[GRASS-user] value differences between landsat images

Jonathan Greenberg greenberg at ucdavis.edu
Tue May 20 14:44:51 EDT 2008


Some responses below:
> Well, that's exactly the problem, i cannot make a complete mosaic for the
> region i'm working on with only images taken on one day, and it adds another
> problem: the images do not overlay well... if they did, i would been able to
> do a radiometric normalization easily (as Johnattan Greenberg suggested)...
By "do not overlay well" I assume you mean they aren't well 
orthorectified to one another -- although I consider this a "must do" 
for mosaicking and for any level of image-to-image work (because of the 
time savings), you COULD manually pick PIFs in the overlap zone (e.g. 
you'd pick a point in the reference image, then visually pick the 
corresponding pixel in the uncorrected image) -- orthorectifying first 
saves you the time of having to pick two points per image -- rather, you 
just pick one point.  How many images do you have in your mosaic?  
Image-to-image rectification with landsat imagery is fairly painless, 
shouldn't take you much more than 30 mins per image once you get going 
with it. 

At any rate, once you pick your PIFs, you can use v.what.rast or 
starspan (starspan.casil.ucdavis.edu) to get this data out to a table to 
do the regression calculations. 
> So my problem now is how to overlay landsat images taken in different days.
> I've already asked for help on this issue (with no solution so far; mail
> subject="mosaic with landsat geotiff"), so guess i'll resume that thread...

>> if it is simply an "automatic contrast" adjustment, you could linearly
>> interpolate between bands so they matched. (but then which is correct?) I
>> don't think it would be though, as LANDSAT has fixed calibration for each
>> band.
> How's that?
This is what I suggested as well -- the question of which one is 
"correct" is somewhat moot -- your goal, as I understand it, is to have 
all of the images in the same radiometric scale, so you can just pick 
one (usually one that's in the middle of your mosaic, and often 
temporally at the "edge" of your time period (either a very early or a 
very recent image).  If
>> if you run i.landsat.rgb on the two images with the same parameters do they
>> match up well?
> what do you mean with 'the same parameters'? (maybe "cropping intensity"...)
>> that doesn't touch the values, only the colors, but it may
>> give you a clue about the cause of the difference.
> How exactly? as an example, i can see a subtle difference in the color of
> water in a river as it pases on to another 'tile', but what does it means, i
> don't have a clue... (the values changes from: |32|40|86| to |20|26|69|, for
> bands 3, 2, 1 respectively)
>> I guess the important thing is the ratio of the bands, not the exact values
>>  of one particular band. I take it you see a hard line at the boundary in
>> the processed image?
> i suspect you mean the same as Johnattan G. with the radiometric
> normalization... anyway, i don't know what you mean with 'proccesed image'.
> The original images come with a null region as a result of the
> georreferencing, and after i use r.null -r to remove the NULL-value bitmap
> file, yes, there is a hard line in the boundary, but what would that mean?

It just means your images aren't radiometrically normalized to one 
another.  If you correctly normalize them you won't see a line in the 
>> Hamish
> JM
> -------------------------------------------------------
> _______________________________________________
> grass-user mailing list
> grass-user at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/grass-user

Jonathan A. Greenberg, PhD
Postdoctoral Scholar
Center for Spatial Technologies and Remote Sensing (CSTARS)
University of California, Davis
One Shields Avenue
The Barn, Room 250N
Davis, CA 95616
Cell: 415-794-5043
AIM: jgrn307, MSN: jgrn307 at hotmail.com, Gchat: jgrn307

More information about the grass-user mailing list