Earthdata Forum

Hello,
I am working on developing a chlorophyll-a algorithm using MODIS data for coastal water. Commonly, reflectance ratio are used in algorithm development where numerator is blue or red band and denominator is GREEN band. Will it be scientifically wrong if green band is used as numerator and red band as denominator in the reflectance ratio?
Also, is there a nice tutorial to learn the complex regression analysis used for the algorithm development? Kindly guide me.

Sincere thanks,
Manzar Abbas
Florida International University

The band ratio algorithms are based on an empirical relationship between the measured quantity and the quantity of interest, in this case reflectance and chlorophyll, respectively.
This particular blue:green ratio relationship was chosen because it has been observed that as the chlorophyll concentration increases the magnitude blue:green ratio decreases.  The dynamic range of the green reflectance is also, typically, smaller than that for the blue wavelengths - and it's nice to have a relatively stable quantity in the denominator.  We also have a reasonable explanation for why this reflectance to chlorophyll concentration relationship exists (pigment absorption). 

Now, to your question as to the scientific validity of moving the ratio from blue:green to green:red...if you can identify an empirical relationship between a green:red, you can still argue that it is rational using the same pigment absorption premise as the blue:green ratio.    Chlorophyll absorbs in the red as well  - but also fluoresces there, so depending on the bandpass of the red band used it may be contaminated by a fluorescence signal.  Others have proposed (and published) green:red ratio relationships, even red:nir relationships. So, yes, it's scientifically valid, but not particularly innovative. 

Sean