[GRASS-user] i.atcorr with wordview2 image return Nan values

Dear Grass users,

i am running i.atcorr on a worldview2 band (red band) and the resulting map is empty, values are only NaN.

the input image is in reflectance, the values are integer between 0-255

i have set the g.region with the input image

the parameters are

16 # indicates that it is worldview2 image
12 11 15.750 -47.115 -22.822 # image taken on the 11th of December at 15:45 GMT in decimals; the center of the image lies at 47.115 °W and -22.822°S
1 # the tropical atmospheric model
1 # the continental aerosol model
0 # I have an estimate of aerosol optical depth so 0
0.100 # aerosol optical depth at 550nm
-.6121813 # the terrain lies 612 meters above sea level [km] * -1
-770 # image taken of a satellite sensor at 770 km
106 # 106 : red band of the worldview2 sensor

I see that the first part is ok, the algorithm compute the estimated visibility (63.66 km) , but then the atmospheric correction did not run and produce and empty raster.

the test with an elevation raster didn’t change nothing.
i have try to set by hand the wl inf and sup in the parameter file (wl inf= 0.350 mic wl sup= 1.100 mic, for the red band ), here the return by verbose is slighty different, but again i have an empty raster

does anyone have an idea of what i am missing ?

Cheers,

Fabien

* Fabien Wagner <wagner.h.fabien@gmail.com> [2017-04-11 15:21:41 -0300]:

Dear Grass users,

i am running i.atcorr on a worldview2 band (red band) and the resulting map
is empty, values are only NaN.

the input image is in reflectance, the values are integer between 0-255

i have set the g.region with the input image

the parameters are

16 # indicates that it is
worldview2 image
12 11 15.750 -47.115 -22.822 # image taken on the 11th of December at
15:45 GMT in decimals; the center of the image lies at 47.115 °W and
-22.822°S
1 # the tropical
atmospheric model
1 # the continental
aerosol model
0 # I have an estimate
of aerosol optical depth so 0
0.100 # aerosol optical depth
at 550nm
-.6121813 # the terrain lies 612
meters above sea level [km] * -1
-770 # image taken of a
satellite sensor at 770 km
106 # 106 : red band of the
worldview2 sensor

I see that the first part is ok, the algorithm compute the estimated
visibility (63.66 km) , but then the atmospheric correction did not run and
produce and empty raster.

the test with an elevation raster didn't change nothing.
i have try to set by hand the wl inf and sup in the parameter file (wl
inf= 0.350 mic wl sup= 1.100 mic, for the red band ), here the
return by verbose is slighty different, but again i have an empty raster

does anyone have an idea of what i am missing ?

Dear Fabien,

this https://trac.osgeo.org/grass/ticket/2059 is still standing, as far
as I can tell.

Nikos

Thank you for your answer Nikos,

may it possible to enter manually the " Red of wv2_spectral_response_for_i.atcorr " in the list of parameters ? For example in the parameter files, as specified in the manual, set the Sensor band to 1 and Enter wlinf, wlsup and user’s filter function s(lambda) by step of 0.0025 micrometer. what do you think ? i will try today and see if it works
cheers,

Fabien

···

2017-04-12 7:58 GMT-03:00 Nikos Alexandris <nik@nikosalexandris.net>:

Dear Grass users,

i am running i.atcorr on a worldview2 band (red band) and the resulting map
is empty, values are only NaN.

the input image is in reflectance, the values are integer between 0-255

i have set the g.region with the input image

the parameters are

16 # indicates that it is
worldview2 image
12 11 15.750 -47.115 -22.822 # image taken on the 11th of December at
15:45 GMT in decimals; the center of the image lies at 47.115 °W and
-22.822°S
1 # the tropical
atmospheric model
1 # the continental
aerosol model
0 # I have an estimate
of aerosol optical depth so 0
0.100 # aerosol optical depth
at 550nm
-.6121813 # the terrain lies 612
meters above sea level [km] * -1
-770 # image taken of a
satellite sensor at 770 km
106 # 106 : red band of the
worldview2 sensor

I see that the first part is ok, the algorithm compute the estimated
visibility (63.66 km) , but then the atmospheric correction did not run and
produce and empty raster.

the test with an elevation raster didn’t change nothing.
i have try to set by hand the wl inf and sup in the parameter file (wl
inf= 0.350 mic wl sup= 1.100 mic, for the red band ), here the
return by verbose is slighty different, but again i have an empty raster

does anyone have an idea of what i am missing ?

Dear Fabien,

this https://trac.osgeo.org/grass/ticket/2059 is still standing, as far
as I can tell.

Nikos

On Wed, Apr 12, 2017 at 12:58 PM, Nikos Alexandris
<nik@nikosalexandris.net> wrote:

* Fabien Wagner <wagner.h.fabien@gmail.com> [2017-04-11 15:21:41 -0300]:

...

Dear Fabien,

this https://trac.osgeo.org/grass/ticket/2059 is still standing, as far
as I can tell.

