Blog of Many Colours

Image featuring the sister cities of Sault Sainte Marie, Ontario, and Sault Sainte Marie, Michigan. ESA’s Proba satellite acquired this image on 11 August 2006 with its Compact High Resolution Imaging Spectrometer (CHRIS), designed to acquire hyperspectral images with a spatial resolution of 18 metres across an area of 14 kilometres. Data courtesy of SSTL through ESA.

Image featuring the sister cities of Sault Sainte Marie, Ontario, and Sault Sainte Marie, Michigan. ESA’s Proba satellite acquired this image on 11 August 2006 with its Compact High Resolution Imaging Spectrometer (CHRIS), designed to acquire hyperspectral images with a spatial resolution of 18 metres across an area of 14 kilometres. Data courtesy of SSTL through ESA.

The aspect of art at school that really stuck with me was learning about the main colours of the rainbow and how they fit together – like with like, such as yellow, green, blue, and like with unlike such as shades of green with a fleck of red to put spark into a picture. Based on these ideas, when I was a teenager I used to construct geometric mandalas coloured in with gouache. As I began studying remote sensing, it seemed natural that hyperspectral imaging would hold a special fascination.

The term Hyperspectral Imaging was coined by Goetz in 1985 and is defined as ‘the acquisition of images in hundreds of contiguous, registered, spectral bands such that for each pixel a radiance spectrum can be derived.’ Put simply, whereas a picture is made using three colour components for television (red, green and blue), for hyperspectral imaging the spectrum is split into many, sometimes hundreds, of different grades of colour for each part of the image. The term made its way into scientific language by way of the intelligence communities – the military became interested in it as it offered them the ability to tell plastic decoys from real metal tanks, as well as an object’s precise colour.

When the first field spectral measurements were conducted in the early 1970s, technology was not advanced enough for it to be put into operation. However, developments in electronics, computing and software throughout the 1980s and into the 1990s, brought the hyperspectral imaging to the EO community.

A series of parallel hardware development began in the 1980’s, such as at NASA JPL with the Airborne Imaging Spectrometer (AIS) in 1983, followed by AVIRIS (Airborne Visible/IR Imaging Spectrometer). The AVIRIS sensor was first flown in 1987 on a NASA aircraft at 20km altitude and to this day, it is still a key provider of high-quality HS data for the scientific community.

The hardware advances were matched by improvements in software capabilities, with the development of the iconic image cube method of handling this type of data, by PhD students Joe Boardman and Kathryn Kierein-Young, from the University of Colorado. Spectral libraries have been amassed for over 2,400 natural and artificial materials, to enable them to be identified. The most famous is the ASTER spectral library which includes inputs from Johns Hopkins University (JHU) Spectral Library, the Jet Propulsion Laboratory (JPL) Spectral Library, and the United States Geological Survey (USGS – Reston) Spectral Library.

Hyperspectral imaging was primarily developed for the mapping of soils and rock types; and the spectra of these are rich in character. Taking regions from the contiguous spectrum makes it possible to identify surface materials by reflectance or emission and also allows precise atmospheric correction which can only be approximated if you are using discrete, wide colour bands. The shape of the reflectance or emittance spectrum yields information about grain size, abundance and composition as well as the biochemistry of vegetation, such as the concentration of chlorophyll and other pigments and life forms in water bodies.

Earth observation hyperspectral imaging really began with NASA’s Earth Observing-1 Mission (EO-1) launched in 2000, with the Hyperion imager on board that has 200 wavelengths. Since then, various other missions have been launched such as the Compact High Resolution Imaging Spectrometer (CHRIS) on the Proba-1 satellite also in 2001, with 63 spectral bands; or the Infrared Atmospheric Sounding Interferometer (IASI) on board the MetOp series of Meteorological satellites whose first version was launched in 2006.

The coming years for hyperspectral imaging looks exciting with a whole series of planned missions including the Italian PRISMA (PRecursore IperSpettrale della Missione Applicativa), German EnMAP (Environmental Mapping and Analysis Program), NASA’s HyspIRI (Hyperspectral Infrared Imager), and JAXA’s (Japan Aerospace Exploration Agency) Hyperspectral Imager Suite (HIUSI).

So for me, and anyone with the same fascination, the future really will be filled with many colours!


Blog written by Dr Louisa Reynolds