can raster create multi-layer objects with different modes? - r

Can a raster object (in R) have layers of different mode (data type)?
On the face of it it seems we are always forced to one type:
library(raster)
## create a SpatialPixelsDataFrame with (trivially) two different "layer" types
d <- data.frame(expand.grid(x = 1:10, y = 2:11), z = 1:100, a = sample(letters, 100, replace = TRUE), stringsAsFactors = FALSE)
coordinates(d) <- 1:2
gridded(d) <- TRUE
## now coerce this to a raster brick or stack and our "a" is crushed to numeric NA
all(is.na(getValues(brick(d)[[2]])))
[1] TRUE
Is there anything like a rasterDataFrame?
Also, note that we presumably cannot use R's factors since the raster#data is a matrix, or otherwise coerced to numeric/integer. Am I missing something?


The raster package provides the ability to create rasters with a categorical variable, and the rasterVis package includes functions for plotting them. The ratify function allows a raster to include a lookup table relating the underlying raster integer values to other values, which can be character. This directly allows the use of any other mode of value in the levels part of the ratified raster.
Here's an example.
library(rasterVis)
r <- raster(xmn = 0, xmx = 1, ymn = 0, ymx = 2, nrow = 10, ncol = 11,
crs = as.character(NA))
r[] <- sample(seq_along(letters[1:5]), ncell(r), replace = TRUE)
## ratify the raster, and set up the lookup table
r <- ratify(r)
rat <- levels(r)[[1]]
rat$value <- letters[1:5]
rat$code <- 1:5
## workaround for limitation as at 2013-05-01
## see https://stat.ethz.ch/pipermail/r-sig-geo/2013-May/018180.html
rat$code <- NULL
levels(r) <- rat
levelplot(r)
There are coming updates to rasterVis that make the workaround above unnecessary.

Related

R function to convert polygon (sf, wkt) into mask (matrix, array)

I have an image stored as matrix with grayscale for each pixel.
On this image I use SLIC algorithm to divide it into areas.
So I get a simple feature (sf) with polygons, I am able to extract in well-known-text (wkt).
But what I really need is a matrix/mask (same dimension as my pixel-image-matrix) storing the id of the polygon each pixel belongs to. For example the pixel image[1,2] belongs to polygon 5, then mask[1,2] <- 5.
I add some code to give example of my porblem (for a random "image"):
mat <- array(runif(10000, min=0, max=500), dim=c(100,100))
# SLIC
library(supercells);
library(sf);
library(terra);
# make spatial raster from matrix
raster <- rast(mat);
rasterSLIC <- supercells(raster, k = 50, compactness = 1, dist_fun = "euclidean", avg_fun = "mean");
plot(raster);
plot(st_geometry(rasterSLIC), add = TRUE, lwd = 0.2);
point <- st_cast(rasterSLIC$geometry[2], to="POINT");
coord <- st_coordinates(point);
# what I want:
goal <- array(c(1,1,1,2,2,1,2,3,3), dim=c(3,3));
image(goal);
goal;
I would like to have something that helps me turning coords into such a mask/matrix I gave a small example for in goal.

You can use terra::rasterize
Example data
library(terra)
# polygons
f <- system.file("ex/lux.shp", package="terra")
v <- vect(f)
# arbitrary raster
r <- rast(v, res=.01)
Solution:
rid <- rasterize(v, r, 1:nrow(r))
#or
v$ID <- 1:nrow(v)
rid <- rasterize(v, r, "ID")
Illustration
plot(rid, type="classes")
text(v)
lines(v)
To get the a matrix of the raster values you can do
m <- as.matrix(rid, wide=TRUE)
With your more specific example, you could do
library(supercells);
library(terra)
set.seed(1)
mat <- array(runif(10000, min=0, max=500), dim=c(100,100))
r <- rast(mat)
SLIC <- supercells(r, k = 50, compactness = 1, dist_fun = "euclidean", avg_fun = "mean");
x <- rasterize(SLIC, r, "supercells")
xm <- as.matrix(x, wide=TRUE)
plot(x);
s <- vect(SLIC)
lines(s)

