Other Space Activities

Hyperspectral Imaging

Aug 14, 2023

Instrument Types

Hyperspectral imaging (HSI) is a process used to obtain high spectral resolution imagery by dividing light into many narrow, contiguous spectral bands across the electromagnetic (EM) spectrum, typically between visible and infrared wavelengths. Different features on Earth’s surface reflect, absorb, scatter and emit light at specific wavelengths, which creates a unique and characteristic spectral fingerprint that hyperspectral imagers are able to identify. HSI has greatly enhanced our perception of the Earth’s surface and its features, other planets, space domain awareness and more, with greater precision than from multispectral imaging. 1) 2)

Figure 1: HSI principles diagram (Image credit: Lincoln Laboratory Journal, 4))

Multispectral imagers capture EM radiation in a small number (typically 4 - 36) of broad spectral bands, whereas hyperspectral imagers collect measurements from a much larger number of spectral bands (up to hundreds), which are adjacent to one another and cover narrow wavebands(typically less than 10 nm). Hyperspectral imagers typically feature spatial resolutions of about 30 m, and their contiguous spectral coverage permits the creation of spectral signatures with no wavelength omission. Traditional optical imagers assign to each pixel a combination of RGB (red, green, blue) light from a scene. Meanwhile, hyperspectral imagers assign an entire spectrum of light for each pixel, providing detailed information into the object’s physical properties and chemical composition. 1) 2) 3)

High resolution radiance measurements obtained by hyperspectral imagers enable the creation of continuous spectra of a target, which are compared against libraries of known material spectra to identify surface materials.

Where surface materials can be identified by comparing these measured spectra with libraries of known material spectra, like the United States Geological Survey (USGS) Spectral Library, and Digital Earth Australia’s National Spectral Database (NSD).

Hyperspectral imaging has demonstrated effectiveness for mapping minerals and soils, vegetation species, composition and health, shallow coastal and coral reef habitats, and water quality. HSI often focuses on a discrete spectral region for specific applications. HSI allows us to see closer to a target not in terms of spatial but spectral resolution, meaning we can see its chemical composition and characteristics as opposed to seeing closer spatially. 1) 2)

Table 1: Hyperspectral imaging spectral regions and applications 1) 5) 6)
Spectral region	Spectral range (nm)	Optimal observations
Thermal Infrared (TIR)	8000 - 15000	Heat sources, land and sea surface temperatures, geothermal mapping, thermal surveys
Infrared (IR)	6000 - 7000	Water vapour, soil moisture, cloud cover, thermography, forest fires and hotspots
Mid-wave Infrared (MIR)	3000 - 5000	Mineral and soil mapping, sea surface temperature, ice formations, geothermal and volcanic activity,
Short-wave Infrared (SWIR)	1100 - 3000	Vegetation mapping, dynamics and physiology, cloud and rock type
NIR (Near Infrared)	700 - 1100	Vegetation vigour, crop and soil moisture, rock and mineral type
Visible	400 - 700	Shallow coastal and coral reef bathymetry, vegetation type, land cover, urban development, ocean colour
Ultraviolet (UV)	100 - 400	Ozone concentration, coral reef health, aerosol distribution, pollution

Spaceborne HSI has been primarily operated by government-led space missions, however the last few years has seen the emergence of startup HSI space companies launching their own constellations of hyperspectral satellites. Companies like Pixxel, who have launched two HSI satellites in recent years with a constellation of 24 planned; Orbital Sidekick, currently launching its constellation of six GHOSt (Global Hyperspectral Observation Satellite) satellites alongside its their demonstrator, Aurora; Satellogic’s NewSat constellation; and undoubtedly more to come. 2)

A hyperspectral image is composed of a stack of images, with each image corresponding to one spectral band, represented as a 3D ‘spectral cube’. These images have two spatial dimensions and one spectral dimension, shown as the stacked image layers. The dimensions of the cubes are derived from the satellite swath widths across-track, along-track, and wavelength range. 2) 7)

Figure 2: AVIRISng (Airborne Visible Infrared Imaging Spectrometer next generation) hyperspectral image cube of Mount Vesuvius, Italy (Image credit: NASA/JPL)

