On Mon, 27 Sep 2021, Rik Ferreira wrote:
Dear Roger, thanks for the explanation.
The error between QGIS and sf drops a lot with sf_use_s2(FALSE) but it's
still bigger than 0. I believe it is due to different parameters passed
because GEOS and PROJ are QGIS dependencies.
No, the differences have nothing to do with GEOS, which only handles
planar geometries, or PROJ, which expresses the coordinate reference
system (here "EPSG:4674", a GEOGCRS in decimal degrees). The differences
are coming from difference ways of estimating area on the sphere (s2) or a
chosen ellipsoid (lwgeom, QGIS) all using different approaches. Try
projecting to UTM zone 23; then GEOS would be in play, because the
geometry is planar.
Hope this clarifies,
Roger
I will run more tests to find out what may be causing this issue but the
problem is solved for now.
Thank you!
Em seg., 27 de set. de 2021 às 12:32, Roger Bivand <roger.biv...@nhh.no>
escreveu:
On Mon, 27 Sep 2021, Rik Ferreira wrote:
Dear list members,
Why does sf::st_area returns a different area than the output of QGIS
$area
for the same geometry?
A reproducible example (state of Minas Gerais, Brazil):
library(sf)
mg <- geobr::read_state(31)
st_area(mg)
sf::st_area: 588358849314 [m^2]
QGIS $area: 586521037474,860 m²
The layer used in QGIS was exported with st_write() in GeoJSON format.
The error is: 1837811840 m².
st_is_longlat(mg)
[1] TRUE
sf_use_s2()
[1] TRUE
sf_use_s2(FALSE)
Spherical geometry (s2) switched off
(o <- st_area(mg))
586520883040 [m^2]
sf_use_s2(TRUE)
Spherical geometry (s2) switched on
(p <- st_area(mg))
588358849314 [m^2]
p - o
1837966275 [m^2]
See https://r-spatial.org/r/2020/06/17/s2.html
When sf_use_s2() is FALSE, see ?st_area ("ellipsoidal distances are
computed using st_geod_distance which uses function 'geod_inverse' from
GeographicLib (part of PROJ)"). You'd have to establish what code QGIS
uses to know why it and GeographicLib differ.
Hope this clarifies
Roger
Thank you for the attention.
Att.
--
Roger Bivand
Emeritus Professor
Department of Economics, Norwegian School of Economics,
Postboks 3490 Ytre Sandviken, 5045 Bergen, Norway.
e-mail: roger.biv...@nhh.no
https://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en
--
Roger Bivand
Emeritus Professor
Department of Economics, Norwegian School of Economics,
Postboks 3490 Ytre Sandviken, 5045 Bergen, Norway.
e-mail: roger.biv...@nhh.no
https://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en
_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
