Error in proj4string ... Geographical CRS given to non-conformant data

[Geoff99]

Hi there

First --- thanks for a really neat package.

I encountered a problem while running a variant of an example from http://vietletruc.com/tag/cartogram/. (on a Windows machine)

quick.carto fails with the following message.

d> world.carto <- quick.carto(world, world@data$Population, blur = 0)
Error in `proj4string<-`(`*tmp*`, value = <S4 object of class "CRS">) : 
  Geographical CRS given to non-conformant data: -189.000000 -192.188202  185.811798
Called from: stop(mess)

If I look at the world object which is a SpatialPolygonsDataFrame I find that it uses (as I expected) a longlat projection. And both the latitude and longitude values are all "legal".

d> world@proj4string
CRS arguments:
 +proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0 
d> range(coordinates(world)[, 1])
[1] -178.1269  179.2159
d> range(coordinates(world)[, 2])
[1] -80.50275  78.60600
d> 

However if I overwrite the proj4string, and then try quick.carto again, everything is fine.

d> world@proj4string <- CRS(as.character(NA))
d> world.carto <- quick.carto(world, world@data$Population, blur = 0)
d> 

I suspect (but have no evidence) that somewhere in the internals of quick.carto a new expanded copy of the SpatialPolygonsDataFrame is made including the surrounding "sea" as described in the Gastner Newman paper. And that the copy simply uses the same proj4string / CRS as the map being morphed. And that when the longlat values are expanded, they exceed the legal values for longitudes and latitudes. Or perhaps simply that at some stage during the morphing process the (warped) long lat values for the cartogram exceed the legal values for longitude and latitude. Hence this error message, which I think comes from a validity check during the creation of a (copy of) SpatialPolygonsDataFrame, when the CRS is longlat .

If I am correct, wiping out the proj4string / CRS in the way I have above is actually the correct thing to do I think --- since after morphing it would be fair to say that the coordinate reference system of the warped map is no longer any type of longitude latitude projection (nor indeed any other projection likely to be known to standard geography and supported in the CRS / proj4 universe).

Not sure what to do, but maybe this issue will help others who may experience a similar problem.

Once again, really like the package, thanks.

Geoff

[ghost]

Hello!

Thank you very much for posting this issue! I'm having the same problem, but unfortunately I'm not having the same results as you.

I'm having the same error:

> cont.carto <- quick.carto(cont, cont$data, blur = 0)
  Error in `proj4string<-`(`*tmp*`, value = <S4 object of class "CRS">) : 
     Geographical CRS given to non-conformant data: -189.000000 -192.177435  185.822565

After applying your solution:

> cont@proj4string <- CRS(as.character(NA))
> cont.carto <- quick.carto(cont, cont$data, blur = 0)
     Error in .local(obj, ...) : NA values in coordinates

Any ideas what can this be? Thank you very much!

[Geoff99]

Hi.

It is quite a while since I last looked at this, so my memory of the details is really vague.

Have you tried looking at the actual values in the coordinates of cont (which I presume is your SpatialPolygonsDataFrame) using coordinates(cont)[,1] and coordinates(cont)[,2] ? I think the error message is saying that it is encountering NA values in there -- but I don't know whether that I happening because cont itself has NA values, or because at some stage one of the intermediate calculations that quick.carto does results in a NA.

Other than that, I don't think I have much advice I can offer. Hope that helps you track the problem down.

Good luck

Geoff

[ghost]

Hi Geoff,

Thank you very much for your reply! I understand that it has been a while, so don't worry. In the end, I had some problems with the polygons of cont so after I made it a valid SpatialPolygonsDataFrame, quick.carto worked :)

Cheers and thanks,

Vanessa

[Geoff99]

Hi Vanessa,

Glad you managed to get it running. Cartograms are pretty neat when you get them working J

Cheers

Geoff

From: Vanessa Pena [mailto:notifications@github.com] Sent: 30 March, 2017 11:18 AM To: chrisbrunsdon/getcartr Cc: Geoff99; Author Subject: Re: [chrisbrunsdon/getcartr] Error in proj4string ... Geographical CRS given to non-conformant data (#1)

Hi Geoff,

Thank you very much for your reply! I understand that it has been a while, so don't worry. In the end, I had some problems with the polygons of cont so after I made it a valid SpatialPolygonsDataFrame, quick.carto worked :)

Cheers and thanks,

Vanessa

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/chrisbrunsdon/getcartr/issues/1#issuecomment-290264705 , or mute the thread https://github.com/notifications/unsubscribe-auth/AH9DBt1bSgrkm3S5wNQuFc99GbMSGKwhks5rqvSggaJpZM4FjJn1 . https://github.com/notifications/beacon/AH9DBt64Y9prmC_J9aoNPZZP6Lt2NGS8ks5rqvSggaJpZM4FjJn1.gif