“Non Finite Transformation Detected” in spTransfor

Question

I am trying to convert geographic coordinates (degrees) into UTM coordinates (meters) and keep getting an error message that a "Non finite transformation detected." Do you know how I can fix this? Here is the code I used:

> GPS.Points <- Gomer.Data[, c('Longitude', 'Latitude')]
> head(GPS.Points)

  Longitude  Latitude
1  23.85474 -19.52211
2  23.85531 -19.52243
3  23.85534 -19.52257
4  23.85580 -19.52346
5  23.85551 -19.52380
6  23.85513 -19.52360

> GPS.Points.Spatial.Data <- SpatialPoints(GPS.Points, 
                                         proj4string=CRS("+proj=longlat +ellps=WGS84"))
> GPS.Points.Spatial.Data[1]

SpatialPoints:
     Longitude  Latitude
[1,]  23.85474 -19.52211
Coordinate Reference System (CRS) arguments: +proj=longlat +ellps=WGS84 

> class(GPS.Points.Spatial.Data)

[1] "SpatialPoints"
attr(,"package")
[1] "sp"

> GPS.Points.UTM.Spatial.Data <- spTransform(GPS.Points.Spatial.Data,
                                           CRS("+proj=utm +south +zone=34 +ellps=WGS84"))

non finite transformation detected:
     Longitude Latitude  
Error in spTransform(GPS.Points.Spatial.Data, CRS("+proj=utm +south +zone=34 +ellps=WGS84")) : 
  failure in points 
In addition: Warning message:
In spTransform(GPS.Points.Spatial.Data, CRS("+proj=utm +south +zone=34 +ellps=WGS84")) :
  3 projected point(s) not finite

Answer 1

I would check the data you are trying to convert. I was not able to access the data you have in your example so I just used the first 3 coordinate points you provided to try to replicate the error and did not get an error. I also checked to see if the error could be caused by having the UTM zone specified not include all points provided by changing the zone number and north/south parameter and still everything worked. I would perhaps create a loop that iterates through the data you want to convert in chunks to see where the problem lies...

library(rgdal)

GPS.Points=data.frame(Longitude=c(23.85474, 23.85531, 23.85534))
GPS.Points=cbind(GPS.Points,Latitude=c(-19.52211, -19.52243, -19.52257))

GPS.Points.Spatial.Data <- SpatialPoints(GPS.Points, 
proj4string=CRS("+proj=longlat     +ellps=WGS84"))
GPS.Points.Spatial.Data[1]

class(GPS.Points.Spatial.Data)

GPS.Points.UTM.Spatial.Data <- spTransform(GPS.Points.Spatial.Data,
                          CRS("+proj=utm +south +zone=34 +ellps=WGS84"))

Since you asked, here is a code to iterate through parts of the data. If you get an error you will at least know where in your data the problem came from:

library(rgdal)
GPS.Points=data.frame(Longitude=c(23.85474, 23.85531, 23.85534, 23.85474, 
23.85531, 23.85534, 23.85474, 23.85531, 23.85534))
GPS.Points=cbind(GPS.Points,Latitude=c(-19.52211, -19.52243, -19.52257, -19.52211, 
-19.52243, -19.52257, -19.52211, -19.52243, -19.52257))

n_chunks=3 #number of pieces you will break you data into
n.points=dim(GPS.Points)[1]
breaks=seq(1,n.points, by=round(n.points/n_chunks))
breaks=c(breaks, n.points) #make sure to include last points as well

i=1
for (i in 1:(length(breaks)-1)){
  cat('\n','converting points', breaks[i], "to", breaks[i+1])  
  temp.GPS.Points=GPS.Points[breaks[i]:breaks[i+1],]
  temp.GPS.Points.Spatial.Data <- SpatialPoints(temp.GPS.Points, 
proj4string=CRS("+proj=longlat +ellps=WGS84"))
  temp.GPS.Points.UTM.Spatial.Data <- spTransform(temp.GPS.Points.Spatial.Data,
                                             CRS("+proj=utm +south +zone=34 
+ellps=WGS84"))
}