alexgenin
commented
6 years ago Hello, thanks for your interest in our work !
Yes, it would be really nice to have functions working directly on raster objects, it's actually a natural application of the package. It would require a bit of work to read raster attributes (resolution, etc.) and adapt the function outputs to them (e.g. having spectra displayed with proper units on the x-axis), but it's definitely doable.
Right now, most of the functions in spatialwarnings can be applied to raster objects by explicitly converting individual raster bands to matrices (using as.matrix), e.g:
library(raster)
#> Loading required package: sp
library(spatialwarnings)
#> This is spatialwarnings 1.2
#> Use options(mc.cores = <n>) to set up multi-core processing (unix only)
# It makes no sense computing indicators on those images, they just are
# example geotiffs
example_files <- c("https://download.osgeo.org/geotiff/samples/spot/chicago/SP27GTIF.TIF",
"https://download.osgeo.org/geotiff/samples/spot/chicago/UTM2GTIF.TIF")
list_of_raster_objects <- lapply(example_files, function(f) {
tmpfile <- tempfile()
download.file(f, destfile = tmpfile)
raster(tmpfile)
})
list_of_matrices_objects <- lapply(list_of_raster_objects, as.matrix)
generic_sews(list_of_matrices_objects)
#> Warning in FUN(X[[i]], ...): Input matrix has continous values but will be
#> coarse-grained anyway. Set subsize=1 to disable coarse graining.
#> Warning in FUN(X[[i]], ...): Input matrix has continous values but will be
#> coarse-grained anyway. Set subsize=1 to disable coarse graining.
#> Generic Spatial Early-Warnings
#>
#> 2 matrices (size: 929x699)
#>
#> Mat. # Mean Moran's I Skewness Variance
#> 1 115 0.64 0.68 1177
#> 2 115 0.63 0.68 1175
#>
#> Use as.data.frame() to retrieve values in a convenient formOf course, for now it means that indicators can only be applied on one band at a time (i.e. on RasterLayer objects or on one band of a Raster{Brick/Stack} object), and raster attributes are not used (extent/resolution, etc).
Cheers
slarge
Hi!
Nice that you have just updated the package :-) Have you thought about raster datasets as input data type to the spatial warnings function? When working with remote sensing data, for instance, this would be super handy!
Cheers, Sophie