Re: [R-sig-Geo] raster: stackApply problems..
Unfortunately the names are not always in ascending order. This is the
result of my data.
names : index_4, index_5, index_6, index_7, index_1, index_2, index_3
min values : 3, 3, 3, 3, 3, 3, 3
max values : 307.0, 297.5, 311.0, 313.0, 468.0, 290.0, 302.0
And worst of all, it is not a proper match with indices.
If I run it with clusterR then the result is different:
names : layer.1, layer.2, layer.3, layer.4, layer.5, layer.6, layer.7
min values : 3, 3, 3, 3, 3, 3, 3
max values : 307.0, 297.5, 311.0, 313.0, 468.0, 290.0, 302.0
The solution is to reorder the layers of the stack so that the
stackApply indices are in ascending order e.g. 1,1,1,2,2,2,3,3,3 ...
My indices of my data was like that:
4 5 6 7 1 2 3 4 5 6 7 1 2 3 5 6 7
I've reported this behavior here
https://github.com/rspatial/raster/issues/82
On 11/20/19 3:05 PM, Ben Tupper wrote:
> Hi,
>
> That is certainly is unexpected to have two different naming styles.
> It's not really solution to take to the bank, but you could simply
> compose your own names assuming that the layer orders are always
> returned in ascending index order.
> Would that work for you
>
> ### start
> library(raster)
>
> # Compute layer names for stackApply output
> #
> # @param index numeric, 1-based layer indices used for stackApply function
> # @param prefix character, prefix for names
> # @return character layers names in index order
> layer_names <- function(index = c(2,2,3,3,1,1), prefix = c("layer.",
> "index_")[1]){
> paste0(prefix, sort(unique(index)))
> }
>
> indices <- c(2,2,3,3,1,1)
>
> r <- raster()
> values(r) <- 1
> # simple sequential stack from 1 to 6 in all cells
> s <- stack(r, r*2, r*3, r*4, r*5, r*6)
> s
>
> beginCluster(2)
> res <- clusterR(s, stackApply, args = list(indices=indices, fun = mean))
> raster::endCluster()
> names(res) <- layer_names(indices, prefix = "foobar.")
> res
>
> res2 <- stackApply(s, indices, mean)
> names(res2) <- layer_names(indices, prefix = "foobar.")
> res2
> ### end
>
>
> On Wed, Nov 20, 2019 at 1:36 AM Leonidas Liakos via R-sig-Geo
> wrote:
>> This is not a reasonable solution. It is not efficient to run stackapply
>> twice to get the right names. Each execution can take hours.
>>
>>
>> Στις 20/11/2019 3:30 π.μ., ο Frederico Faleiro έγραψε:
>>> Hi Leonidas,
>>>
>>> both results are in the same order, but the name is different.
>>> You can rename the first as in the second:
>>> names(res) <- names(res2)
>>>
>>> I provided an example to help you understand the logic.
>>>
>>> library(raster)
>>> beginCluster(2)
>>> r <- raster()
>>> values(r) <- 1
>>> # simple sequential stack from 1 to 6 in all cells
>>> s <- stack(r, r*2, r*3, r*4, r*5, r*6)
>>> s
>>> res <- clusterR(s, stackApply, args = list(indices=c(2,2,3,3,1,1), fun =
>>> mean))
>>> res
>>> res2 <- stackApply(s, c(2,2,3,3,1,1), mean)
>>> res2
>>> dif <- res - res2
>>> # exatly the same order because the difference is zero for all layers
>>> dif
>>> # rename
>>> names(res) <- names(res2)
>>>
>>> Best regards,
>>>
>>> Frederico Faleiro
>>>
>>> On Tue, Nov 19, 2019 at 4:15 PM Leonidas Liakos via R-sig-Geo <
>>> r-sig-geo@r-project.org> wrote:
>>>
I run the example with clusterR:
no_cores <- parallel::detectCores() -1
raster::beginCluster(no_cores)
?? res <- raster::clusterR(inp, raster::stackApply, args =
list(indices=c(2,2,3,3,1,1),fun = mean))
raster::endCluster()
And the result is:
> res
class?? : RasterBrick
dimensions : 180, 360, 64800, 3?? (nrow, ncol, ncell, nlayers)
resolution : 1, 1?? (x, y)
extent : -180, 180, -90, 90?? (xmin, xmax, ymin, ymax)
crs?? : +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0
source : memory
names?? : layer.1, layer.2, layer.3
min values : 1.5, 3.5, 5.5
max values : 1.5, 3.5, 5.5??
layer.1, layer.2, layer.3 (?)
So what corrensponds to what?
If I run:
res2 <- stackApply(inp,c(2,2,3,3,1,1),mean)
The result is:
> res2
class : RasterBrick
dimensions : 180, 360, 64800, 3 (nrow, ncol, ncell, nlayers)
resolution : 1, 1 (x, y)
extent : -180, 180, -90, 90 (xmin, xmax, ymin, ymax)
crs: +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0
source : memory
names : index_2, index_3, index_1
min values : 1.5, 3.5, 5.5
max values : 1.5, 3.5, 5.5
There is no consistency with the names of the output and obscure
correspondence with the indices in the case of clusterR
[[alternative HTML version deleted]]
___
R-sig-Geo mailing list
R-sig-Geo@r-project.org
Re: [R-sig-Geo] raster: stackApply problems..
