[GRASS-dev] i.sentinel improvements

lucadelu · December 17, 2018, 9:38am

Hi all,

I would like to do some improvements in i.sentinel
Since I'm not the main author I would like to get approval before to
start to work on it.
- i.sentinel.download add capability to filter out scenes with small
percent (user could decide the percentage to keep) of area matching
the area of interest
- i.sentinel.import add capability to cut the scene on the footprint
to simplify scenes patch
- i.sentinel.preproc add capability to compute topographic correction
using the following prcedure

###
eval `r.sunmask -sg ele=DTM year=${dat:0:4} month=$(echo $dat:4:2 |
sed 's/^0*//') day=${dat:6:2} hour=${dat:9:2} minute=${dat:11:2}
second=${dat:13:2} timezone=0`
z=`echo "90 - $sunangleabovehorizon" | bc -l`
echo "${inpmap} $z $sunazimuth"
i.topo.corr -i basemap=DTM zenit=$z azimuth=$sunazimuth output=dtm_illu

for m in `g.list type=raster pattern=${inpmap}*_cor`; do g.region
raster=$m; r.mapcalc expr="${m}_d=double(${m})"; done

i.topo.corr basemap=dtm_illu input=`g.list type=raster sep=comma
pattern=${inpmap}*_cor_d` output=tcor zenith=$z method=c-factor

g.remove -f type=raster pattern=${inpmap}*_cor_d
###

let me know.

--
ciao
Luca

www.lucadelu.org

martinl · December 17, 2018, 1:18pm

Hi,

po 17. 12. 2018 v 10:39 odesílatel Luca Delucchi <lucadeluge@gmail.com> napsal:

- i.sentinel.download add capability to filter out scenes with small
percent (user could decide the percentage to keep) of area matching
the area of interest

sounds great

- i.sentinel.import add capability to cut the scene on the footprint
to simplify scenes patch

great again

- i.sentinel.preproc add capability to compute topographic correction
using the following prcedure

unrelated, probably this module could be renamed to
i.sentinel.import.preproc or similar since it's importing data. Any
idea/opinion?

Ma

--
Martin Landa
http://geo.fsv.cvut.cz/gwiki/Landa
http://gismentors.cz/mentors/landa

veroandreo · December 17, 2018, 1:57pm

Hi Lu,

El lun., 17 dic. 2018 a las 10:39, Luca Delucchi (<lucadeluge@gmail.com>) escribió:

Hi all,

I would like to do some improvements in i.sentinel
Since I’m not the main author I would like to get approval before to
start to work on it.

i.sentinel.download add capability to filter out scenes with small
percent (user could decide the percentage to keep) of area matching
the area of interest

i.sentinel.import add capability to cut the scene on the footprint
to simplify scenes patch

i.sentinel.preproc add capability to compute topographic correction

+1! (no idea about differences among topographic correction algorithms, though… is c-factor the best one?)

What will the module do with L2A data? Only topographic correction?

unrelated, probably this module could be renamed to
i.sentinel.import.preproc or similar since it’s importing data. Any
idea/opinion?

Yeah, that would be an option… What about keeping i.sentinel.import for the import part and make i.sentinel.preproc only about pre-processing, i.e., atcorr and topcorr?

best,

Vero

Roberta_Fagandini · December 19, 2018, 9:04am

Hi Luca,

Il giorno lun 17 dic 2018 alle ore 10:39 Luca Delucchi <lucadeluge@gmail.com> ha scritto:

Hi all,

I would like to do some improvements in i.sentinel
Since I’m not the main author I would like to get approval before to
start to work on it.

i.sentinel.download add capability to filter out scenes with small
percent (user could decide the percentage to keep) of area matching
the area of interest

i.sentinel.import add capability to cut the scene on the footprint
to simplify scenes patch

i.sentinel.preproc add capability to compute topographic correction
using the following prcedure

Of course, it’s ok for me!

