Leaves, when healthy and vigour show a characteristic green colour. This visual effect evident to humans is caused by the co-existence of two evolutionarily facts: the specific interaction of the chlorophyll pigment in living leaves to the visible spectrum (VIS, 400-700 nm wavelength) of light emitted by the sun and the sensitivity of our human eye to the same sub-spectrum. According to fundamental physical laws of radiation (Stefan Boltzmann law of blackbody radiation and Wien’s displacement law), the VIS sub-spectrum corresponds to the radiation maximum of the sun, a hot blackbody with a surface heat of about 6000 K. Living leaves are structured in specific layers exhibiting characteristic interaction with light. The chloroplasts located in the so-called palisade layer, make use of the blue and the red part of sunlight for photosynthesis, the unique process of transforming light to create energy (carbohydrates) from water and carbon dioxide. This leads to the specific behaviour of leaves to absorb large portions (up to 90%) of the blue and red part of the electromagnetic spectrum and reflect nearly 100% of the green light. The peak reflectance in green light makes leaves (and plants in general) appear in green colour in our visual perception.
A second, by no means less characteristic, feature of leaves is the specific response to near infrared (NIR, at around 700 nm wavelength) light in the mesophyll tissue (transmittance, scattering and reflectance). Only a small fraction of NIR is being absorbed.
This combination of two specific spectral characteristics, the absorption in VIS (red colour) by chlorophyll a in palisade layers, and the reflectance of NIR in the spongy tissue, makes the spectral profiles of plants and vegetation exhibiting a very characteristic shape, the so-called red edge. This absorption edge between red and NIR light is sharper for higher intensity green reflectance and brighter green tones (such as grassland or bright deciduous forest) than for less intensive reflectance and darker tones (coniferous forest).
The red edge may shift for the same vegetation type due to plant maturity or plant stress. This effect we call the red shift. The red shift is sensitive to crop maturity (headed stage) and may indicate harvesting time. Notably, there is also a blue shift, indicating green plants’ exposure to geochemical stress, which causes the absorption spectra to shift towards shorter wavelengths.
Plants usually do not appear in isolation but form a canopy with a certain degree of coverage (e.g., crown closure in forests), and a certain part of understorey or soil per area unit. The resulting canopy reflectance is therefore a spectral mix of soil and vegetation (or even different types of vegetation) and generally lower than the reflectance of a pure vegetation sample under lab conditions.
To capture most of these plant-typical spectral characteristics, the so-called normalised difference vegetation index (NDVI) was developed. NDVI is an arithmetic band combination of red and NIR bands in a normalised value range.
The NDVI is calculated as:
NDVI=((NIR-R))/((NIR+R))
The (hypothetic) value range of the NDVI is [-1 | +1]. Under real-world conditions, the NDVI ranges from values of around -0.2 to 0.6 or 0.7. To discriminate principal land cover classes such as water, non-vegetation (soil, sealed, etc.) and vegetation the following thresholds in the continuous range are used:
NDVI < ~ 0: water
~ 0 < NDVI < ~ 0.2: non-vegetation (soil, sealed surfaces, bare rock, etc.)
~ 0.2 < NDVI: vegetation.
Notably, these class limits are just a very rough approximation (indicated by the ~ sign), due to the mixed pixels effect, canopy reflectance, the abundance of water plants and suspending particles, and the illumination effect of specific atmospheric or topographic conditions.
We can use the NDVI to generally mask out vegetation from other land cover types and, more specifically, to indicate vegetation vigour and health. It is also suitable for monitoring plant phenology as the relationship between vegetative growth and the (changing) conditions of the environmental conditions. A range of variations has been suggested, enhancing one or the other mathematical or statistical behaviour of the index, or making it even more sensitive to specific plant behaviour. A well-known example is the enhanced vegetation index (EVI).