R 中的坐标值

Question

I'm new to R and trying to learn how it's done for some species distribution modeling. I'm trying to create a mask of Italy using the wrld_simpl map included with the maptools package. When I run the code, I get "Error in.local(obj, ...): NA values in coordinates" when I'm trying to create spxy. I suspect that this issue might be rooted in the rasterization step but I'm not sure... What am I doing wrong?

Apologies for the code dump

# Create library
library(raster)
library(dismo)
library(sf)
library(maptools)
library(rgdal)
library(sp)
library(rgeos)

# Call the cleaned nivale csv
nivale <- read.table('C:/Users/David/Documents/nival_1980_GeoDat.csv', header=T, sep = ',')
nivale <- nivale[,2:3]

# Create a simple map of Italy, plot nivale data points
data(wrld_simpl)
plot(wrld_simpl, xlim=c(0,20), ylim=c(40,50), axes=TRUE, col="light yellow")
box()
points(nivale$lon, nivale$lat, col='orange', pch=20, cex=0.75)
points(nivale$lon, nivale$lat, col='black', cex=0.75)

# set CRS of nivale equal to wrld_simpl
coordinates(nivale) <- ~lon+lat
crs(nivale) <- crs(wrld_simpl)
projection(nivale) <- CRS('+proj=longlat +datum=WGS84')
class(nivale)
class(wrld_simpl)

# Sampling Bias Assessment
r <- raster(nivale)
res(r) <- 1
r <- extend(r, extent(r)+1)
nisel <- gridSample(nivale, r, n=1)
p <- rasterToPolygons(r)
plot(p, border='blue')
points(nivale)
points(nisel, cex=1, col='red', pch='x')

# Pseudo-Absences
# Create shapefile from wrld_simpl
italy <- wrld_simpl[is.element(wrld_simpl$NAME, 'Italy'),]
set.seed(1963)
crsi = crs('+proj=longlat +datum=WGS84')
exti = extent(italy)
# Create template for rasterization
rst_temp <- raster(ncols = 1000, nrows = 1000, 
                   crs = crsi, 
                   ext = exti)
# Rasterize italy
rst_italy <- rasterize(italy, rst_temp)
# Random point generation
rand_point <- randomPoints(rst_italy, 250)

#Pseudo-Absence Points
x <- circles(nivale, d=50000, lonlat=TRUE)
pol <- polygons(x)
samp1 <- spsample(pol, 250, type='random', iter=25)
cells <- cellFromXY(rst_italy, samp1)
xy <- xyFromCell(rst_italy, cells)
plot(pol, axes=TRUE)
points(xy, cex=0.75, pch=20, col='blue')
#
# Error - NA values in coordinates
spxy <- SpatialPoints(xy, proj4string=CRS('+proj=longlat +datum=WGS84'))
o <- over(spxy, geometry(x))
xyInside <- xy[!is.na(o), ]
v <- extract(mask, x@polygons, cellnumbers=T)
v <- do.call(rbind, v)
v <- unique(v[,1])
head(v)
m <- italy
m[] <- NA
m[v] <- 1
plot(m, ext=extent(x@polygons)+1)
plot(x@polygons, add=T)

Answer 1

I don't have your nivale dataset, so I am not 100% sure, but I think the error arises after you are sampling points within polygons ( spsample(pol, 250, type='random', iter=25) ) created using circle() and not because of the rasterization, which looks fine to me. My guess is that, after this, you sample random points in these polygons, which may fall outside the raster of Italy ( samp1 <- spsample(pol, 250, type='random', iter=25) ). Possibly, you can just remove such points by excluding NAs values in xy :

xy <- xy[.is,na(xy[. 1]) &,is.na(xy[, 2])]

Try this and see if it works. If you need a certain number of points (eg xy needs to have 250 lines), you can put a while loop until you have what you want; this may take (a lot of) time, though.

/Emilio