Dn values after applying attmospheric correction method

[gisareafor]

Hi all

As you know atmospheric effect cuses real Dn values to be reduced.

So if we apply an atmospheric correction method like quic correction or log residual (using envi software)

resulted dn valuse should be in original range.

But what is the problem?

for example after applying quac method,dn values range grow to 10 or 20 time grater than original value.

What is the problem?Do you know?

Than you

[am2]

I have this problem too,Is there any suggestion?

[mamadouba]

I'm relatively new to this site. I came across this topic and did not see an answer or response. Digital number (DN) counts are integer values that are relatively meaningless until they are rescaled to an actual unit of measurement such as top of atmosphere (TOA) radiance or reflectance by applying a linear gain and offset calibration. The next step of the preprocessing chain would be to convert TOA radiance to surface reflectance, which requires an atmospheric correction algorithm such as FLAASH or QuAC. FLAASH input requires TOA radiance values because it's a physical based model, but QuAC can be performed directly on DN values because it's an empirical scene based model. Now that the background context has been provided, here's the answer to your questions. The output of most atmospheric correction models is unitless surface reflectance that can be scaled in any manner, such as floating point scale from 0.0 to 1.0, with 0.0 being no reflectance or 0% and 1.0 being full reflectance, or 100% reflectance. In the case of QuAC or FLAASH, the data are scaled by a 16-bit integer factor of 10000, hence the large output numbers. For example, a value of 3564 can be perceived as 0.3546 or 35.46% reflectance.