[R-sig-Geo] Problem with cell numbers when using a raster extent object

[Robert J. Hijmans]

Ned, that's a bug that only occurred with large rasters (than are not
processed in one chunk). This has been fixed for version 2.1-71
Note that Andrew correctly pointed out that in any case it should be
"ext=commonExt", not "extent=commonExt".
Thanks for the helpful example. Best, Robert

On Mon, Nov 25, 2013 at 11:28 AM, Ned Horning <horning at amnh.org> wrote:
> Robert,
>
> Here are examples of the two versions.
>
> ### This one does not select points within the extent ###
>
> library(raster)
> crs_predImage <- CRS("+proj=utm +zone=33 +datum=WGS84 +units=m +no_defs
> +ellps=WGS84 +towgs84=0,0,0")
> predImage <- raster(nrows=7561, ncols=8521, xmn=240585, xmx=496215,
> ymn=5297985, ymx=5524815)
> projection(predImage) <- crs_predImage
> predImage[] <- 1
>
> commonExt <- extent(366760.6 , 425233.5, 5396052, 5450158)
>
> sampleCells <- sampleRandom(predImage, size=20, sp=TRUE, extent=commonExt)
>
> plot(predImage)
> plot(commonExt, add=T)
> plot(sampleCells, add=T)
>
>
> ### This one does work ###
>
> library(raster)
> crs_predImage <- CRS("+proj=utm +zone=33 +datum=WGS84 +units=m +no_defs
> +ellps=WGS84 +towgs84=0,0,0")
> predImage <- raster(nrows=7561, ncols=8521, xmn=240585, xmx=496215,
> ymn=5297985, ymx=5524815)
> projection(predImage) <- crs_predImage
> predImage[] <- 1
>
> commonExt <- extent(366760.6 , 425233.5, 5396052, 5450158)
>
> sampleCells <- sampleRandom(crop(predImage,commonExt), size=20, sp=TRUE,
> ext=commonExt, na.rm=TRUE)
>
> plot(predImage)
> plot(commonExt, add=T)
> plot(sampleCells, add=T)
>
>
>
> On 11/25/2013 01:04 PM, Robert J. Hijmans wrote:
>>
>> Ned,
>>
>> Hard to see what does not work here. Perhaps it is just the [i] in
>> xyFromCell(predImage, sampleCells[i])?
>> Can you please provide a self-contained example, perhaps based on the
>> below?
>>
>> library(raster)
>> r <- raster(system.file("external/rlogo.grd", package="raster"))
>> e <- extent(r) + 100
>> s <- sampleRandom(r, size=100, cells=TRUE, extent=e)[,1]
>> xy <- xyFromCell(r, s)
>>
>> plot(e)
>> plot(r, add=T, legend=F)
>> points(xy)
>>
>>
>>
>> On Mon, Nov 25, 2013 at 9:42 AM, Ned Horning <horning at amnh.org> wrote:
>>>
>>> Hi -
>>>
>>> I have a script that calculates a common extent between two images than
>>> then
>>> uses the resulting extent to restrict the random selection of cells from
>>> a
>>> raster using the “sampleRandom” function. I use the following function:
>>>
>>> sampleCells <- sampleRandom(predImage, size=numSamps, cells=TRUE,
>>> extent=commonExt)[,1]
>>>
>>> Next I use "xyFromCell(predImage, sampleCells[i])" to calculate the
>>> coordinates of the center of a cell and all of the cell centers fall
>>> outside
>>> of the extent that I am using. Using "sp=TRUE" I get a similar result -
>>> all
>>> of the points are outside of the extent:
>>>
>>> sampleCells <- sampleRandom(predImage, size=numSamps, sp=TRUE,
>>> ext=commonExt)
>>>
>>> To get around this problem I tried to physically clip the image before
>>> running “sampleRandom” and the results are similar. The following line
>>> also
>>> produces cells that are outside of the extent “commonExt” when I use
>>> “xyFromCell”:
>>>
>>> sampleCells <- sampleRandom(crop(predImage,commonExt), size=numSamps,
>>> ext=commonExt, na.rm=TRUE)
>>>
>>> However, if I run the following line using"clip" and "sp=TRUE" the sample
>>> points fall within the extent:
>>>
>>> sampleCells <- sampleRandom(crop(predImage,commonExt), size=numSamps,
>>> sp=TRUE, ext=commonExt, na.rm=TRUE)
>>>
>>> So, there is a work-around but I'd prefer not to have the overhead of
>>> clipping the image.
>>>
>>> This seems somewhat similar to a problem I reported a few months ago and
>>> was
>>> fixed. Can anyone replicate this problem? Am I doing something wrong?
>>>
>>> I am using R version: 3.0.2 and Raster package version: 2.1-49 on Ubuntu
>>> 12.04 64 bit
>>>
>>> Ned
>>>
>>> _______________________________________________
>>> R-sig-Geo mailing list
>>> R-sig-Geo at r-project.org
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>