[GRASS-dev] New GRASS addon i.sentinel.mask is now available (GSoC 2018 project)

[Ivan Barka]

Hi Roberta,

thank you for your effort. I tried i.sentinel.mask just now with one
Sentinel-2 granule. I have considerably better results with Level 2A
Sentinel product, topographically and atmospherically corrected from Level
1C by sen2cor with default settings.

Regarding strange results of i.topo.corr, I also had a problem few months
ago. In the first pass, if the resolution of DEM raster differs from
current region settings, you can create strange illumination model. The
solution was to set region from the DEM (e.g. with 25 m resolution in case
of EU-DEM), then to create illumination model with "i.topo.corr -i ..."),
restore region settings (e.g. with 30 m for Landsat) and finally to apply
topo correction in the second pass.
For some unknown reason, i.topo.corr did not change region settings
automatically, despite you can find "The illumination model (cos_i) with
flag -i uses the actual region as limits and the resolution of the
elevation map." in its manual (Notes section).

Best regards,

Ivan


2018-06-14 13:08 GMT+02:00 Stefan Blumentrath <Stefan.Blumentrath at nina.no>:

> i.topo.corr performed quite well for us in the Norwegian mountains with
> minnaert method…
>
> What options did you try in i.topo.corr?
>
>
>
> Cheers
>
> Stefan
>
>
>
> *From:* Roberta Fagandini <robifagandini at gmail.com>
> *Sent:* torsdag 14. juni 2018 13:01
> *To:* Stefan Blumentrath <Stefan.Blumentrath at nina.no>
> *Cc:* Moritz Lennert <mlennert at club.worldonline.be>; GRASS developers
> list <grass-dev at lists.osgeo.org>
>
> *Subject:* Re: [GRASS-dev] New GRASS addon i.sentinel.mask is now
> available (GSoC 2018 project)
>
>
>
> Hi Stefan!
>
> Certainly applying the topographic correction could be very useful to
> avoid misclassification.
>
> I did several tests with i.topo.corr but I always got strange results
> especially in areas with steeper terrain.
>
> I also tested the topographic correction tool implemented in arcsi [0]
> but again the results were not satisfactory at all. This is why I have not
> further investigated this issue within my PhD research.
>
>
>
> In order to remove misclassifications, e.g. due to topographic shadows, I
> implemented the shadow mask cleaning procedure. It intersects the rough
> shadow mask with the cloud mask that is shifted according to the sun
> position. In this way, all the areas identified as shadows without a
> corresponding cloud are removed from the final shadow mask.
>
>
>
> I'm still very interested in investigating the topographic correction so
> if you have any suggestion please, let me know!!
>
>
>
> Ciao,
>
> Roberta
>
>
>
> [0] https://www.arcsi.remotesensing.info/index.html
>
>
>
> 2018-06-14 10:55 GMT+02:00 Stefan Blumentrath <Stefan.Blumentrath at nina.no
> >:
>
> Hi Roberta,
>
>
>
> What about topographic correction (i.topo.corr [1])?
>
> Should that be run before as well?
>
> I would assume it has some impact on shadow detection in steeper terrain…
>
>
>
> Cheers
>
> Stefan
>
>
>
> 1: https://grass.osgeo.org/grass74/manuals/i.topo.corr.html
>
>
>
> *From:* grass-dev <grass-dev-bounces at lists.osgeo.org> *On Behalf Of *Roberta
> Fagandini
> *Sent:* torsdag 14. juni 2018 10:29
> *To:* Moritz Lennert <mlennert at club.worldonline.be>
> *Cc:* GRASS developers list <grass-dev at lists.osgeo.org>
> *Subject:* Re: [GRASS-dev] New GRASS addon i.sentinel.mask is now
> available (GSoC 2018 project)
>
>
>
> Hi Moritz!
>
>
>
> 2018-06-14 10:03 GMT+02:00 Moritz Lennert <mlennert at club.worldonline.be>:
>
> On 13/06/18 17:35, Roberta Fagandini wrote:
>
> Hi all!
> The first version of i.sentinel.mask is now available and can be installed
> from the official svn repository.
> i.sentinel.mask is a module for cloud and shadow masks computation and it
> is the first step of my GSoC project.
>
>
> Thank you, Roberta. Great job !
>
>
>
> Thank you!
>
>
>
>
> One question: ISTR that you mentioned that the input to this module has to
> be atmospherically corrected images, i.e. land surface reflectance. Is this
> true ? If yes, then this needs to be mentioned in the man page.
>
>
>
> Yes, it's true! I'm going to add it.
>
>
>
> Roberta
>
>
>
>
>
> Moritz
>
>
>
>
>
> _______________________________________________
> grass-dev mailing list
> grass-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/grass-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/grass-dev/attachments/20180618/415b8688/attachment.html>