Problem using 'masker' while plotting KDE - R

Question

The Problem

I am trying to plot the kernel density estimate for conflict data in Nigeria. I am using GISTools, rgdal, and sp loaded into R.

I want to generate the KDE, plot it out, and then mask the plot so that all excess space is clipped from the plot.

I am able to create the kernel density without problem. It's generating the mask that seems to be giving me the problem. I get the following error:

Error in `proj4string<-`(`*tmp*`, value = CRS(proj4string(input.poly))) : 
Geographical CRS given to non-conformant data: -997.36173977 1014.70658626
-995.75375669 1013.91618459
In addition: Warning message:
In RGEOSBinTopoFunc(spgeom1, spgeom2, byid, id, drop_lower_td, unaryUnion_if_byid_false,  :
spgeom1 and spgeom2 have different proj4 strings

Code

Here is what I have done:

I import the shapefile for the administrative districts 2

nks_gis2 <- readOGR(dsn = "Map/Data/nks_admin2.shp")
nks_gis1 <- readOGR(dsn = "Map/Data/nks_admin1.shp")
nks_gis0 <- readOGR(dsn = "Map/Data/nks_admin0.shp")

I then check their CRS

proj4string(nks_gis2)

[1] "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"

proj4string(nks_gis1)

[1] "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"

proj4string(nks_gis0)

[1] "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"

I import the incident data in as a spatial data frame

nks_longlat  <- cbind(nks$longitude, nks$latitude)
nks_points <- SpatialPointsDataFrame(nks_longlat, nks, proj4string = CRS("+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"))

I check the CRS for the points

proj4string(nks_points)

[1] "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"

I now generate the kernel density estimate

nks_dens <- kde.points(subset(nks_points, country %in% "Nigeria"), 
                   lims = subset(nks_gis2, NAME_0 %in% "Nigeria"))

And once again, I check the CRS

proj4string(nks_dens)

[1] "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"

I plot the density using level.plot

level.plot(nks_dens)

Now, here is where the problem occurs. I try to create a 'masker' so that it clips the plot to fit my Nigeria shapefile (I'll use the admin2 file):

nga_masker <- poly.outer(nks_dens, subset(nks_gis2, NAME_0 %in% "Nigeria"), extend = 1000)

I receive this error:

Error in proj4string<-(*tmp*, value = CRS(proj4string(input.poly))) :

Geographical CRS given to non-conformant data: -997.36173977 1014.70658626 -995.75375669 1013.91618459

In addition: Warning message:

In RGEOSBinTopoFunc(spgeom1, spgeom2, byid, id, drop_lower_td, unaryUnion_if_byid_false, : spgeom1 and spgeom2 have different proj4 strings

I have checked both my latitude and longitude, but do not see any long/lat that is even close to -997 or 1014.

Answer 1

You have:

nga_masker <- poly.outer(nks_dens, subset(nks_gis2, NAME_0 %in% "Nigeria"), 
  extend = 1000)

and that extend=1000 is expanding your polygon by 1000 degrees:

extend: A buffer used to extend the mask if it is required to be
        larger than ‘exo.object’

that explains the output coordinates being around the 1000 mark, and not being valid lat-long degrees.