eval r.sunmask -sg ele=DTM year=${dat:0:4} month=$(echo $dat:4:2 | sed 's/^0*//') day=${dat:6:2} hour=${dat:9:2} minute=${dat:11:2} second=${dat:13:2} timezone=0
z=echo "90 - $sunangleabovehorizon" | bc -l
echo “${inpmap} $z $sunazimuth”
i.topo.corr -i basemap=DTM zenit=$z azimuth=$sunazimuth output=dtm_illu

for m in g.list type=raster pattern=${inpmap}*_cor; do g.region
raster=$m; r.mapcalc expr=“${m}_d=double(${m})”; done

i.topo.corr basemap=dtm_illu input=g.list type=raster sep=comma pattern=${inpmap}*_cor_d output=tcor zenith=$z method=c-factor

g.remove -f type=raster pattern=${inpmap}*_cor_d

I don’t have much experience with topographic correction but I’m not sure that c-factor is the best. When I tested it, I got anomalous results.

However, I would add the topographic correction as an option (e.g -flag “also perform topo cor”)

Yeah, that would be an option… What about keeping i.sentinel.import for the import part and make i.sentinel.preproc only about pre-processing, i.e., atcorr and topcorr?

+1.

Robi

neteler · December 19, 2018, 11:36am

On Wed, Dec 19, 2018 at 10:04 AM Roberta Fagandini
<robifagandini@gmail.com> wrote:

Il giorno lun 17 dic 2018 alle ore 10:39 Luca Delucchi <lucadeluge@gmail.com> ha scritto:

...

- i.sentinel.preproc add capability to compute topographic correction
using the following prcedure

...

I don't have much experience with topographic correction but I'm not sure that c-factor is the best. When I tested it, I got anomalous results.

However, I would add the topographic correction as an option (e.g -flag "also perform topo cor")

Since there are different method in i.topo.corr, perhaps instead of a
flag use a parameter? Like:
topo_method=c-factor|minnaert|...

Markus

lucadelu · February 12, 2019, 10:34am

On Mon, 17 Dec 2018 at 10:38, Luca Delucchi <lucadeluge@gmail.com> wrote:

Hi all,

Hi devs,

- i.sentinel.preproc add capability to compute topographic correction
using the following prcedure

done in r74080, please test it

--
ciao
Luca

www.lucadelu.org

Roberta_Fagandini · February 13, 2019, 2:02pm

Il giorno mar 12 feb 2019 alle ore 11:35 Luca Delucchi <lucadeluge@gmail.com> ha scritto:

On Mon, 17 Dec 2018 at 10:38, Luca Delucchi <lucadeluge@gmail.com> wrote:

Hi all,

Hi devs,

Hi Luca,

i.sentinel.preproc add capability to compute topographic correction
using the following prcedure

done in r74080, please test it

Unfortunately I have no time to test it in the next days but I will as soon as possible.
I took a look at the interface and I saw a new text form for the output suffix name…if the -c flag is checked, does the module create two different output (13 atmospherically corrected bands and 13 topographically corrected bands)?
maybe only one output (i.e. 13 bands to which both corrections are applied) would be better.

–
ciao
Luca

ciao
Roberta

www.lucadelu.org

grass-dev mailing list
grass-dev@lists.osgeo.org
https://lists.osgeo.org/mailman/listinfo/grass-dev

lucadelu · February 13, 2019, 2:14pm

On Wed, 13 Feb 2019 at 15:02, Roberta Fagandini <robifagandini@gmail.com> wrote:

I took a look at the interface and I saw a new text form for the output suffix name..if the -c flag is checked, does the module create two different output (13 atmospherically corrected bands and 13 topographically corrected bands)?

right now yes

maybe only one output (i.e. 13 bands to which both corrections are applied) would be better.

could make sense, probably a new flag to keep/remove atmospherically
corrected bands could be useful

ciao
Roberta

--
ciao
Luca

www.lucadelu.org