Spectral Indices
Measurement Types
Spectral indices are numerical ratios created by combining values from different spectral bands of multispectral imagery to highlight specific features of Earth’s surface. Earth observation satellites enable the consistent creation of spectral index data products through their ability to collect calibrated and repeatable multispectral measurements globally and over long timescales.
Multispectral satellite data is captured in bands, each with a given spectral range that correlates to an observed radiance of the Earth. Radiances in certain spectral ranges can be used to determine qualities such as soil moisture, vegetation mass, and geological features. When sunlight is incident on Earth’s surface, different wavelengths of light are absorbed or reflected depending on the unique spectral characteristics of the incident material. This reflectance, as well as emitted radiation, is observable using EO satellites.
Related resources
Spectral indices can be used to enhance the signatures of specific phenomena by emphasising differences in bands. This is done by combining specific bands within the data and eliminating components of the spectral data that are less representative of the target. They can also reduce atmospheric, instrument noise, and sun angle effects, allowing for more consistent spatial and temporal comparisons. Some common applications of spectral indices include vegetation health (Normalised Difference Vegetation Index, NDVI) and burnt area mapping (Normalised Burn Ratio, NBR), water content (Normalised Difference Water Index, NDWI), and other biophysical parameters such as biomass and geological mapping.
The rise in hyperspectral imaging is providing information on many more spectral bands, allowing the creation of new spectral indices for highly specific purposes. 1) 2) 3) 4)
History
The first spectral indices created for Earth observation were simple vegetation products developed in 1969 prior to the launch of the first Landsat satellite. The Ratio Vegetation Index (RVI) and Vegetation Index Number (VIN) were ratios of the Landsat Red and Near-Infrared (NIR) bands that contrasted the difference between vegetation and other ground objects.
One of the most common spectral indices used today is the Normalised Difference Vegetation Index (NDVI), calculated by the difference between Near-Infrared (NIR) and red reflectance of Earth’s surface divided by their sum. NDVI is widely used to assess vegetation conditions, as the chlorophyll in healthy, photosynthesising plants strongly reflect NIR light, whereas plants affected by drought or disease reflect more red light. As a normalised index, NDVI is dimensionless (ranging from 1 to -1), allowing consistent comparisons across different datasets and measurement conditions. 12) 13)
Vegetation indices can also be derived from polarimetric SAR imagery by combining measurements from different polarisation channels. These measurements can highlight different characteristics, such as those associated with the structure and density of forests. The Radar Forest Degradation Index (RFDI), for example, represents the normalised difference of co- and cross-polarisation backscatter. This index differentiates between vegetation types and assesses the strength of the SAR ‘double bounce’ mechanism. 11)
Example Products
Normalised Difference Vegetation Index
The Normalised Difference Vegetation Index (NDVI) is the ratio of the red and near infrared (NIR) values, according to the formula: (NIR - R)/(NIR + R). It is used to quantify vegetation greenness, aiding understanding of vegetation density and plant health. Copernicus NDVI products can be accessed here. 5)
Enhanced Vegetation Index
The Enhanced Vegetation Index (EVI), similar to NDVI, quantifies vegetation greenness. It also corrects for atmospheric conditions and canopy background noise, as different leaf structures, particularly broadleaf and needle leaf canopies are found to produce differences in Soil Adjusted Vegetation Index (SAVI) values. Landsat EVI products can be accessed here. 6) 15)
Modified Normalised Difference Water Index
The Modified Normalised Difference Water Index (MNDWI) is designed to enhance open water features and suppress noise from built-up land, vegetation and soil. These features producing noise tend to reflect more short-wave infrared (SWIR) light than NIR wavelengths, and water tends to absorb more SWIR than NIR light. MNDWI was originally developed for use with Landsat-4 and -5 Thematic Mapper (TM) bands 2 and 5, as an improvement of the Normalised Difference Water Index (NDWI) with the use of SWIR bands, which provide better accuracy for urban water mapping over the NIR data used in NDWI. MNDWI can be produced by any sensor with a green band between 0.5 - 0.6 µm and an SWIR band between 1.55 - 1.75 µm. 7)
Normalised Burn Ratio
The Normalised Burn Ratio (NBR) product is used to identify burned areas and the severity of fire damage. It is derived from red and NIR band data, using the same logic as NDVI, quantifying vegetation greenness to determine burned area extent. NBR is produced according to the formula: (NIR - R)/(NIR + R), with Landsat NBR data accessible here. 8)
WorldView New Iron Index
