Hi Elodie,
glad to see that you move to explore also the BB approach.
However, I am sorry but I don't see the point of comparing the "traditional"
kernel estimation with the BB approach, because the BB approach lies on a quite
different assumption (time spent in an area) so comparison with "traditional"
kernel could make no sense, although as stated in some papers, usually the
Kernel estimation and the BB estimation are quite comparable.
I would guess that the problem here is with the different methods used to
calculate the overlap, e.g. BA and UDOI. Might be useful to carefully go
further into the details of such methods as they are quite different in their
outputs.
Additionally, according to what I can figure out from the behaviour of your
species (penguin), I personally think that the BB approach should be more
"realistic" than the kernel as "foraging" movements from one patch of food
(schools of fish) to another should be probably better "described" by the
former approach.
From the code you posted, it seems like everything is fine.
One additional tip, what happened if you plot the two approach (traditional
kernel vs BB)?
A visual inspection of what the models do can drive you into the topic better
than numbers.
All the best
Stef
Il Lunedì 18 Agosto 2014 12:00, "animov-requ...@faunalia.it"
<animov-requ...@faunalia.it> ha scritto:
Send AniMov mailing list submissions to
animov@faunalia.it
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov
or, via email, send a message with subject or body 'help' to
animov-requ...@faunalia.it
You can reach the person managing the list at
animov-ow...@faunalia.it
When replying, please edit your Subject line so it is more specific
than "Re: Contents of AniMov digest..."
Today's Topics:
1. interpretation of kerneloverlap and kerneloverlaphr
(Elodie Camprasse)
----------------------------------------------------------------------
Message: 1
Date: Mon, 18 Aug 2014 14:34:07 +1000
From: "Elodie Camprasse" <elodie.campra...@gmail.com>
To: <animov@faunalia.it>
Subject: [AniMov] interpretation of kerneloverlap and kerneloverlaphr
Message-ID: <003301cfba9d$a8463940$f8d2abc0$@gmail.com>
Content-Type: text/plain; charset="us-ascii"
G'day everyone,
I am working on tracking data from penguins and would like to determine the
consistency of space use from each of my individuals from one trip to the
next. I calculated kerneloverlap between consecutive trips (standard kernel)
and wanted to compare the results of the kerneloverlaphr after using the
Brownian bridge model. I am not understanding the results I am getting,
however, and I was hoping someone could help me understand them. I pretty
much get opposite results (in the case of kerneloverlap, the individuals are
consistent from one trip to the next, and in the case of kerneloverlaphr,
they are not).
I have created a reproducible example (csv attached) where an individual has
GPS coordinates for two consecutive trips.
I used the following code:
test2_kernel<-read.csv("reproducible_example_kerneloverlap2.csv", header=T)
# Try the kerneloverlap approach
test2_kernel_df<-SpatialPointsDataFrame(coords=test2_kernel[, c(2,3)],
data=test2_kernel[, c(1,4)], proj4string = CRS("+proj=utm +zone=55 +south
+ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs"), bbox = NULL)
test2_kernel_BA<-kerneloverlap(test2_kernel_df[,1], grid=200, meth="BA")
test2_kernel_BA
test2_kernel_UDOI<-kerneloverlap(test2_kernel_df[,1], grid=200, meth="UDOI")
test2_kernel_UDOI
# Try the kernelbb and then kerneloverlaphr approach
test2_kernel$timestamp
<-as.POSIXct(strptime(as.character(test2_kernel$timestamp),"%d/%m/%Y
%H:%M"), "GMT")
test2_coords<-test2_kernel[, c(2,3)]
test2_coords_meters<-project(as.matrix(test2_coords), "+proj=utm +zone=55
+south +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs")
test2_coords_meters<-as.data.frame(test2_coords_meters)
colnames(test2_coords_meters)<-c("Longitude_meters", "Latitude_meters")
test2_kernel$Longitude_meters<-test2_coords_meters$Longitude_meters
test2_kernel$Latitude_meters<-test2_coords_meters$Latitude_meters
test2_ltraj<-as.ltraj(xy =
test2_kernel[,c("Longitude_meters","Latitude_meters")], date =
test2_kernel$timestamp, id=test2_kernel$id)
lik <- liker(test2_ltraj, sig2 = 4, rangesig1 = c(0, 10))
test2_kernelbb<-kernelbb(test2_ltraj, sig1=5, sig2=4, grid = 200)
test2_kernelbb_BA<-kerneloverlaphr(test2_kernelbb, grid=200, meth="BA")
test2_kernelbb_BA
test2_kernelbb_UDOI<-kerneloverlaphr(test2_kernelbb, grid=200, meth="UDOI")
test2_kernelbb_UDOI.
In the kerneloverlap approach, I am getting high indices, like this:
test2_kernel_BA
1st_trip 2nd_trip
1st_trip 0.9998753 0.8604860
2nd_trip 0.8604860 0.9998635
> test2_kernel_UDOI
1st_trip 2nd_trip
1st_trip 1.4040774 0.7812684
2nd_trip 0.7812684 1.3015604
In the kerneloverlap approach, I am getting low indices, like this:
test2_kernelbb_BA
1st_trip 2nd_trip
1st_trip 1.0000000 0.1375754
2nd_trip 0.1375754 1.0000000
> test2_kernelbb_UDOI
1st_trip 2nd_trip
1st_trip 6.46508189 0.02059335
2nd_trip 0.02059335 2.75579098
Could anyone tell me what I'm missing please? Your help would be greatly
appreciated!
Kind regards,
Elodie
Elodie Camprasse
6/187 Auburn Road
Hawthorn, VIC 3122
Australia
Email: <mailto:elodie.campra...@gmail.com> elodie.campra...@gmail.com
Website: <http://hors-des-sentiers-battus.e-monsite.com/>
http://hors-des-sentiers-battus.e-monsite.com/
Mobile: <tel:%28%2B61%29%20049%20794%200793> (+61) 049 794 0793
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.faunalia.it/pipermail/animov/attachments/20140818/ba58755b/attachment-0001.html>
------------------------------
Subject: Digest Footer
_______________________________________________
AniMov mailing list
AniMov@faunalia.it
http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov
------------------------------
End of AniMov Digest, Vol 102, Issue 8
**************************************
_______________________________________________
AniMov mailing list
AniMov@faunalia.it
http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov