11 Aug 2016 05:52 AM |
|
Hi,
I have ASTER L1T data that I need to work on and seems to be in DN values. I have read from AST_L1T user guide and they state that to convert them it needs to use the formula: Radiance at-sensor = (DN value – 1) x Unit conversion coefficient where for that case Unit conversion coefficient for the data from HDF file are:
Band 1: 0.676000
Band 2: 0.708000
Band 3: 0.862000
the DN values are:
Band 1 0 255
Band 2 0 255
Band 3 0 251
I understand that when applying gain and offset in ENVI I can convert the values to Radiance (correct me if I am wrong). My worry is that when applying gain and offset I get -ve minimum values which I think it doesn't make sense.That negative value is been picked from offset value that I plug in as shown below:
Band 1: -0.676000
Band 2: -0.708000
Band 3: -0.862000
It seems to me the ENVI apply gain and offset uses the same formular as the one required only that it does the calculations to both minimum and maximum DN values.
My question is: Are the negative initial values make sense? If not what can be done so that I remain with minimum values of 0?
Thanks
|
|
|
|
MariM Veteran Member
Posts:2396  
11 Aug 2016 11:03 AM |
|
It looks like the same coefficients are entered as gains and offsets in the ENVI Header. However, if you use the Radiometric Calibration tool with ASTER L1T, it will apply the equation you mentioned and only the gains are used. So no band has a negative radiance value.
You can compare this by using band math and using the equation:
(b1-1)*0.6760
to calibrate band 1 of the ASTER dataset. Then compare this to the results from Radiometric Calibration (without using FLAASH Settings).
|
|
|
|
Deleted User New Member
Posts:  
11 Aug 2016 12:45 PM |
|
Mari,
The band math equation gives me exactly the same results as apply gain and offset function. See the below statistics:
Min Max
Band 1 -0.676000 171.703995
My concern here is that; does the -ve minimum value make sense or not? If not what should be done to correct this? I have no problem with the maximum value.
Regards
|
|
|
|
MariM Veteran Member
Posts:2396  
12 Aug 2016 05:44 AM |
|
When I used Apply Gain and Offset, my background pixels, which are 0, retained the 0 value. However, I do not believe this equation takes into account the -1 in the equation.
When using Band Math and the equation:
(b1-1)*0.676
You will apply the equation to all pixels in the image including the background. The background pixels that have a 0 will then take on the value of -0.067. So yes, it is expected that the new image minimum is -0.067. You can create a mask of the background and apply the mask to convert the pixels back to 0. You can also use the Radiometric Calibration tool which applies the correct equation and retains the background pixels of 0.
|
|
|
|
Deleted User New Member
Posts:  
12 Aug 2016 06:31 AM |
|
Ok. Thanks Mari
|
|
|
|