The WorldView New Iron Index (WV-II) uses the WorldView-2 WorldView-110 Camera to identify iron-oxide rich pixels in an image. This index uses a spectral matched filtering (SMF) technique with a Gothite spectrum selected as the iron oxide mineral end member. This demonstrates the use of spectral indices for geological surveying. WV-II is produced according to the formula: WV-II = (G * Y)/(B * 1000). 9)
WorldView provides over 100 spectral indices for the WorldView 1-4 and Worldview Legion satellites, for a wide range of vegetation properties and harmonisation with other sensors. 16)
Normalised Difference Snow Index
The Normalised Difference Snow Index (NDSI) can be derived from green and SWIR band multispectral data. NDSI products indicate the probability of snow being present in a given area, where vegetation, soils and lithology endmembers register as NDSI pixel values closer to 1. Sentinel-2 NDSI products can be accessed here. 10)
Spectral Index | Formula | Description |
NDVI | (NIR - R)/(NIR + R) | Used to quantify vegetation greenness |
EVI | G * ((NIR - R) / (NIR + C1 * R – C2 * B + L)) | Similar to NDVI, but corrects for atmospheric conditions and canopy noise |
MNDWI | (G - SWIR)/(G + SWIR) | Enhances open water features and suppresses noise from built-up land, vegetation and soil |
NBR | (NIR - R)/(NIR + R) | Identifies burned areas and estimates severity of fire damage |
WV-II | (G * Y)/(B * 1000) | Identifies iron-oxide rich pixels |
NDSI | (G - SWIR)/(G + SWIR) | Estimates the probability of snow being present in a given pixel |
Related Missions
As spectral indices are derived by combining reflectance or backscatter measurements from two or more spectral bands or channels, they are not mission-specific products in themselves. Rather, they can be generated from a wide range of EO missions, provided the sensor offers the necessary spectral coverage, calibration quality, spatial resolution and revisit characteristics. Multispectral optical missions are most commonly used, particularly those with bands in the visible, near-infrared (NIR) and shortwave infrared (SWIR), while hyperspectral missions enable the development of more specialised indices using narrow contiguous spectral bands. In some cases, analogous indices can also be derived from SAR observations, for example by combining different polarisation channels to highlight vegetation structure or land-surface properties.
The missions listed below are representative examples of satellite systems whose data are commonly used to derive spectral indices, and do not represent an exhaustive or exclusive set. Their relevance depends not only on spectral band placement, but also on spatial resolution, temporal frequency, radiometric performance, data continuity and accessibility.
MODIS
The moderate resolution imaging spectrometer (MODIS) is a 36-channel VIS/IR radiometer, carried by the Terra and Aqua missions. MODIS provides data for NDVI and EVI for vegetation and NDWI and NDII (Normalised Difference Infrared Index) for water content. MODIS data is accessible here.
Sentinel-2
Copernicus Sentinel-2 is a constellation of optical imaging satellites each carrying a Multispectral Imager (MSI). MSI images across 13 bands in VNIR and SWIR, with a spectral range of 0.4 µm - 2.4 µm. This combination of spectral coverage, spatial resolution and open data access has made Sentinel-2 one of the most widely used missions for deriving common spectral indices such as NDVI, EVI, MNDWI, NBR and NDSI, as well as red-edge-based vegetation metrics.
Sentinel-3
Sentinel-3 carries the Ocean and Land Colour Imager (OLCI) and the Sea and Land Surface Temperature Radiometer (SLSTR). OLCI has 21 spectral bands ranging over the VNIR (visual and near-infrared) and SWIR (short-wave infrared) wavelengths. Sentinel-3 data is used to produce Leaf Area Index (LAI), Fraction of Green Vegetation Cover (FCOVER), and Burnt Area products.
PRISMA
PRISMA is a hyperspectral imaging satellite launched in March 2019 developed and operated by the Italian Space Agency (ASI). The mission carries two instruments, the Hyperspectral Camera (HYC) (a prism spectrometer with 239 channels) and the Panchromatic Camera (PAN). The hyperspectral imagery provided by HYC covers 239 wavelength ranges in VIS, NIR and SWIR, meaning it can be used for identification of highly specific spectral indices, such as those for individual mineral signatures.
WorldView-2
WorldView-2 (WV-2) is a commercial imaging satellite owned by Maxar Technologies, launched in October 2009. It aims to provide high spatial resolution imaging of the Earth’s surface. WV-2 carries a single instrument, the WorldView-110 camera (WV110), a nine channel VIS/NIR radiometer and high resolution optical imager. The multispectral imagery provided by WV110 can be used for the development of a number of spectral products, including WV-II and the WorldView Soil Index (WV-SI).
HYSIS
The Hyperspectral Imaging Satellite (HYSIS) is an ISRO EO mission launched in November 2018. The satellite carries two pushbroom hyperspectral imagers, operating in the VNIR and SWIR regions. The VNIR sensor covers 60 continuous bands, and the SWIR sensor covers 256, with both instruments providing 30 m spatial resolution.