Compile multiple raster extractions to one table

I have 3 rasters in which I have extracted data from using a polyline from a shapefile. Currently, I have the extraction as 3 separate lists. Is there A way I can do an extraction from all three rasters and compile them to one table with different columns for the data from each raster?
This is the current code I am using
Harney_Transects <- readOGR(dsn = ".", layer = "Transect_HN")
MeanTreeHeightHarneyBefore=raster('HN_TrMean_B_Clip.tif')
ScanAngleHarneyBefore= raster('HNScanAngle_B_Clip.tif')
MeanShrubHeightHarneyBefore= raster('HN_MeanShrub_B_Clip.tif')
Extraction_Shrub_Harney= extract(MeanShrubHeightHarneyBefore,Harney_Transects)
Extraction_Tree_Harney= extract(MeanTreeHeightHarneyBefore,Harney_Transects)
Extraction_ScanAngle_Harney= extract(ScanAngleHarneyBefore,Harney_Transects)

In short, you can stack() all the rasters you want to extract data from, and extract from the stack.
Here's a fully reproducible example using two rasters and a SpatialLines object, like you have in your question. Skip to the last code chunk for a direct answer to your question.
library(sp)
library(raster)
# function to generate random rasters
gen_raster <- function(){
r <- raster(nrows = 10, ncols = 10, res = 1,
xmn = 0, xmx = 10, ymn = 0, ymx = 10,
vals = rnorm(100, 5, 1))
return(r)
}
# generate 2 random rasters
r1 <- gen_raster()
r2 <- gen_raster()
# view
par(mfrow = c(1,2))
plot(r1, main = "raster 1"); plot(r2, main = "raster 2")
dev.off()
# generate transect (`SpatialLines` object)
m <- as.matrix(data.frame(x = 5.5, y = seq(0, 10, 1)))
l <- list(Lines(Line(m), "m"))
l <- SpatialLines(l)
# view the transect
plot(r1, main = "raster 1 with transect"); lines(l)
Running extract on the stacked rasters returns a list with a matrix in it. The last thing you'll want is to pull this out as a data.frame, which is a bit tricky.
rs <- stack(r1, r2) # stack any amount of rasters to extract from
re <- extract(rs, l) # extract at locations `l`
do.call(rbind.data.frame, re) # convert to data.frame
layer.1 layer.2
1 4.586890 5.115136
2 4.780503 5.093281
3 6.877302 3.337345
4 5.913230 3.755099
5 4.907834 4.887160
6 5.576908 5.386136
7 3.572350 5.225392
8 4.778727 5.391765
9 6.600041 4.205841
10 6.946321 5.544172
The names of the columns are the names of the raster layers in the stack. You can access these names with names(rs), and modify them with names(rs) <- c("new_name_1", "new_name_2").

Removing the third dimension in SpatialPointsDataFrame

I have a SpatialPoints object with 3 dimensions :
x <- c(1,1,1,2,2,2,3,3,3)
y <- c(1,2,3,1,2,3,1,2,3)
z <- c(1,3,5,2,1,2,1,2,3)
xyz <- cbind(x,y,z)
ss <- SpatialPoints(xyz)
dimensions(ss)
And a raster object:
rr <- raster(matrix(round(runif(49,0,10)),7,7), xmn=0, xmx=4, ymn=0, ymx=4, crs=NA, template=NULL)
I want to extract the raster values using the SpatialPoints object:
extract(rr,ss)
#Error in .xyValues(x, coordinates(y), ..., df = df) :
# xy should have 2 columns only.
#Found these dimensions: 9, 3
You can visualize the data if you want:
plot(rr)
plot(ss, add=T)
So the problem is that the extract function of the raster package require a 2 dimension SpatialPoints object. Mine (in my real data) in 3 dimensional. Is there a way to drop the 3rd dimension of my point shape? I've tried:
coordinates(ss) <- coordinates(ss)[,-3]
#Error in `coordinates<-`(`*tmp*`, value = c(1, 1, 1, 2, 2, 2, 3, 3, 3, :
# setting coordinates cannot be done on Spatial objects, where they have #already been set
I don't want to have to rebuild my shape from scratch.

