accept WKT "POLYGON FULL", for full polygon (entire Earth), if coordinates are geodetic

[edzer]

The full polygon is not part of the simple feature standard, but is needed when coordinates are geodetic. The "hidden" implementation used by s2geometry is POLYGON((0 -90,0 -90))

st_as_sfc("POLYGON((0 -90,0 -90))", crs = 4326) |> st_area()
# 5.100661e+14 [m^2]

but that is rather ugly from the user side. A more reasonable approach would be

library(sf)
# Linking to GEOS 3.12.1, GDAL 3.8.4, PROJ 9.4.0; sf_use_s2() is TRUE
x = st_as_sfc(c("POINT(0 1)", "POLYGON FULL"), crs = 'OGC:CRS84')
x
# Geometry set for 2 features 
# Geometry type: GEOMETRY
# Dimension:     XY
# Bounding box:  xmin: 0 ymin: -90 xmax: 0 ymax: 1
# Geodetic CRS:  WGS 84 (CRS84)
# POINT (0 1)
# POLYGON FULL
st_area(x)
# Units: [m^2]
# [1] 0.000000e+00 5.100661e+14

I suppose it should only be accepted when it is clear that the CRS is unprojected / coordinates are geodetic.

[edzer]

T.b.d. :

• st_make_valid()
• st_bbox()
• st_is_empty()
• st_is_simple()
• st_dimension()

[edzer]

As discussed with @evetion

[evetion]

Will there be a WKB equivalent? And if not, how does one store this WKT in files etc?

[edzer]

Good questions, this is less obvious! The "cheap" option is to use the WKB of POLYGON((0 -90,0 -90)), which is, in hex

> st_as_sfc("POLYGON((0 -90,0 -90))", crs = 'OGC:CRS84') |> st_as_binary(hex = TRUE)
[1] "01030000000100000002000000000000000000000000000000008056c0000000000000000000000000008056c0"

I'm afraid other options would require modification of the big upstream libraries.

[edzer]

I suppose it should only be accepted when it is clear that the CRS is unprojected / coordinates are geodetic.

Probably more useful would be that POLYGON FULL passes also when no CRS is specified - for those working by default in geodetic coordinates.

[edzer]

Here's a test run (in sf/tests/full.R):

suppressPackageStartupMessages(library(sf))

# create full polygon:
(f = st_as_sfc("POLYGON FULL"))
# Geometry set for 1 feature 
# Geometry type: POLYGON
# Dimension:     XY
# Bounding box:  xmin: -180 ymin: -90 xmax: 180 ymax: 90
# CRS:           NA
# POLYGON FULL
g = st_sfc(st_polygon(list(matrix(c(0,-90,0,-90), 2, byrow = TRUE))))
identical(f, g)
# [1] TRUE
old = sf_use_s2(FALSE)
# Spherical geometry (s2) switched off
try(st_as_sfc("POLYGON FULL")) # errors
# OGR: Corrupt data
# Error : OGR error
sf_use_s2(old)
# Spherical geometry (s2) switched on
(f = st_as_sfc(c("POLYGON FULL", "POLYGON((0 0,1 0,1 1,0 1,0 0))")))
# Geometry set for 2 features 
# Geometry type: POLYGON
# Dimension:     XY
# Bounding box:  xmin: -180 ymin: -90 xmax: 180 ymax: 90
# CRS:           NA
# POLYGON FULL
# POLYGON ((0 0, 1 0, 1 1, 0 1, 0 0))
st_is_full(f)
# [1]  TRUE FALSE
st_bbox(f[1])
# xmin ymin xmax ymax 
# -180  -90  180   90 
st_bbox(f[2])
# xmin ymin xmax ymax 
#    0    0    1    1 
st_is_valid(f) # full polygon NA: right, we don't know the CRS
# [1]   NA TRUE
st_crs(f) = 'OGC:CRS84' # geodetic:
st_is_valid(f)
# [1] TRUE TRUE
st_crs(f) = NA
try(st_make_valid(f))
# Error in (function (msg)  : 
#   IllegalArgumentException: Invalid number of points in LinearRing found 2 - must be 0 or >= 3
st_crs(f) = 'OGC:CRS84' # geodetic:
st_make_valid(f)
# Geometry set for 2 features 
# Geometry type: POLYGON
# Dimension:     XY
# Bounding box:  xmin: -180 ymin: -90 xmax: 180 ymax: 90
# Geodetic CRS:  WGS 84 (CRS84)
# POLYGON FULL
# POLYGON ((0 0, 1 0, 1 1, 0 1, 0 0))
# mixed geometries:
(f = st_as_sfc(c("POLYGON FULL", "POLYGON((0 0,1 0,1 1,0 1,0 0))", "POINT(0 1)"), crs = 'OGC:CRS84'))
# Geometry set for 3 features 
# Geometry type: GEOMETRY
# Dimension:     XY
# Bounding box:  xmin: -180 ymin: -90 xmax: 180 ymax: 90
# Geodetic CRS:  WGS 84 (CRS84)
# POLYGON FULL
# POLYGON ((0 0, 1 0, 1 1, 0 1, 0 0))
# POINT (0 1)
st_bbox(f[1])
# xmin ymin xmax ymax 
# -180  -90  180   90 
st_bbox(f[3])
# xmin ymin xmax ymax 
#    0    1    0    1 
st_is_valid(f)
# [1] TRUE TRUE TRUE
st_make_valid(f)
# Geometry set for 3 features 
# Geometry type: GEOMETRY
# Dimension:     XY
# Bounding box:  xmin: -180 ymin: -90 xmax: 180 ymax: 90
# Geodetic CRS:  WGS 84 (CRS84)
# POLYGON FULL
# POLYGON ((0 0, 1 0, 1 1, 0 1, 0 0))
# POINT (0 1)
st_make_valid(f[2:3])
# Geometry set for 2 features 
# Geometry type: GEOMETRY
# Dimension:     XY
# Bounding box:  xmin: 0 ymin: 0 xmax: 1 ymax: 1
# Geodetic CRS:  WGS 84 (CRS84)
# POLYGON ((0 0, 1 0, 1 1, 0 1, 0 0))
# POINT (0 1)

[edzer]

Do we need a MULTIPOLYGON FULL?

[evetion]

Do we need a MULTIPOLYGON FULL?

That doesn't make sense conceptually, as there would be multiple overlapping full polygons? Or is it a multi with a single polygon?

[edzer]

Or is it a multi with a single polygon?

yes.

The problem is that SFA already has all types empty (7 empty geometry types) that all point to the same concept ("no points for which...")