[GRASS-dev] [GRASS GIS] #3860: GRASS GIS producing different slope than GDAL
GRASS GIS
trac at osgeo.org
Mon Jun 10 00:52:16 PDT 2019
#3860: GRASS GIS producing different slope than GDAL
------------------------+-------------------------
Reporter: mazingaro | Owner: grass-dev@…
Type: defect | Status: new
Priority: major | Milestone: 7.6.2
Component: Raster | Version: unspecified
Resolution: | Keywords: slope
CPU: x86-64 | Platform: Linux
------------------------+-------------------------
Comment (by mlennert):
Replying to [comment:2 mazingaro]:
> Replying to [comment:1 mlennert]:
> > What does GRASS GIS give you when you work at the resolution of the
data, i.e. 1m, instead of 0.5m ?
> The result is the same (with the -a flag)
Weird. Are you sure you are working on the same region extent ?
I just tried with the NC demo data elevation map, setting the region with
g.region rast=elevation in GRASS. Here are the univar stats of the
difference map between the two:
{{{
> r.univar diff
100%
total null and non-null cells: 8100000
total null cells: 22784
Of the non-null cells:
----------------------
n: 8077216
minimum: -9.67979e-05
maximum: 9.72748e-05
range: 0.000194073
mean: -1.50386e-09
mean of absolute values: 1.30826e-05
standard deviation: 1.69973e-05
variance: 2.88908e-10
variation coefficient: -1.13025e+06 %
sum: -0.0121469677105779
}}}
IOW: very small (IMHO negligeable) differences.
However, when I calculate the slope in GRASS with the same region extent,
but a resolution set to 5m, I get much more significant differences (diff
and univar stats calculated at 5m resolution):
{{{
> r.univar diff_restest
100%
total null and non-null cells: 2025000
total null cells: 5696
Of the non-null cells:
----------------------
n: 2019304
minimum: 0
maximum: 19.4712
range: 19.4712
mean: 3.74369
mean of absolute values: 3.74369
standard deviation: 2.68803
variance: 7.22549
variation coefficient: 71.8015 %
sum: 7559653.93641658
}}}
--
Ticket URL: <https://trac.osgeo.org/grass/ticket/3860#comment:3>
GRASS GIS <https://grass.osgeo.org>
More information about the grass-dev
mailing list