Skip main navigation

Vegetation indices

An overview of commonly vegetative that use multispectral data, such as NDVI and Leaf Area Index

Hyperspectral imaging enables the capture of vast amounts of data at a range of wavelengths both inside and outside the range of colours detectable by the human eye. This allows for measurements of both ultraviolet and infrared radiation intensities and everything in between.

The drawback to this approach, however, is the computational and storage costs of dealing with such large amounts of data. Furthermore, in many practical cases, measurement of the full spectral response curve is not needed to capture a feature of interest, rather we more often need a specific point or set of points on the curve. This is where the use of multi-spectral imaging and vegetation indices comes in.

Vegetation indices

A vegetation index is a calculation based on the reflectance of two or more light wavelengths that aims to approximate some aspect of plant biology. This could be the amount of canopy cover, the age or growth stage of a plant, or the monitoring of disease. Such measures are particularly useful when collecting data at very large spatial scales, i.e., at a field scale or above. Data like this is usually captured via remote sensing, either using a UAV (Unmanned Aerial Vehicle or drone) or satellite. Due to the practical challenges of recording and transmitting data remotely, multi-spectral rather than fully hyper-spectral data is often recorded. It is also dependant on the sensors fitted to the imaging device, which of course may have been built many years ago, or have limitations in payload weight. For example, Sentinel2, an earth observation satellite launched and maintained by the European Space Agency records and transmits data at 13 bands in the visible and infrared parts of the spectrum (https://www.esa.int/Applications/Observing_the_Earth/Copernicus/Sentinel-2).

Example data of a river estuary taken from Sentinel-2 showing RGB 'true' colour and data at the shortwave-infrared wavelength 'B12'

Example data taken from Sentinel-2 using the Python API. https://sentinelhub-py.readthedocs.io/en/latest/
In remote sensing applications, because of the distance between camera and object, it is often difficult or impossible to recover much information about the physical shape of plants, or even distinguish between individual plants. Vegetation indices can therefore be a useful approximation of plant life over large spatial scales. Numerous vegetation indices have been developed, and each provides a subtly different measure of the plant biomass in a given area.

NDVI

Perhaps the most commonly used vegetative index in remote sensing is Normalised Difference Vegetation Index (NDVI). This is often used in agriculture and agricultural research to monitor and model crop growth and health. Generally speaking, NDVI is used to determine the amount of live green plant biomass in a given pixel and is calculated as follows:

[NDVI = frac{NIR-RED}{NIA+RED}]

where NIR is the measurement of light in the near-infrared region, and RED is the usual visible red-light measurement. This is better than simply using green visible light as chlorophyll in healthy plant cells actively absorb red light and reflect near-infrared light. While the absorption of red light means plants appear green to our eyes, the inclusion of invisible near-infrared in the NDVI measure is actually a better proxy for overall plant health and biomass.

Interpretation of NDVI

Mathematically, NDVI will always fall between -1 and 1, with water having an NDVI of near to -1, rock or other barren landscape around 0, and areas with very dense vegetation such as rainforest having NDVI approaching 1. Because NDVI is a single number, images showing NDVI can easily be displayed as a grayscale, though because of the nature of the data it is often displayed as a colour map, for example with -1 appearing red or black, 0 pixels appearing yellow, and 1 appearing as green.

Other indices

Several multi-spectral indices have also been developed based on NDVI for use in specific contexts such as SAVI (Soil Adjusted Vegetation Index), which adjusts for soil reflectance in areas with lots of exposed soil, and EVI (Enhanced Vegetation Index), which adjusts for atmospheric and canopy background noise in regions with high biomass. Another index you might encounter is FAPAR (Fraction of Absorbed Photosynthetically Active Radiation), which measures exactly what you might expect from the name and is often used to estimate carbon dioxide uptake by plants. Although most of these indices are designed to be computationally lightweight and utilise multispectral data, improvements in technology have enabled the development of vegetation indices that use hyper-spectral data, such as MDI (Moment Distance Index).

Leaf area index

Finally, another measure you might encounter in this context is LAI (leaf area index). This is defined as the fraction of leaf area to ground area and so is effectively the ‘true’ value of plant canopy cover.

A diagram showing the calculation of Leaf Area Index or LAI by dividing image area containing leaf by the total image area.

If, for example, you were to take a picture of a plant or plants from above, to calculate LAI you would need to calculate the area of the image containing leaf material relative to the total area of the image. In practice, if the camera is a long distance away from your area of interest it is often difficult or impossible to distinguish even individual plants, and so simpler measures like NDVI may be used instead to estimate vegetation cover. Often in fact, NDVI measurements are used to estimate the true LAI values over large scales and over time.

Summary

As always in research the correct data acquisition and use of these indices will depend on context (and also budget!), but hopefully this article provides some useful jumping off points if such an approach would be relevant to your research. For a list of vegetation indices see https://en.wikipedia.org/wiki/Vegetation_Index

This article is from the free online

Introduction to Image Analysis for Plant Phenotyping

Created by
FutureLearn - Learning For Life

Our purpose is to transform access to education.

We offer a diverse selection of courses from leading universities and cultural institutions from around the world. These are delivered one step at a time, and are accessible on mobile, tablet and desktop, so you can fit learning around your life.

We believe learning should be an enjoyable, social experience, so our courses offer the opportunity to discuss what you’re learning with others as you go, helping you make fresh discoveries and form new ideas.
You can unlock new opportunities with unlimited access to hundreds of online short courses for a year by subscribing to our Unlimited package. Build your knowledge with top universities and organisations.

Learn more about how FutureLearn is transforming access to education