Hi,
That is certainly is unexpected to have two different naming styles.
It's not really solution to take to the bank, but you could simply
compose your own names assuming that the layer orders are always
returned in ascending index order.
Would that work for you
### start
library(raster)
# Compute layer names for stackApply output
#
# @param index numeric, 1-based layer indices used for stackApply function
# @param prefix character, prefix for names
# @return character layers names in index order
layer_names <- function(index = c(2,2,3,3,1,1), prefix = c("layer.",
"index_")[1]){
paste0(prefix, sort(unique(index)))
}
indices <- c(2,2,3,3,1,1)
r <- raster()
values(r) <- 1
# simple sequential stack from 1 to 6 in all cells
s <- stack(r, r*2, r*3, r*4, r*5, r*6)
s
beginCluster(2)
res <- clusterR(s, stackApply, args = list(indices=indices, fun = mean))
raster::endCluster()
names(res) <- layer_names(indices, prefix = "foobar.")
res
res2 <- stackApply(s, indices, mean)
names(res2) <- layer_names(indices, prefix = "foobar.")
res2
### end
On Wed, Nov 20, 2019 at 1:36 AM Leonidas Liakos via R-sig-Geo
wrote:
>
> This is not a reasonable solution. It is not efficient to run stackapply
> twice to get the right names. Each execution can take hours.
>
>
> Στις 20/11/2019 3:30 π.μ., ο Frederico Faleiro έγραψε:
> > Hi Leonidas,
> >
> > both results are in the same order, but the name is different.
> > You can rename the first as in the second:
> > names(res) <- names(res2)
> >
> > I provided an example to help you understand the logic.
> >
> > library(raster)
> > beginCluster(2)
> > r <- raster()
> > values(r) <- 1
> > # simple sequential stack from 1 to 6 in all cells
> > s <- stack(r, r*2, r*3, r*4, r*5, r*6)
> > s
> > res <- clusterR(s, stackApply, args = list(indices=c(2,2,3,3,1,1), fun =
> > mean))
> > res
> > res2 <- stackApply(s, c(2,2,3,3,1,1), mean)
> > res2
> > dif <- res - res2
> > # exatly the same order because the difference is zero for all layers
> > dif
> > # rename
> > names(res) <- names(res2)
> >
> > Best regards,
> >
> > Frederico Faleiro
> >
> > On Tue, Nov 19, 2019 at 4:15 PM Leonidas Liakos via R-sig-Geo <
> > r-sig-geo@r-project.org> wrote:
> >
> >> I run the example with clusterR:
> >>
> >> no_cores <- parallel::detectCores() -1
> >> raster::beginCluster(no_cores)
> >> ?? res <- raster::clusterR(inp, raster::stackApply, args =
> >> list(indices=c(2,2,3,3,1,1),fun = mean))
> >> raster::endCluster()
> >>
> >> And the result is:
> >>
> >>> res
> >> class?? : RasterBrick
> >> dimensions : 180, 360, 64800, 3?? (nrow, ncol, ncell, nlayers)
> >> resolution : 1, 1?? (x, y)
> >> extent : -180, 180, -90, 90?? (xmin, xmax, ymin, ymax)
> >> crs?? : +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0
> >> source : memory
> >> names?? : layer.1, layer.2, layer.3
> >> min values : 1.5, 3.5, 5.5
> >> max values : 1.5, 3.5, 5.5??
> >>
> >>
> >> layer.1, layer.2, layer.3 (?)
> >>
> >> So what corrensponds to what?
> >>
> >>
> >> If I run:
> >>
> >> res2 <- stackApply(inp,c(2,2,3,3,1,1),mean)
> >>
> >> The result is:
> >>
> >>> res2
> >> class : RasterBrick
> >> dimensions : 180, 360, 64800, 3 (nrow, ncol, ncell, nlayers)
> >> resolution : 1, 1 (x, y)
> >> extent : -180, 180, -90, 90 (xmin, xmax, ymin, ymax)
> >> crs: +proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0
> >> source : memory
> >> names : index_2, index_3, index_1
> >> min values : 1.5, 3.5, 5.5
> >> max values : 1.5, 3.5, 5.5
> >>
> >> There is no consistency with the names of the output and obscure
> >> correspondence with the indices in the case of clusterR
> >>
> >>
> >> [[alternative HTML version deleted]]
> >>
> >> ___
> >> R-sig-Geo mailing list
> >> R-sig-Geo@r-project.org
> >> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
> >>
> >
>
>
> --
> Λιάκος Λεωνίδας, Γεωγράφος
> https://www.geographer.gr
> PGP fingerprint: 5237 83F8 E46C D91A 9FBB C7E7 F943 C9B6 8231 0937
>
> ___
> R-sig-Geo mailing list
> R-sig-Geo@r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
--
Ben Tupper
Bigelow Laboratory for Ocean Science
West Boothbay Harbor, Maine
http://www.bigelow.org/
https://eco.bigelow.org
___
R-sig-Geo mailing list
R-sig-Geo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
