[GRASS-user] no output when running i.atcorr

[Markus Neteler]

Hi Annekatrien,

news regarding i.atcorr!

On Fri, Jan 30, 2009 at 7:01 PM, Annekatrien Debien
<annekatrien.debien at gmail.com> wrote:
> Hi all,
>
> I want to do an atmospheric correction of my Landsat ETM+ images using the
> i.atcorr module in Grass 6.3 on Ubuntu, but I never get an output. When I
> use a DEM in ialt, the continuous version or the categorized, of the region,
> I just get "wavelength less than 0.25 micron              let's take
> s(1)=s(0.25)". I don't insert an altitude value in my icdn file however,
> since I've read that that value would overwrite the DEM.

@Yann: do you have any pointer for us?
Since the DEM *is* read in in the code, is the documentation statement true?
I don't understand the code well.


> When I do put in an
> altitude value, I get a list with the parameters for the 6S-algorithm, and
> then 'percent complete', so that seems to be ok, but the algorithm stops
> running, so I don't get a corrected image.

As discussed, it just runs "forever", say it takes really long. But there is
hope:

With great help from Yann Chemin,  I have submitted to all branches/trunk
some changes:
- cache of 1024 instead of 128
- bugfix in example (channel 1 and 4 was mixed; fixed target height)
- example update (suggesting and using integer DEM)

The use of an integer DEM seems to be the trick.
It is relatively fast now with the North Carolina Landsat (earlier it
didn't finished in several hours and I gave up):

g.region rast=lsat7_2002_40 -p
r.info lsat7_2002_40

# using an integer DEM greatly accelerates the i.atcorr computations
r.mapcalc "elev_int = round(elevation)"

# find mean elevation (target above sea level)
r.univar elev_int

# create control file for channel 4 (NIR)
echo "8                            - geometrical conditions=Landsat ETM+
5 24 14.30 -78.691 35.749    - month day hh.ddd longitude latitude
("hh.ddd" is a decimal hour GMT)
2                            - atmospheric mode=midlatitude summer
1                            - aerosols model=continental
50                           - visibility [km] (aerosol model concentration)
-.110                        - target at 110m above sea level
-1000                        - sensor on board a satellite
64                           - 4th band of ETM+ Landsat 7" > icnd.txt

# run atmospheric correction (note the -o for optimization)
i.atcorr -o lsat7_2002_40 ialt=elev_int icnd=icnd.txt oimg=lsat7_2002_40_atcorr

This takes  20 seconds (!) on a "normal" laptop. Without -o it takes
long, after
10 minutes it is still at 0%...

Cheers
Markus