[GRASS-user] i.landsat.toar and zero values in GRASS 7.0.0

[Nikos Alexandris]

Katerina-Karolina Karapanou:

> I have a Landsat TM image (2007) and I work in GRASS 7.0.0. in Windows
> 7. Once again, trying to apply the module i.landsat.toar in the right
> way, via GUI, I have the outcome of 7 bands under the name Btoar1,
> Btoar2 etc.

> Also trying to check the results via r-info for all bands I realized
> that bands have different values and also 0 values.

It would be suspicious, and most likely an error, if all bands have had
the _same_ values. And, zero values are also normally there.  Check what
the zero pixels actually are (maybe create a new raster map with
r.mapcalc for all zero pixel values).  Are they just a few, in- or
near-water, for example?  Too dark surfaces?

> For example for Band1:

>  |   Range of data:    min = 39.0688188976378  max = 193                      
>  |                                                                            
>  |   Data Description:                                                        
>  |    generated by i.landsat.toar                                            
>  |                                                                            
>  |   Comments:                                                                
>  |     Radiance of Landsat-5 TM (method uncorrected)                          
>  |    -----------------------------------------------------------------       
>  |     Acquisition date (and time) ........... 2007-06-24 (-0.0000 h)         
>  |     Production date ....................... 2014-05-09                     
>  |                                                                            
>  |     Earth-sun distance (d) ................ 0.0000000                      
>  |     Sun elevation (and azimuth) ........... -0.00000 (0.00000)             
>  |     Digital number (DN) range ............. 0 to 0                         
>  |     Calibration constants (Lmin to Lmax) .. +0.00000 to +0.00000           
>  |     DN to Radiance (gain and bias) ........ -0.00000 and +0.00000          
>  |     Mean solar irradiance (ESUN) .......... -0.000                         
>  |     Radiance to Reflectance (divide by) ... -0.00000                       
>  |    ------------------------------------------------------------------      
>  |                                                                            
>  |    i.landsat.toar -r input="B" output="Btoar" metfile="C:\Users\User\Desktop\thessaloniki\2007tm\thessaloniki2007tm\LT51840322007175MOR00_M\ 
>  |    TL.txt" sensor="tm5" method="uncorrected" percent=0.01 pixel=1000 rayleigh=0.0 scale=1.0         


The all zero values for band #1 may, inded, worth re-checking your
processing. But the rest, below, looks ok to me.  It's not unusual to
have a few "erroneous" negative values.  Check how many.  I guess just a
few. There are various reasons why this can happen (search in the
literature). 

> Also for band2: min: 27.571968503937 max: 365
>                band3: min: 13.4456692913386 max: 264
>                band4:min:  -0.633976377952756 max:221
>                band5:min: 0.231771653543307 max: 30.2
>                band6:min: 8.87961417322835 max: 12.3681771653543 
>                band7:min: -0.15 max: 16.5
> 
> What I can do???

Remember, these are Radianve valuse, not Reflectances. So, they seem
quite normal. Maybe you can set the begative pixels to zero. This will
not harm, I think, your processing chain.

Nikos