Visible vs. Near Infrared (NIR) Bands Applied to Crop Agriculture
Related page: Why apply NIR to crops Individual band correlation to soybean yield
Most remote sensing data applied to agriculture, whether from satellites, aircraft, vehicles, or hand-held sensors, include visible (blue, green, red) and near infrared (NIR) bands (green, red, and NIR are applied more often than blue; short-wave infrared can also be applied in useful ways). Certain types of sensors (hyperspectral) acquire so many (tens to hundreds) very narrow adjacent bands (ranges) of wavelengths that the data can be considered continuous across a portion of the electromagnetic spectrum. More commonly though, as with Landsat, data is acquired for relatively few broader wavelength bands.
Visible bands can be very useful in assessing crop condition. Chlorophyll in plant leaves absorb much of the blue, green and red light; more green light is reflected than other visible bands though which is why plants appear green. (When considering the effect of shadows in a canopy lowering visible wavelength reflectance, a closed canopy reflects less than ten percent of visible light [reflectance of blue, green, and red can average less than 5 percent].) As a result, there is a negative correlation with crop condition and yield; the lower the values of blue, green, or red reflectance, the more chlorophyll and healthy vegetation and the higher the crop condition. However, this correlation can only be extracted significantly enough from imagery for a time after the canopy is full enough (to diminish the effect of the background soil enough) to a time near but prior to the canopy is closed. When a canopy becomes closed, visible light is unable to detect differences very well across a field due to saturation (pixel values are the same or nearly the same throughout a field even if crop condition is different).
NIR reflectance is mainly controlled by the structure of the spongy mesophyll and has a positive correlation with crop condition and yield; NIR reflectance of a closed green canopy is typically greater than 50 percent. NIR radiation can be scattered upward or downward; downward scatter can continue to transmit through the canopy and reflect off the soil (Campbell, 2007) which has lower reflectance than green vegetation. As a result, if a green canopy is closed but is thinner in areas, NIR radiation (unlike visible light) can detect the difference (NIR radiation does not saturate like visible band radiation). Also, as plants are subjected to stress (e.g. disease, insects, or drought), change is more noticeable in the NIR than visible spectrum (Campbell, 2007). It is important to note that the NIR band is not effective enough at detecting corn condition differences after tassel (tassels obscure green vegetation too much) but can detect soybean crop condition very well through most of the reproductive stages.
The "red edge" spectrum, which is in the transition area after low red reflectance and before high NIR reflectance, can also be very useful for crop assessment.
Reference
Campbell, J. 2007. Introduction to Remote Sensing: Fourth Edition. The Guilford Press: New York, London.