Most researchers in the field of remote sensing working to converting DN values to reflectance before making andy process. Can we discuss that, why and what is the estimated differences ?
If you're only using a single date, with no prior spectral signatures or thresholds from previous study, there's no real need to transform the original DNs to radiance or reflectance. Even a reflectance value might need to be BRDF normalized if sun and view angles where not the same between date.
Data in DN is a singular extratification that is performed for each band of the sensor, according to the radiometric resolution of it (2^bits). Therefore, we do not recommend algebraic operations between bands when in DN value. In addition, DN information, being a radiometric parameter, is influenced by the intensity of solar energy on the surface, so that images of different dates should not be compared.
DN could be different between two cameras for the same target. So the DN is with instrumental characteristics.
The reflectance is the property of the target which is supposed to be not changed by different radiometers' observing under the same geometric conditions.
Sometimes, in a special case, DN is also OK to investigate spectral or intensity difference.
Usually, reflectance is better than DN.
The reflectance is better comparing to DN values, since the first is a fixed characteristic of target, while the last is a function of radiometric resolution.
DN are values from satelital product which depends from the imagery resolution (4 bits, 16 bits, 32 bits .....) and its use directly to compute environmental information is not recommended because they are mathematical values or representations of spectral data without physical magnitudes, by this reason we need to transform DN to TOA (for example) to convert unphysical units to physical units and have better representation of objects spectral response .
- Badges
- Science topic