Just overwrite the coords slot of the S4 object:
ss#coords <- ss#coords[, 1:2]
I don't know how your SpatialPoints object is created, but if you use rgdal::readOGR there is a pointDropZ argument (default FALSE)

#rcs answer is better, but this works as well:
ss <- SpatialPoints(coordinates(ss)[,-3])
and if you have a SpatialPointsDataFrame:
ss <- SpatialPointsDataFrame(coordinates(ss)[,-3], ss#data,proj4string=CRS(proj4string(ss)))

get coordinates of a patch in a raster map (raster package in R)

I have a raster map with many patches (clumps of continguous cells with the same value). What I need to do is to obtain the coordinates of the center (or close to the center) of each patch.
I am very unexperienced with raster package but it seems I can get coordinates only if I know the position of the cells in the map. Is there any way to get coordinates giving a value of the cells instead? Thank you

If by patch you mean clumps, Raster package allows you to find , and isolate, clumps. Taking the clump() raster package example, and extending it:
library(raster)
library(igraph)
detach("package:coin", unload=TRUE)
r <- raster(ncols=12, nrows=12)
set.seed(0)
r[] <- round(runif(ncell(r))*0.7 )
rc <- clump(r)
clump_id <- getValues(rc)
xy <- xyFromCell(rc,1:ncell(rc))
df <- data.frame(xy, clump_id, is_clump = rc[] %in% freq(rc, useNA = 'no')[,1])
df[df$is_clump == T, ]
plot(r)
plot(rc)
text(df[df$is_clump == T, 1:2], labels = df[df$is_clump == T, 3])
May not be as interesting as you could expect.
You do it all over with directions = 4
rc <- clump(r, directions = 4)
clump_id <- getValues(rc)
xy <- xyFromCell(rc,1:ncell(rc))
df <- data.frame(xy, clump_id, is_clump = rc[] %in% freq(rc, useNA = 'no')[,1])
df[df$is_clump == T, ]
to get
and maybe clump 'centroids'
dfm <- ddply(df[df$is_clump == T, ], .(clump_id), summarise, xm = mean(x), ym = mean(y))
plot(rc)
text(dfm[, 2:3], labels = dfm$clump_id)
Notes:
There will be an error if you try to use clump() without first
detach modeltools library. modeltools is called by coin and maybe
other statistical libraries.

You could take the mean of the coordinates of each patch:
# some dummy data
m <- matrix(c(
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0,0,4,4,0,
0,0,0,0,0,0,0,0,1,1,1,1,1,0,0,0,0,0,0,4,4,0,
0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0,4,4,0,
0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,
0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,
0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,
0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,
0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,
0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,
0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,
0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,
0,0,0,0,0,0,0,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,1,1,1,1,1,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0,0,0,0,0,
0,0,2,3,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,
0,0,2,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0), nrow=20, byrow=T)
# create a raster
r <- raster(m)
# convert raster to points
p <- data.frame(rasterToPoints(r))
# filter out packground
p <- p[p$layer > 0,]
# for each patch calc mean coordinates
sapply(split(p[, c("x", "y")], p$layer), colMeans)