The Worldview2 filter functions are apparently containing some wrong values.

Today I have re-generated the CSV table from a spreadsheet (see r70887
and the ticket) and replaced the Worldview2 filter functions
accordingly using the code generated by create_iwave.py.

Upon successful testing in current trunk this needs to be backported to 7.2.x.
Please let us know if the updated version now works.

Markus

On Sun, Apr 16, 2017 at 10:34 PM, Markus Neteler <neteler@osgeo.org> wrote:

On Wed, Apr 12, 2017 at 12:58 PM, Nikos Alexandris
<nik@nikosalexandris.net> wrote:

* Fabien Wagner <wagner.h.fabien@gmail.com> [2017-04-11 15:21:41 -0300]:

...

Dear Fabien,

this https://trac.osgeo.org/grass/ticket/2059 is still standing, as far
as I can tell.

The Worldview2 filter functions are apparently containing some wrong values.

Today I have re-generated the CSV table from a spreadsheet (see r70887
and the ticket) and replaced the Worldview2 filter functions
accordingly using the code generated by create_iwave.py.

Upon successful testing in current trunk this needs to be backported to 7.2.x.
Please let us know if the updated version now works.

Never mind, I accidentially committed in relbr72. Now forward-ported
to trunk, too, in r70888.

Testing can now be done both in relbr72 and trunk.

If ok, the ticket can be closed.

Markus

Dear Markus and Nikos,
i have Ubuntu 16.04.2 LTS. We have removed grass 7.2 and install it again but there are no changes in the result from i.atcorr. I have two questions :
1 - do we need to compile grass again ?
2 - where can we find the source code to compile the new version ?
Cheers,
Fabien

···

2017-04-16 17:39 GMT-03:00 Markus Neteler <neteler@osgeo.org>:

On Sun, Apr 16, 2017 at 10:34 PM, Markus Neteler <neteler@osgeo.org> wrote:

On Wed, Apr 12, 2017 at 12:58 PM, Nikos Alexandris
<nik@nikosalexandris.net> wrote:

Dear Fabien,

this https://trac.osgeo.org/grass/ticket/2059 is still standing, as far
as I can tell.

The Worldview2 filter functions are apparently containing some wrong values.

Today I have re-generated the CSV table from a spreadsheet (see r70887
and the ticket) and replaced the Worldview2 filter functions
accordingly using the code generated by create_iwave.py.

Upon successful testing in current trunk this needs to be backported to 7.2.x.
Please let us know if the updated version now works.

Never mind, I accidentially committed in relbr72. Now forward-ported
to trunk, too, in r70888.

Testing can now be done both in relbr72 and trunk.

If ok, the ticket can be closed.

Markus

2 - where can we find the source code to compile the new version ?

For a fresh svn checkout see:

https://trac.osgeo.org/grass/wiki/DownloadSource

-----
best regards
Helmut
--
View this message in context: http://osgeo-org.1560.x6.nabble.com/i-atcorr-with-wordview2-image-return-Nan-values-tp5316891p5317492.html
Sent from the Grass - Users mailing list archive at Nabble.com.

On Mon, Apr 17, 2017 at 2:04 PM, Fabien Wagner
<wagner.h.fabien@gmail.com> wrote:

Dear Markus and Nikos,
i have Ubuntu 16.04.2 LTS. We have removed grass 7.2 and install it again
but there are no changes in the result from i.atcorr.

Yes, because the official release does not contain it yet.

I have two questions :
1 - do we need to compile grass again ?

Yes and no:
- no: you can take a generic binary snapsjot created on the GRASS GIS server:
   https://grass.osgeo.org/grass72/binary/linux/snapshot/
   --> I just forced an update there, will be ready in 1h from now. It
comes with an install script.

- or, compile yourself (that's a bit more of effort), see below.

2 - where can we find the source code to compile the new version ?

As stated by Helmut, you need to get the code from SVN:
https://trac.osgeo.org/grass/wiki/DownloadSource

HTH,
Markus

(re-adding the list)

On Tue, Apr 18, 2017 at 2:58 PM, Fabien Wagner
<wagner.h.fabien@gmail.com> wrote:

Thank you for your answers,
We have install the generic binary from
https://grass.osgeo.org/grass72/binary/linux/snapshot/
then launch grass (in the console the name of grass is now GRASS 7.2.1svn)
some error appears because this version was not able to locate some library
but we manage to correct this.
then run i.atcorr after inporting the red band raster and selecting the
g.region of the red band raster.
i.atcorr run without returning any error. he stops after complete 100% of
the atmospheric correction and return the empty raster with values ranginf
from -nan to -nan
The first part of the algorithm run as it computes the visibily in km (from
the aerosol optical depth I give in the parameter file in txt)
if someone want to test i can send the image and the parameter file, because
i don't really know what i can do now

Yes, I could try to check it.

The best would be a zipped location/mapset along with the parameter
file (since the data are non-free, I suggest to not send the download
link over this list).

Best
Markus

cheers,
Fabien