Landsat-7
Landsat-7, launched in April 1999 and jointly operated by NASA and USGS, aims to enhance the multispectral imagery of previous Landsat missions. To achieve this, the satellite carries the Enhanced Thematic Mapper Plus (ETM+), a whiskbroom scanning radiometer that images in eight bands – four VNIR, two SWIR, one TIR and one PAN. Landsat-7’s multispectral data contributes to a variety of spectral indices including NBR, NDSI, and EVI, accessible here.
CHIME
The Copernicus Hyperspectral Imaging Mission for the Environment (CHIME) is a planned ESA EO mission that will provide hyperspectral data for environmental and resource monitoring. The mission will consist of two satellites, each carrying a single hyperspectral imager (HSI), acquiring imagery across more than 200 bands with a spectral range of 420 - 2500 nm. This high spectral resolution data will enable the derivation of spectral indices requiring highly precise identification of different spectral signatures. CHIME data will be used for the creation of NDVI, LAI, FCOVER and soil moisture products.
MULA
The Multispectral Unit for Land Assessment (MULA) is a planned Filipino medium resolution optical imagery satellite. The satellites will carry the Surrey Satellite Technology Ltd. (SSTL) TrueColour multispectral optical imager, imaging in nine bands with a spectral range of 433 - 907 nm and providing a 5 m ground sampling distance (GSD). As the mission covers the NIR and VIS regions of the electromagnetic spectrum, collected data can be used for creation of crop health and vegetation density spectral products, such as NDVI and EVI.
Overall, spectral indices should be understood as a general analytical approach applicable across many EO missions, rather than as products tied to specific satellites. The choice of mission depends on the requirements of the application: medium-resolution public missions such as Landsat and Sentinel-2 are often preferred for broad operational use; coarser-resolution systems such as MODIS and Sentinel-3 are valuable for frequent, large-area monitoring; commercial missions such as WorldView-2 provide finer spatial detail; and hyperspectral missions such as PRISMA, HySIS and CHIME enable more specialised analyses based on narrow spectral features.
References
1) Kablamo, “What are spectral indices in remote sensing?”, URL: https://engineering.kablamo.com.au/posts/what-are-spectral-indices/
2) Applied Remote Sensing Training Program, “Spectral Indices for Land and Aquatic Applications”, URL: https://appliedsciences.nasa.gov/sites/default/files/2023-10/Spectral_Indices_Part1.pdf
3) OnGeo Intelligence, “Spectral Bands: A Guide to Popular Index Formulas”, URL: https://ongeo-intelligence.com/blog/spectral-bands-list-of-formulas-for-the-most-popular-spectral-indices
4) GeoAwesome, “An Overview of Indices in Remote Sensing”, URL: https://geoawesome.com/an-overview-of-indices-in-remote-sensing/
5) USGS, “Landsat Normalised Difference Vegetation Index”, URL: https://www.usgs.gov/landsat-missions/landsat-normalized-difference-vegetation-index
6) USGS, “Landsat Enhanced Vegetation Index”, URL: https://www.usgs.gov/landsat-missions/landsat-enhanced-vegetation-index
7) NV5, “Miscellaneous Indices Background”, URL: https://www.nv5geospatialsoftware.com/docs/backgroundotherindices.html
8) USGS, “Landsat Normalised Burn Ratio”, URL: https://www.usgs.gov/landsat-missions/landsat-normalized-burn-ratio
9) NV5, “Geology Indices Background”, URL: https://www.nv5geospatialsoftware.com/docs/backgroundgeologyindices.html#WVSI
10) USGS, “Landsat Normalised Difference Snow Index”, URL: https://www.usgs.gov/landsat-missions/normalized-difference-snow-index
11) P. Balaji, Medium, “SAR Series Part6: SAR Vegetation Indices — quick reference sheet”, URL: https://medium.com/@preet.balaji20/sar-series-part6-sar-vegetation-indices-quick-reference-sheet-d71a5d63da25
12) K. Solymosi et al., “The Development of Vegetation Indices: a Short Overview”, URL: https://scispace.com/pdf/the-progression-of-vegetation-indices-a-short-overview-bgfd5oh8uo.pdf
13) GeoAwesome, “An Overview of Indices in Remote Sensing”, URL: https://geoawesome.com/an-overview-of-indices-in-remote-sensing/
14) Awesome Spectral Indices, “Spectral Indices- Expressions”, URL: https://awesome-ee-spectral-indices.readthedocs.io/en/latest/index.html#expressions
15) X.Giao et al., “Optical–Biophysical Relationships of Vegetation Spectra without Background Contamination”, URL: https://www.sciencedirect.com/science/article/pii/S0034425700001504?casa_token=v2qqQrukka4AAAAA:k48ccjWq9YLgawh_oaOl_ueY2QjwuDraKluK-s258GM9UUnssaXaE5kSWb1nK5PAHFq0UCYYvg
16) Geopera, “WorldView Legion Spectral Indices,” URL: https://docs.geopera.com/en/docs/spectral-indices/worldview-legion