extract value from raster stack from spatialpolygondataframe

I have a raster stack with 27 rasters in it. I have 27 corresponding polygons in a spatial polygon data frame. I want to take polygon[i] overlay it on raster[i], extract and sum the values from raster [i], get a count of the number of cells within the polygon[i] and then divide the sum value by the # of cells. In other words, the raster is a utilization distribution or a kernel density of use. I want to know much use is occurring in the area of the polygon where it is overlapping the raster. I want to divide by the number of cells in the polygon to take into account the size of the polygon.
I have a script that was given to me that does this, only it was written with the intention of extracting data from 1 raster only by any number of spatial polygons in the data frame. It works, its ugly, and I now would like to convert it to something more stream line. I only wish I had someone around me who could help because this might take a while?
This is code Ive been given and my summary of what I think is going on:
msum99Kern07 = SpatialPolygonDataFrame (many polygons)
KERNWolfPIX07m = Raster (this is a single raster, I have 27 rasters I put into a stack
)
#Extracting value from raster to many polygons
sRISK_Moose07m<- extract(KERNWolfPIX07m, msum99Kern07,df=FALSE,method='bilinear')
#Calculate THE SUM FOR EACH polygon#
sRISK_Moose07m<-unlist(lapply(sRISK_Moose07m, function(x) if (!is.null(x)) sum(x, na.rm=TRUE) else NA ))
sRISK_Moose07m<-as.data.frame(sRISK_Moose07m)
#Im not sure why these next commands are needed Im only guessing
#data.frame(levels) as there are many polygons creating a dataframe to put the info into
ID_SUM_07<-as.data.frame(levels(as.factor(msum07locs$ID2)))
#ADD ID TO THE risk data frame
sRISK_Moose07m$ID<-ID_SUM_07[,1]
#NUMBER OF CELLS WITHIN POLYGON EXTRACT CELLS/ POLYGON
NB_SUM2007m<-cellFromPolygon(KERNWolfPIX07m, msum99Kern07)
NB_SUM07m<-unlist(lapply(NB_SUM2007m, function(x) if (!is.null(x)) length(x) else NA ))
#####CONVERT TO DATA FRAME
NB_SUM07m<-as.data.frame(NB_SUM07m)
###ADD THE NB OF CELLS TO THE RISK_SUM FILE###
sRISK_Moose07m$NB_CELLS<-NB_SUM07m[,1]
###DIVIDING VALUE by NB CELLS##
sRISK_Moose07m$DIVID<-sRISK_Moose07m$sRISK_Moose07m/sRISK_Moose07m$NB_CELLS
Now, I have my spatial polygon data frame with 27 polygons and my raster stack with 27 rasters. I want to select the raster[i] and polygon[i] and extract, sum, and calculate the kernel density of the overlapping area. One side thing to keep in mind, I may get an error because it is possible that the polygon and raster do not overlap...I don't know how to check for this in R at all.
My script I have started:
moose99kern = spatial polygon data frame 27 moose
Rastwtrial = stack of 27 rasters having the same unique name as the ID in moose99kern
mkernID=unique(moose99kern$id)
for (i in length(mkernID)){
r = Rastwtrial[Rastwtrial[[i]]== mkernID[i]] #pick frm Rasterstack the raster that has the same name
mp = moose99kern[moose99kern$id == mkernID[i]] #pick from spatialpolygondataframe the polygon that has the same name
RISK_MooseTrial<- extract(r, mp, df=T, method'bilinear')
risksum = (RISK_MooseTrial, function(x) if (!is.null(x)) sum(x, na.rm=TRUE) else NA )#sum all the values that were extracted from the raster
My script doesn't even start to work because I don't know how to index a raster stack. But even still, going through 1 raster/1polygon at a time, Im not sure what to do next in the code. If this is too much for StackOverflow I apologize. Im just seriously stuck and have no where to turn.
Here is test data with 2 individuals for polygons
dput(mtestpoly)
new("SpatialPolygonsDataFrame"
, data = structure(list(id = structure(1:2, .Label = c("F01001_1", "F07002_1"
), class = "factor"), area = c(1259.93082578125, 966.364499511719
)), .Names = c("id", "area"), row.names = c("F01001_1", "F07002_1"
), class = "data.frame")
, polygons = list(<S4 object of class structure("Polygons", package = "sp")>,
<S4 object of class structure("Polygons", package = "sp")>)
, plotOrder = 1:2
, bbox = structure(c(6619693.77161797, 1480549.31292137, 6625570.48348294,
1485861.5586371), .Dim = c(2L, 2L), .Dimnames = list(c("x", "y"
), c("min", "max")))
, proj4string = new("CRS"
, projargs = NA_character_
dput(Rastwtest)
new("RasterStack"
, filename = ""
, layers = list(<S4 object of class structure("RasterLayer", package = "raster")>,
<S4 object of class structure("RasterLayer", package = "raster")>)
, title = character(0)
, extent = new("Extent"
, xmin = 1452505.6959799
, xmax = 1515444.7110552
, ymin = 6575235.1959799
, ymax = 6646756.8040201
)
, rotated = FALSE
, rotation = new(".Rotation"
, geotrans = numeric(0)
, transfun = function ()
NULL
)
, ncols = 176L
, nrows = 200L
, crs = new("CRS"
, projargs = NA_character_
)
, z = list()
, layernames = "Do not use the layernames slot (it is obsolete and will be removed)\nUse function 'names'"
)

