[GRASS-dev] Unexpected EVI range from "i.vi"

Nikos Alexandris nik at nikosalexandris.net
Tue Jun 11 03:43:06 PDT 2013


Markus Neteler wrote:

> Hi Nikos,

Salut!

> if you have a i.landsat.toar corrected scene, please extract the values for

--%<---
I have to report-back, though, that using corrected (DOS1) bands (== 
reflectances), EVI looks OK!, i.e.

	min = -1.27297025550878  max = 1.4896100159326
--->%--

> - water pixel
> - green vegetation
> - asphalt
> 
> ... all bands except for TIR.
> 
> I want then to synthetic channels of one pixel to do
> the i.vi testing:
> 
> # prepare comp. region
> g.region rast=lsat7_2002_10 rows=1 cols=1 -p
> 
> # create synthetic channels (for the letters, we need your values)
> r.mapcalc "water_b7 = aa"
> r.mapcalc "water_b5 = bb"
> r.mapcalc "water_b4 = cc"
> r.mapcalc "water_b3 = dd"
> r.mapcalc "water_b2 = ee"
> r.mapcalc "water_b1 = ff"
> 
> Likewise for the other "landuses". This allows us to create a tiny test
> environment to check the output. I am just lacking a prepared Landsat
> scene (and the time to process it). Since you already have it, you may
> extract the values easily with r.what or the wxGUI.


Noted (for tonight).

With the exception of using DNs (which is not good anyway for VIs), I 
wrongly(?) expected numbers ranging in [-1,1].  For example, when using 
"i.landsat.toar method=uncorrected" or any other transformed set of values 
(meaning radiance or reflectance).

Anyhow, will try to check tonight.

Thanks, Nikos


More information about the grass-dev mailing list