AVIRIS is regarded as the first operational hyperspectral instrument, and was flown from 1986 as an airborne hyperspectral imager to identify, measure and monitor constituents of the Earth's surface and atmosphere. AVIRIS was the first Earth-viewing imaging spectrometer to observe the entire solar reflectance spectrum in contiguous spectral bands. 4) 8)

Hyperspectral Imaging vs Imaging Spectroscopy

Hyperspectral imaging and imaging spectroscopy are two similar remote sensing methods, both used to obtain information of the Earth’s surface in great detail. The terms are often used interchangeably, but there are subtle differences. 3)

‘Imaging spectroscopy’ is a more comprehensive term that encompasses HSI, and is the process of gathering spectral information from each pixel of an image. Although imaging spectroscopy includes HSI, it also covers multispectral imagery which collects information over fewer, wider and discrete spectral bands. Imaging spectroscopy focuses on the in-depth analysis of spectral information, which can be gathered from singular point sources using a spectrometer and spectrum analyser. HSI only refers to measurements made over many narrow and contiguous spectral bands. 3) 10)

Example Products

Due to HSI providing an abundance of information about a scene in hundreds of spectral bands, imagery becomes inherently difficult to visualise. In order to analyse all the information from all spectral bands, elaborate image processing is required. Common data formats used by HSI are spectral cubes and reflectance curves.

Spectral Cubes

From successive cross-track scans by an air or spaceborne sensor, a 3D cube can be created by stacking 2D spatial images of different wavelengths on top of each other. Figure 3 (b) illustrates this with a spatial x-y plane and spectral z axis, which delineates high reflectance as red and low reflectance as blue. 4)

Figure 3: Hyperspectral data cube construction (Image credit: Lincoln Laboratory Journal, 4))

Reflectance Curves

For each wavelength band measured in a hyperspectral image, a spectral reflectance image is produced. This means that an entire reflectance curve can be plotted for each pixel in the image. Reflectance curves allow in-depth scene analysis pixel-by-pixel, and are often paired alongside a spectral cube for analyses of areas of interest. Reflectance curves can be imagined as taking a slice out of a spectral cube in the z (spectral) axis, isolating reflectance information for a specific pixel or area of the 2D image.

Figure 5 shows an exploded view of a hyperspectral data cube, with a 2D single-band image taken from the 3D stack, and a reflectance curve of a single pixel in the image.

Figure 5: Exploded view of a hyperspectral data cube (Image credit: Duo Wang, 9))

ISS: EMIT (Earth surface Mineral dust source InvesTigation)

EMIT is a hyperspectral imaging instrument onboard the ISS (International Space Station), with an objective to determine mineral compositions from natural sources of dust aerosols around the world and model their climatic effects. The NASA (National Aeronautics and Space Administration) led mission launched in July 2022 carries one of the most sophisticated Earth-facing spectrometers in space, that can acquire over 100,000 unique spectral signatures of dust source minerals every second.

EO-1 (Earth Observing-1)

The first hyperspectral imaging satellite, EO-1, was a NASA Goddard Space Flight Centre (GSFC) technology demonstration mission that carried Hyperion, a pushbroom hyperspectral imager. The imager was capable of resolving 220 spectral bands from VNIR (Visible and Near Infrared) to SWIR at a 30 m spatial resolution, in order to identify surface minerals, vegetation type, forest vigour, and volcanic activity on the Earth’s surface. With Hyperion, EO-1 proved the value of hyperspectral imagery to the scientific community and preceded a new frontier in satellite imagery.

EnMAP (Environmental Mapping and Analysis Program)

EnMAP is a German minisatellite mission carrying a dedicated hyperspectral imager to observe biophysical, biochemical and geochemical variables, a wide range of ecosystem parameters encompassing agriculture forestry, soil/geological environments, coastal zones and inland waters. EnMAP’s pushbroom HSI sensor makes observations with 228 spectral bands in both the VNIR and SWIR spectrum to produce high quality, high spatial and spectral resolution data used to model and develop understanding of biospheric and geospheric processes in a range of applications.