Maybe I miss something , but I think you over complicated the problem. For me you have :
stack of raster : a list of raster : ss
a list of polygons of the same size as ss : polys
You need to apply extract for each pair(layer,poly) from (ss,polys)
sapply(1:nlayers(ss), function(i) {
m <- extract(ss[[i]],polys[i], method='bilinear', na.rm= T)[[1]]
d <- ifelse (!is.null(m) , sum(m)/length(m), NA)
d
})
Here an example of 2 legnths since you don't give a reproducible example :
## generate some data
library(raster)
cds1 <- rbind(c(-180,-20), c(-160,5), c(-60, 0), c(-160,-60), c(-180,-20))
cds2 <- rbind(c(80,0), c(100,60), c(120,0), c(120,-55), c(80,0))
## In your case you need something like SpatialPolygons(moose99kern)
polys <- SpatialPolygons(list(Polygons(list(Polygon(cds1)), 1),
Polygons(list(Polygon(cds2)), 2)))
r <- raster(ncol=36, nrow=18)
r[] <- 1:ncell(r)
r1 <- raster(ncol=36, nrow=18)
r1[] <- seq(-1,-2,length.out=ncell(r1))
ss <- stack(r,r1)
## density compute
sapply(1:nlayers(ss), function(i) {
## sum of values of the cells of a Raster ss[[i]] covered by the poly polys[i]
m <- extract(ss[[i]],polys[i], method='bilinear', na.rm= T)[[1]]
d <- ifelse (!is.null(m) , sum(m)/length(m), NA)
})
[1] 387.815789 -1.494714

When you are asking questions about R, always use simple reproducible examples, not your own data; unless perhaps what you want to do works for such an example, but not for your data, but then still show the example that works and the error message you are getting. You can typically start with the examples in the help files, as in below from ?extract
r <- raster(ncol=36, nrow=18)
r[] <- 1:ncell(r)
s <- stack(r, r*2)
cds1 <- rbind(c(-180,-20), c(-160,5), c(-60, 0), c(-160,-60), c(-180,-20))
cds2 <- rbind(c(80,0), c(100,60), c(120,0), c(120,-55), c(80,0))
polys <- SpatialPolygons(list(Polygons(list(Polygon(cds1)), 1),
Polygons(list(Polygon(cds2)), 2)))
v <- extract(s, polys, small=TRUE)
#cellnumbers for each polygon
sapply(v, NROW)
# mean for each polygon
sapply(v, function(x) apply(x, 2, mean, na.rm=T))
the functions in sapply need to be refined if some of your polgyons our outside of the raster (i.e. returning NULL, but the "small=TRUE" option should avoid problems with very small polygons inside the raster. Also note that there is no "method" argument when extracting with SpatialPolygon* objects.
Do not use a loop, unless to prevent memory problems if you have very many cells for each polygon.

Resources