raised by this
> problem: for example, the 95% home range is defined as the minimum area
> where the probability to find the animal is equal to 95%. In your case,
> the home range is not an area, but a point. What is the meaning of the
> home range in this context? I do not think that any statistical method
> will succeed to estimate the home-range of an animal that does not move.
> HTH,
>
> Clément
>
>
>
>>
>> #demo: no error after small change in one location
>> XX<- c(rep(699889, 9), 699890)
>> YY<- c(rep(3161559, 9), 3161560)
>> Name<- rep("Brock", 10)
>> test2<- data.frame (Name, XX, YY)
>> coordinates(test2)<- c("XX", "YY")
>> ud2<- kernelUD(test2)
>>
>> Apologies in advance if this problem arises from newbie misunderstanding. However unchanging locations (over a limited time) are a valid occurrence in my data so I will be very grateful for the solution.
>>
>> Thanks and best wishes,
>>
>> Julia
>>
>> Julia Hazel Ph.D.
>> Adjunct Research Fellow
>> School of Earth and Environmental Sciences
>> James Cook University, Townsville
>> QLD 4811
>> Phone: 0407 431 382 (international +61 407 431 382)
>> Email: Julia.Hazel at jcu.edu.au
>> _______________________________________________
>> AniMov mailing list
>> AniMov at faunalia.it
>> http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov
>>
>
>
> --
> Clément CALENGE
> Cellule d'appui à l'analyse de données
> Direction des Etudes et de la Recherche
> Office national de la chasse et de la faune sauvage
> Saint Benoist - 78610 Auffargis
> tel. (33) 01.30.46.54.14
>
>
> _______________________________________________
> AniMov mailing list
> AniMov at faunalia.it
> http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov

-- 

~$ whoami
Mathieu Basille, Post-Doc

~$ locate
Laboratoire d'Écologie Comportementale et de Conservation de la Faune
+ Centre d'Étude de la Forêt
Département de Biologie
Université Laval, Québec

~$ info
http://ase-research.org/basille

~$ fortune
``If you can't win by reason, go for volume.''
Calvin, by Bill Watterson.


More information about the AniMov mailing list