Spectral Indices

Spectral index is the ratio of broadband spectral bands or as the normalized differences between two bands.

There are various hyperspectral sensors to capture hyperspectral data whose band centers are usually slightly different. Thus defining bands gives you the freedom to apply spectral indices to a wide range of sensors. The Band Definition section provide details of band definitions used in the Hyperspectral Imaging Library for Image Processing Toolbox™.

Spectral indices characterize the specific features of interest of a target by exploiting its biophysical and chemical properties. These features of interest enable you to identify plant, water, soil, and various forms of built-up regions such as road, house, railway track, and parking lot. For more information on supported spectral indices, see List of Supported Spectral Indices section.

Band Definition

Hyperspectral Imaging Library for Image Processing Toolbox uses various band definitions to compute spectral indices. The toolbox selects the nearest wavelength to the center of each band available in input hyperspectral data.

Depending on the range of the wavelengths for a band, the band definition can be of two types.

• Broadband — Bands generally have wider wavelength ranges.

• Narrowband — Bands generally have narrow wavelength ranges.

This table lists the broadband definitions used in the toolbox.

Band	Minimum	Center	Maximum
B	400 nm	470 nm	500 nm
G	500 nm	550 nm	600 nm
R	600 nm	650 nm	700 nm
NIR	760 nm	860 nm	960 nm
SWIR1	1550 nm	1650 nm	1750 nm
SWIR2	2080 nm	2220 nm	2350 nm

This table lists the narrowband definitions used in the toolbox.

Band	Minimum	Center	Maximum
$$B_{531}$$	525 nm	531 nm	550 nm
$$B_{550}$$	540 nm	550 nm	560 nm
$$B_{570}$$	560 nm	570 nm	575 nm
$$B_{670}$$	650 nm	670 nm	690 nm
$$B_{700}$$	680 nm	700 nm	730 nm
$$B_{795}$$	720 nm	795 nm	800 nm
$$B_{800}$$	780 nm	800 nm	865 nm
$$B_{819}$$	815 nm	819 nm	824 nm
$$B_{990}$$	830 nm	990 nm	995 nm
$$B_{1510}$$	1500 nm	1510 nm	1515 nm
$$B_{1599}$$	1590 nm	1599 nm	1620 nm
$$B_{1680}$$	1670 nm	1680 nm	1690 nm
$$B_{2000}$$	1980 nm	2000 nm	2040 nm
$$B_{2100}$$	2085 nm	2100 nm	2110 nm
$$B_{2200}$$	2170 nm	2200 nm	2220 nm

List of Supported Spectral Indices

Hyperspectral Imaging Library for Image Processing Toolbox supports various spectral indices used to identify vegetation, minerals, burned areas, and built up regions. This table lists the hyperspectral indices supported by the spectralIndices function. The equations of these indices uses the band definitions as specified in Band Definition. You can compute the indices CMR, EVI, GVI, MNDWI, NBR, NDBI, NDVI, OSAVI, and SR, which use broadband band definitions, for both narrowband hyperspectral images and broadband multispectral images. You can compute the indices CAI, MCARI, MTVI, MSI, NDMI, NDNI, and PRI, which use narrowband band definitions, for only narrowband hyperspectral images.

