SOmap_auto to handle target=NULL

[raymondben]

Reminder for the thing we discussed the other day: SOmap_auto(raster_layer, target = NULL) should inherit target from raster_layer, not try and guess its own.

[mdsumner]

that's done, should be

  library(SOmap)
  x <- raster::raster(system.file("nc/reduced.nc", package = "stars"), varname = "sst")
  a <- SOmap_auto(x, target = NULL)
  a

but I'm getting fig errors on plot ...

[mdsumner]

Wowsers, if I don't "rotate()" I get freaky fig errors ...

library(SOmap)
#> Loading required package: raster
#> Loading required package: sp
x <- raster::rotate(raster::raster(system.file("nc/reduced.nc", package = "stars"), varname = "sst"))
#> Loading required namespace: ncdf4
a <- SOmap_auto(x, target = NULL)
#> Warning in CPL_proj_direct(as.character(c(from[1], to[1])),
#> as.matrix(pts)): one or more projected point(s) not finite
#> dist is assumed to be in decimal degrees (arc_degrees).
#> although coordinates are longitude/latitude, st_intersection assumes that they are planar
a

^{Created on 2019-06-04 by the reprex package (v0.3.0)}

[Maschette]

Was I part of this discussion, I forget. Sounds fun!

Is it getting confused by being over the northern hemisphere?

[mdsumner]

More examples, found a bug

library(SOmap)
#> Loading required package: raster
#> Loading required package: sp
a <- SOmap_auto(ice, target = NULL)
a

^{Created on 2019-06-04 by the reprex package (v0.3.0)}

[Maschette]

What is the reason for setting target = NULL? surely that tells it there is no target?

[mdsumner]

It tells it to use the input raster grid as-is even if that grid is in longlat.

[Maschette]

Ok so it is not actually target=NULL but target=projection(x) internally.
That makes sense but I think the default should still be our standard projection.

[raymondben]

If x as a raster object and target is left as its default (stere) then SOmap_auto(x) will use the extent of x but NOT its projection - i.e. following your "default should still be our standard projection". If you set target = NULL it is saying to use the projection of x (and that is equivalent to specifying target = projection(x)).

[Maschette]

Weird output for the new SOmap_auto with class of spatialpointsdataframe

library(rgdal)
#> Loading required package: sp
#> rgdal: version: 1.4-4, (SVN revision 833)
#>  Geospatial Data Abstraction Library extensions to R successfully loaded
#>  Loaded GDAL runtime: GDAL 2.2.3, released 2017/11/20
#>  Path to GDAL shared files: C:/Users/dale_mas/Documents/R/R-3.6.0/library/rgdal/gdal
#>  GDAL binary built with GEOS: TRUE 
#>  Loaded PROJ.4 runtime: Rel. 4.9.3, 15 August 2016, [PJ_VERSION: 493]
#>  Path to PROJ.4 shared files: C:/Users/dale_mas/Documents/R/R-3.6.0/library/rgdal/proj
#>  Linking to sp version: 1.3-1
library(SOmap)
#> Loading required package: raster
#> Registered S3 methods overwritten by 'ggplot2':
#>   method         from 
#>   [.quosures     rlang
#>   c.quosures     rlang
#>   print.quosures rlang
brokew<-readOGR("C:\\Users\\dale_mas\\Documents\\Projects\\Mapping\\data\\Krill_surveys_EastAntarctica\\Krill_surveys_EastAntarctica\\BROKEWest_transects.shp")
#> OGR data source with driver: ESRI Shapefile 
#> Source: "C:\Users\dale_mas\Documents\Projects\Mapping\data\Krill_surveys_EastAntarctica\Krill_surveys_EastAntarctica\BROKEWest_transects.shp", layer: "BROKEWest_transects"
#> with 3890 features
#> It has 6 fields
#> Integer64 fields read as strings:  transect segment
plot(brokew)

SOmap_auto(brokew)
#> Error in st_cast_sfc_default(x): list item(s) not of class sfg

^{Created on 2019-06-04 by the reprex package (v0.3.0)}

[mdsumner]

I can't reproduce that, can you print a summary? (or send me the data)

library(raster)
print(brokew)

[mdsumner]

Also I'm sympathetic to the issue of SOmap being about stereographic. Maybe it's time to take the general stuff out.

The issue atm is

• we default to projecting input data, with target = stereographic (1)
• we determine the extent and orientation from your data (2)
• we provide custom choice of projection type (target), optionally centre, and optionally full specification (3)
• if the input data is projected, we use its extent and projection, ignoring target if it is the same type (4)
• or, if not the same type we work out a new extent in the input target (5)
• if the input data is not projected, we go to (2), unless target = NULL (6)

It's helpful to write that all down, I want to flow-diagram it.

I feel like customizing target in (5) is too much, maybe we shouldn't allow that and just set target = "stere" and rather than NULL only allow target = "source".

But, this is functional enough that we should just push through the gg stuff as well, then review.

(I just love being able to do anything I want with any input, but it is getting complicated again. If we can seal this off sensibly I'll pull out the general stuff and make "panmap" or something)

[raymondben]

The "fig errors on plot" (second comment, above) are fixed in ceffb43. But there is a lot more discussion here, what do we want to do with it? @mdsumner @Maschette

[Maschette]

Kill it with fire.
The actual issue is fixed. It does highlight though a need for a check on the data, in what was the first example the raster is 0-360, everything in SOmap is -180 to 180 so it plots half the continents but not the other. We could add a warning telling people what the issue is.