Newsat (Aleph-1) Constellation

Newsat (ÑuSat) is a constellation of 38 identical microsatellites operated by the Argentine space company, Satellogic, providing imagery to customers for applications in agriculture and food production, oil and gas monitoring, climate and resource monitoring, disaster response, and infrastructure monitoring. The satellites carry a hyperspectral imager operating in the VNIR spectrum, with 29 spectral bands and a 25 m spatial resolution, which paired with the panchromatic and multispectral payloads, can collect up to 1,000 scenes per day.

PACE (Plankton, Aerosol, Cloud, ocean Ecosystem) Mission

PACE is a NASA mission approved for launch in January 2024 with the intent to gather data for ocean ecology and global biogeochemistry with OCI (Ocean Color Instrument), a hyperspectral instrument operating from the ultraviolet (UV) to SWIR range. Through the analysis of light spectra reflected from the ocean surface, OCI will uncover essential marine biological elements and provide valuable insights for various sectors relying on water quality, fisheries, and food security.

PRISMA (Hyperspectral Precursor and Application Mission)

PRISMA, a hyperspectral minisatellite mission led by ASI (Italian Space Agency), has the objective to deliver global hyperspectral imagery from 2019 to 2024.

Resurs-P (Resurs-Prospective)

Resurs-P is a Roscosmos multi-satellite mission with the objective to obtain multipurpose hyperspectral, multispectral, and panchromatic imagery from 2013 to 2028.

GHOSt (Global Hyperspectral Observation Satellite)

GHOSt, by start-up Orbital Sidekick Inc. (OSK), is a constellation of the highest resolution commercial hyperspectral satellites, operating from 2023 to at least 2027.

References

1) “Hyperspectral Imagers,” CEOS Earth Observation Handbook 2014, URL: http://www.eohandbook.com/eohb2014/sat_earth_obs_hyperspectral.html

2) David Hodes, “Hyperspectral Imaging Attracts a Host of Space Startups,” Via Satellite, 23 May 2023, URL: https://interactive.satellitetoday.com/via/june-2023/hyperspectral-imaging-attracts-a-host-of-space-startups/

3) Gary Shaw, Hsiao Burke, “Spectral Imaging for Remote Sensing,” Lincoln Laboratory Journal, Volume 14, 2003, URL: https://courses.cs.washington.edu/courses/cse591n/07sp/papers/Shaw2003.pdf

4) Anshu Miglani, “Hyperspectral Remote Sensing -an Overview,” Geospatial World, 12 July 2010, URL: https://www.geospatialworld.net/article/hyperspectral-remote-sensing-an-overview/

5) “Observing in Infrared,” NASA Earth Observatory, 4 March 2014, URL: https://earthobservatory.nasa.gov/features/FalseColor/page5.php

6) “Earth Observation Satellite and Mechanism of Observation,” JAXA, URL: https://www.satnavi.jaxa.jp/en/satellite-knowledge/whats-eosatellite/observation/index.html

7) “Hyperspectral image ‘data cube’”, European Space Agency, 8 April 2014, URL: https://www.esa.int/ESA_Multimedia/Images/2014/04/Hyperspectral_image_data_cube

8) “AVIRIS (Airborne Visible/Infrared Imaging Spectrometer),” eoPortal, 15 June 2014, URL: https://www.eoportal.org/other-space-activities/aviris

9) Wang, D.; Chen, Z.; Zhang, X.; Fu, T.; OuYang, R.; Bi, G.; Jin, L.; Wang, X. “A High Optical Throughput Spectral Imaging Technique Using Broadband Filters,” Sensors 2020, Volume 20, Article 4387, URL: https://www.mdpi.com/1424-8220/20/16/4387

10) “Hyperspectral and Multispectral Imaging,” Edmund optics, URL: https://www.edmundoptics.co.uk/knowledge-center/application-notes/imaging/hyperspectral-and-multispectral-imaging/

The information compiled and edited in this article was provided by Herbert J. Kramer from his documentation of: "Observation of the Earth and Its Environment: Survey of Missions and Sensors" (Springer Verlag) as well as many other sources after the publication of the 4th edition in 2002. - Comments and corrections to this article are always welcome for further updates (eoportal@symbios.space).