Index Name	Equation	Description
Cellulose absorption index (CAI)	$$CAI=0.5\left(B_{2000}+B_{2200}\right)-B_{2100}$$	CAI identifies dried plant materials relative to the cellulose sensitive wavelengths in the range 2000 nm to 2200 nm. Use this index to monitor crop residue, plant health, and fuel conditions in an ecosystem. The index value varies in the range (-3, 4).
Clay minerals ratio (CMR)	$$CMR=\frac{SWIR1}{SWIR2}$$	CMR identifies hydrothermally altered rocks containing clay and alunite. Use this index to map minerals in rock surfaces.
Enhanced vegetation index (EVI)	$$EVI=\frac{\left(NIR-R\right)}{\left(NIR+6\ast R-7.5\ast B+1\right)}$$	EVI identifies vegetation regions with a high leaf area index. It uses blue reflectance to correct the soil background by including atmospheric influences. For vegetation pixels, the index value varies in the range (0, 1).
Green vegetation index (GVI)	$$\begin{array}{l}GVI=\left(-0.2848\ast B\right)+\left(-0.2435\ast G\right)+\left(-0.5436\ast R\right)+\mathrm{...}\\ \left(0.7243\ast NIR\right)+\left(0.0840\ast SWIR1\right)+\left(-0.1800\ast SWIR2\right)\end{array}$$	GVI identifies green vegetation by reducing the background soil effect. The index value varies in the range (-1, 1).
Modified chlorophyll absorption ratio index (MCARI)	$$MCARI=\left(B_{700}-B_{670}\right)-\left(0.2\ast \left(B_{700}-B_{550}\right)\right)\ast \left(\frac{B_{700}}{B_{670}}\right)$$	MCARI identifies the vegetation regions that contain chlorophyll by minimizing the combined effects of soil and non-photosynthetic surfaces.
Modified triangular vegetation index (MTVI)	$$MTVI=1.2\left(\left(1.2\ast \left(B_{800}-B_{550}\right)\right)-\left(2.5\ast \left(B_{670}-B_{550}\right)\right)\right)$$	MTVI identifies vegetation regions. This index includes the 800 nm wavelength, which is influenced by changes in leaf and canopy structure.
Modified normalized difference water index (MNDWI)	$$MNDWI=\frac{G-SWIR1}{G+SWIR1}$$	MNDWI identifies open water surfaces, reducing background noise from soil, vegetation, and built-up areas.
Moisture stress index (MSI)	$$MSI=\frac{B_{1599}}{B_{819}}$$	MSI maps the level of leaf water content in vegetation canopies. The index value varies in the range (0, 3).
Normalized burn ratio (NBR)	$$NBR=\frac{NIR-SWIR2}{NIR+SWIR2}$$	NBR identifies burned areas in larger fire zones. You can obtain a burn sensitive image by subtracting the pre-fire and post-fire NBR images.
Normalized difference built-up index (NDBI)	$$NDBI=\frac{SWIR1-NIR}{SWIR1+NIR}$$	NDBI identifies urban areas with high reflectance in the SWIR region as compared to the NIR region.
Normalized difference mud index (NDMI)	$$NDMI=\frac{B_{795}-B_{990}}{B_{795}+B_{990}}$$	NDMI identifies shallow water or muddy surfaces.
Normalized difference nitrogen index (NDNI)	$$NDNI=\frac{\log \left(\frac{1}{B_{1510}}\right)-\log \left(\frac{1}{B_{1680}}\right)}{\log \left(\frac{1}{B_{1510}}\right)+\log \left(\frac{1}{B_{1680}}\right)}$$	NDNI maps the amount of nitrogen content in vegetation canopies. Because NDNI is a logarithmic index, use data values in the range 0 to 1 for accurate results.
Normalized difference vegetation index (NDVI)	$$NDVI=\frac{NIR-R}{NIR+R},$$	NDVI identifies vegetation canopies. The index value varies in the range (-1, 1). A value close to 1 indicates healthy vegetation, 0 indicates unhealthy vegetation, and -1 indicates no vegetation.
Optimized soil adjusted vegetation index (OSAVI)	$$OSAVI=\frac{\left(NIR-R\right)}{\left(NIR+R+0.16\right)}$$	OSAVI identifies sparse vegetation, where soil is visible through the canopy.
Photochemical reflectance index (PRI)	$$PRI=\frac{B_{531}-B_{570}}{B_{531}+B_{570}}$$	PRI maps the photosynthetic efficiency of a region by detecting the changes of carotenoid pigments in live foliage. The index value varies in the range (-1, 1).
Simple ratio (SR)	$$SR=\frac{NIR}{R}$$	SR finds the ratio of vegetation and chlorophyll absorption wavelength features.

References

[1] Bannari, A., D. Morin, F. Bonn, and A. R. Huete. “A Review of Vegetation Indices.” Remote Sensing Reviews 13, no. 1–2 (August 1995): 95–120. https://doi.org/10.1080/02757259509532298.

[2] Xue, Jinru, and Baofeng Su. “Significant Remote Sensing Vegetation Indices: A Review of Developments and Applications.” Journal of Sensors 2017 (2017): 1–17. https://doi.org/10.1155/2017/1353691.

[3] Haboudane, D. “Hyperspectral Vegetation Indices and Novel Algorithms for Predicting Green LAI of Crop Canopies: Modeling and Validation in the Context of Precision Agriculture.” Remote Sensing of Environment 90, no. 3 (April 15, 2004): 337–52. https://doi.org/10.1016/j.rse.2003.12.013.

[4] Thenkabail, Prasad S., John G. Lyon, and Alfredo Huete, eds. Hyperspectral Indices and Image Classifications for Agriculture and Vegetation. Boca Raton: CRC Press, 2018.