[GRASS-user] i.atcorr with Sentinel2

[Stefan Blumentrath]

For completness:
https://trac.osgeo.org/grass/ticket/3469

-----Original Message-----
From: grass-user [mailto:grass-user-bounces at lists.osgeo.org] On Behalf Of Stefan Blumentrath
Sent: torsdag 21. desember 2017 08.18
To: Markus Metz <markus.metz.giswork at gmail.com>; Moritz Lennert <mlennert at club.worldonline.be>
Cc: GRASS user list <grass-user at lists.osgeo.org>
Subject: Re: [GRASS-user] i.atcorr with Sentinel2

Thanks Markus,

As a next try I updated my gcc and g++ to 5.4.1 (on Ubuntu 14.04) and I applied your compiler flags, Markus.
Unfortunately, I still get only NULL cells in i.atcorr output…

Will try to test more recent OS (Ubuntu 16) and maybe get a Fedora live system…

Cheers
Stefan

From: Markus Metz [mailto:markus.metz.giswork at gmail.com]
Sent: onsdag 20. desember 2017 22.52
To: Moritz Lennert <mlennert at club.worldonline.be>
Cc: Stefan Blumentrath <Stefan.Blumentrath at nina.no>; Sajid Pareeth <spareeth at gmail.com>; GRASS user list <grass-user at lists.osgeo.org>
Subject: Re: [GRASS-user] i.atcorr with Sentinel2

Adding to the confusion. I used

On Wed, Dec 20, 2017 at 5:41 PM, Moritz Lennert <mlennert at club.worldonline.be<mailto:mlennert at club.worldonline.be>> wrote:
>
> On 19/12/17 15:24, Stefan Blumentrath wrote:
>>
>>
>> In case someone would be willing to test my data (which I would be very grateful for), you can get it directly here:
>>
>> dem: http://www.filedropper.com/dem_1
>>
>> S2 band 8: 
>> http://www.filedropper.com/s2aoperprdmsil1cpdmc20160907t044118r008v20
>> 160905t10402220160905t104245b08
>>
>> 62 params: http://www.filedropper.com/p6s
with latest 7.4 and 7.5, command is

i.atcorr in=b08 range=1,28001 elevation=dem parameters=p6s.txt out=b08_atcorr

and I get reasonable results:

r.info<http://r.info> -r b08_atcorr_g74
min=0.003231863
max=255

r.info<http://r.info> -r b08_atcorr_g75
min=0.003231863
max=255
my OS is Fedora 27 with compiler gcc (GCC) 7.2.1 20170915 (Red Hat 7.2.1-2) relevant compiler flags are -g -Wall -Wextra -Wpedantic -Wshadow -Wno-sign-compare -fno-common -fexceptions -Werror=implicit-function-declaration -Wp,-D_FORTIFY_SOURCE=2 -O3 -fno-fast-math

-fno-fast-math might be important

running i.atcorr through valgrind, I get lots of warnings like

==14080== Conditional jump or move depends on uninitialised value(s)
==14080==    at 0x42B18D: os(float, float, float, float, float, float (&) [51][49], Gauss&, Altitude const&, GeomCond const&) (computations.cpp:831)
==14080==    by 0x42E9E3: atmref(float, float, float, float, float, OpticalAtmosProperties&, Gauss&, GeomCond const&, AerosolModel const&, Altitude const&) (computations.cpp:1408)
==14080==    by 0x42FCBC: discom(GeomCond const&, AtmosModel const&, AerosolModel const&, AerosolConcentration const&, Altitude const&, IWave const&) (computations.cpp:1654)
==14080==    by 0x40C032: init_6S(char*) (6s.cpp:100)
==14080==    by 0x405A3C: main (main.cpp:618)
...
later on

==14080== Conditional jump or move depends on uninitialised value(s)
==14080==    at 0x42DED8: iso(float, float, float, float, float, float (&) [3], Gauss&, Altitude const&) (computations.cpp:1262)
==14080==    by 0x42F011: scatra(float, float, float, float, float, OpticalAtmosProperties&, Gauss&, GeomCond const&, Altitude const&) (computations.cpp:1578)
==14080==    by 0x42FD08: discom(GeomCond const&, AtmosModel const&, AerosolModel const&, AerosolConcentration const&, Altitude const&, IWave const&) (computations.cpp:1659)
==14080==    by 0x406882: pre_compute_h(float) (6s.cpp:148)
==14080==    by 0x406372: process_raster (main.cpp:369)
==14080==    by 0x406372: main (main.cpp:632)
...
IMHO, these valgrind warnings need to be investigated.
Markus M

>>
>
> Just a very quick test, in 7.4 RC1, using
>
> i.atcorr in=b08 range=1,28001 elevation=dem 
> parameters=BROL/STEFAN/p6s.txt out=b08_atcorr
>
> gives
>
> r.info<http://r.info> -r b08_atcorr
> min=-nan
> max=-nan
>
> So I can confirm that there seems to be a problem.
> Don't have much time to do more testing.
>
> A bug report with example data would probably be warranted.
>
> Moritz
>
> _______________________________________________
> grass-user mailing list
> grass-user at lists.osgeo.org<mailto:grass-user at lists.osgeo.org>
> https://lists.osgeo.org/mailman/listinfo/grass-user
_______________________________________________
grass-user mailing list
grass-user at lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/grass-user