On Fri, 6 Aug 2010, Laetitia Blanc wrote:
Can the KDE be run with the correct coordinates in the specified
projection on the input data, so that thw window matches the outlines? Can
you re-write the code making res to generate a SpatialGridDataFrame? The
input points must have a projection. Without a shared projection (CRS,
coordinate reference system) this is not possible.
Roger
I have taken into account your advice.
I have loaded my input data with the correct CRS (lambert II) in order to
match with the map :
d <- read.csv('total.csv', header = TRUE) #, sep = ",", quote="\"",
dec=".",fill = TRUE, comment.char="", ...)
d <- d[-(1:10),]
d2 <- data.frame (d)
spd2 <- d2
#transformation into SpatialPointDataFrame
coordinates ( spd2 ) <- ~ x + y
proj4string ( spd2 ) <- CRS("+init=epsg:27572")
Then, i have realised my KDE :
source("stkde2.txt")
res <- array(0,c(20,20,length(unique(spd2$an)))) # créer 30
(length(unique(d$an))) tableaux de 20x20.
for (i in 1:length(unique(spd2$an))) {
spd3 <- subset(spd2,spd2$an == 1979+i) # résumé en conservant que 1973+i où
i=30 donc en gardant que 2009
lynxkde<-stkde(xlong=spd3$x, ylat=spd3$y, ztime=spd3$an,
xgrids=20,ygrids=20,bwmethod="cv.ml")
res[,,i] <- lynxkde$dens
}
At that moment, the window where the KDE is running has te correct unit
(meters) (xrange : >from 823 000 to 981 000 ; yrange : from 2 079 500 to 2
305 500)
No, this is a misunderstanding. Until you get this right, please forget
the simulation, make sure everything is correct for a single period
first. I am pretty sure that lynxkde$dens is a matrix, and if you
take the trouble to read help(image), you'll see that if the first
argument is a matrix, then the window is set to 0,1. You need to pass the
cell centres of the xgrids and ygrids to the image call. Beware that if
the spd3$x and spd3$y change by i, the grid will be different each time,
evenj though it is 20x20, it will start somewhere else, and have a
different cell dimension.
I think that you are using a function from the stam package - you need to
adapt it to pass the x.seq and y.seq values now generated internally in as
arguments, which will permit you to make sure that you have full control
of the location of the output grids. If you then pass the same x.seq and
y.seq to image() as the first two arguments, and the matrix as the third
argument, it will use them to specify its axes, giving you the correct
window. You need to be sure that your x.seq and y.seq contain all of the
xlong and xlat for all periods. If stkde took a GridTopology object as a
grid definition, the output could easily be used to create a
SpatialGridDataFrame object, which would be very easy to plot in register
with your boundaries, but this is something for the package maintainer.
Hope this helps,
Roger
PS. CC-ing the stam maintainer.
But then, when i create my animation, the window has the x and y ranges :
0:1. I think the >probleme comes from my animation. Moreover, when i try to
insert my map in my animation, i still >have an alternation between the map
and the kernel >rather than the kernel ON the map (like
overlay/superposition...) :
library(animation)
library(MASS)
# create a HTML page from my animation :
pres.ani <- function()
{
for (i in 1:length(unique(spd2$an))) {
image(res[,,i],xlab='',ylab='',main=unique(spd2$an)[i])
contour(res[,,i],add=T)
plot(france,xlim=range(823000:981000),ylim=range(2079500:2305500),add=TRUE)
Sys.sleep(.5)
}
}
ani.start(interval=.5,title = "movie presence lynx", outdir = getwd())
pres.ani()
ani.stop()
I think i must solve the problem by changing my animation but i don't know
how. Do you have an >idea ? Do you think that it can be because i have
spatial objects ?
--
View this message in context:
http://r-sig-geo.2731867.n2.nabble.com/clip-a-SpatialPolygonsDataFrame-tp5376909p5379892.html
Sent from the R-sig-geo mailing list archive at Nabble.com.
_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
--
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: roger.biv...@nhh.no
_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
