*This page introduces the scientific background of NDVI and its application to the [PLOTS visible/infrared camera tool](/wiki/dual-camera-kit-guide).* 
Figure 1. The absorption of different wavelengths of visible light by photosynthetic pigments and the resulting spectral response of photosynthetic rate. Leaves appear green because the pigments do not absorb green wavelengths and reflect them all back for us to see. Modified from [here](http://en.wikipedia.org/wiki/File:Par_action_spectrum.gif)
Vegetation is green because plant leaves reflect green light -- they don?t use it for photosynthesis (Figure 1). Instead they use almost all of the blue and red wavelengths in sunlight. The pigments in leaves absorb this light to power photosynthesis which converts CO2, water, and nutrients into carbohydrates (food). In general, you can estimate the productivity or vigor of vegetation by how much blue and red light it is absorbing. Photosynthetic pigments do not use the longer, invisible wavelengths of infrared light and reflect almost all of it away (this helps prevent the leaves from overheating). About a year after the launch of the first Landsat satellite in 1972, scientists began using the data from its sensors to estimate the productivity of vegetation by comparing the amount of red light reflected (there is not much from healthy plants) to the amount of near infrared light reflected (there is a lot). The amount of infrared light reflected from vegetation is a good indicator of how bright the sunlight was at any moment (leaves reflect all IR), and comparing that to the amount of reflected red light can tell us what proportion of the sunlight was being absorbed by the plants. Figure 2. The equation for computing NDVI, the Normalized Difference Vegetation Index.
That relationship is a good measure of the amount of photosynthetically active biomass. They quickly settled on an index of plant productivity called NDVI for Normalized Difference Vegetation Index. Instead of just using the difference between the amounts of red and near infrared light, they normalized that difference by dividing it by the total amount of red plus infrared light (Figure 2). That allowed the index from different areas and different times of the day or year to be compared with each other.  _Figure 3. The width of the red and near infrared bands sensed by different satellites and used to calculate NDVI. The earliest satellites are to the left. At far right are the approximate bands captured by the visible/IR camera pair developed for the PLOTS KickStarter campaign. Data from [here](http://rangeview.arizona.edu/Glossary/ndvi.html) and elsewhere._ It was soon demonstrated that NDVI was a pretty good proxy for vegetation health or the amount of photosynthetically active biomass. Dozens of subsequent satellites have returned data about the red and near infrared light reflected from Earth?s surface, and for 40 years NDVI computed from these two bands has been a standard way of describing vegetation health or productivity. Over the years, the exact bands of wavelengths used to compute NDVI differed from satellite to satellite (Figure 3), but the index remained a fairly robust indicator of primary productivity. The most recent satellites sense relatively narrow bands of red and near infrared light. Figure 4. A Bayer filter array allowing each pixel on the silicon sensor to record information from either red, green, or blue light.
It is possible to produce a good approximation of NDVI by using a normal consumer grade digital camera to capture visible light (including red), and another similar camera modified to capture only near infrared light. Silicon sensors (both CCD and CMOS) are more or less equally sensitive to all visible wavelengths, but by themselves cannot distinguish among different colors. So red, green, or blue filters are placed over each pixel to record information about just one color (Figure 4). An algorithm then assigns information about all three color channels to each pixel in the image file.