[GRASS-dev] i.atcorr @ GRASS6.4 using RapidEye: strange output

Hi,

I'm reporting again on the mixed performance of i.atcorr for RapidEye satellite data in GRASS6.4.2 on SuSE.

In a test case RapidEye satellite data for the Spearfish,SD demo site is used.

Rapideye data comes in 16bit.

The value ranges (digital numbers) for the first three bands are:
red: [1286 - 58603]
green: [2464 - 54248]
blue: [4069-56975]

So two variants were tried:

Variant A) Using r.rescale to scale the data down to 8 bit

Example [red band]:
i.atcorr iimg=spearfish_3.255.rad iscl=0,255 oscl=0,255 icnd=re_20100509_3.atcorr_20120814 oimg=spearfish_3.255.rad.atcorr_20120814 --o

The output map consists mostly of zeroes and NULLs with a few 1s and 2s for both red and blue band. "i.atcorr-ing" the green band returns just NULLs.

Variant B) i.atcorrs "iscl"-option is used to provide the 16bit data range for each band.

Example [red band]:
i.atcorr iimg=spearfish_3 iscl=1286,58603 oscl=0,255 icnd=re_20100509_3.atcorr_20120814 oimg=spearfish_3.atcorr_test1_1286-58603_0-255 --o

Now the results for the red band seem reasonable, the results for the blue band look like a "salt and pepper" pattern and the green band again goes NULL.

Ideas, anyone ? Does this kind of behaviour also occur for other sensors ?

Best,
Peter

--
Dr. Peter Löwe
<peter.loewe@gmx.de>