Hyperspectral Imaging for finer surface characterization
- Arpit Shah
- May 31, 2022
- 15 min read
Updated: Apr 14
Introduction
It is the investigative nature of Satellite Imagery Analytics that, I suppose, draws enthusiasts to this field - the possibility to extract meaningful insights by processing remotely-sensed data acquired by spaceborne instruments hundreds of kilometers above the earth's surface.
In this post, you will get familiar with a relatively-novel technique of acquiring Earth observation data known as Hyperspectral Imaging (technically, Imaging Spectroscopy) through a workflow demonstration video. At the outset, I will cover Remote Sensing fundamentals and the characteristics of the more commonly-used Multispectral and Radar Imaging techniques which will lend itself well to your understanding of where the utility of Hyperspectral Imaging lies.
Hyperlinks to Sections
Credits: Much thanks to EO College's MOOC series and the webinars by NASA ARSET which helped me to get acquainted to the world of Hyperspectral Remote Sensing
Remote Sensing Background
Objects, materials and surfaces on earth can be imaged using two ways-
Passively - by capturing reflected Solar Radiation
Actively - by transmitting illumination through a satellite sensor and capturing the backscatter
Figure 1 is a natural-color rendition of a passively-acquired Optical Imagery dataset - the dataset is formed when the Imaging instrument captures reflected Solar Radiation (Sunlight constitutes energy waves from the visible light, infrared and ultraviolet portions of the Electromagnetic Spectrum) across one or more Spectral 'bands' i.e. ranges of wavelength. If the Imaging instrument captures reflected Solar Radiation across more than three Spectral bands (see its types here), then the dataset can be called a Multispectral Imagery.
Different types of surfaces respond differently to Solar Radiation. Some energy wavelengths would reflect upon interacting with the object, while the remaining may be absorbed or transmitted.
Besides the characteristics of the wave itself, the extent of reflectance is influenced by-
Reflective properties of the object (whether the surface is rough or smooth),
Geometric effect of reflection (angle of illumination and angle of reflection), and
Biogeochemical characteristics of the object (moisture content, mineral properties, size etc.)
That being said, even before an energy wave interacts with an object, it has to pass through the atmosphere (subject to the imaging instrument being spaceborne) where it may be influenced by clouds and gases (the wavelengths that constitute Solar Radiation are particularly susceptible to these). The same is applicable after surface interaction when the reflected wave exits the atmosphere
As can be gathered from the example in Figure 3, the rate of reflection varies considerably when Soil, Vegetation and Water are exposed to the various wavelengths that constitute Solar Radiation. For example, Water has 0% reflection, Soil hovers around the 30-35% mark, and Vegetation has the highest reflection of them all touching 50% when interacting with Near Infrared (band 4) energy waves. This is in contrast to how these three material types respond to the other wavelength ranges, be it within Visible Light (band 1, 2 and 3) or within Intermediate Infrared (band 5, 7). As a result, through the use and processing of Multispectral Satellite Imagery datasets, one can exploit these varying solar energy reflectance characteristics of different objects to detect, delineate and classify distinct geologic features such as farmlands, lakes and forests.
Sunlight, the source of illumination for Multispectral Satellite Imagery acquisition for Earth Observation, has some drawbacks - the most obvious being the imagery cannot be acquired during night-time. Even during the day, Solar Radiation can be obstructed by clouds, aerosols and other atmospheric constituents. Also, while the three surface types in Figure 3 are broad and distinct categories, if one were to attempt more granular classifications - such as distinguishing palm trees from coconut trees in a region - the variability of reflectance of these two objects to the same band of solar wavelength which a multispectral imager typically acquires may not be very large, which would hinder us from accurately delineating one from another in regional-scale Earth Observation studies.
This aspect serves to highlight the utility of utilizing modes of imaging that have a high Spectral Resolution i.e. which capture reflected energy from the Earth's surface across more and narrower Spectral bands - this would help to perform finer classification of surface materials, something which spaceborne Multispectral imagers are not particularly good at, as of today.
Besides Spectral Resolution, there are other types of Resolution as well which are important considerations by themselves for any research study which involves the use and processing of remotely-sensed data. Here's a short summary of the four resolutions-
Spectral Resolution - range of energy wavelengths which the sensor is sensitive to - for example, Cartosat-1 satellite captures reflectances in between 500 nm and 850 nm i.e. it has a single spectral band with a width of 350 nanometres
Spatial Resolution - measure of the smallest surface feature which the sensor can detect i.e. it is a dimension of the pixel size - for example, the dimensions of a pixel in R, G, and B bands of Sentinel-2 satellites are 10 x 10 metres, hence the spatial resolution for these is 10 metres
Temporal Resolution - refers to how quickly the sensor can image the same object again i.e. the orbital cycle of the satellite - for example, Sentinel-5P satellite has an orbital cycle of 16 days which means that the same spot on the earth's surface can be imaged by it once every 16 days
Radiometric Resolution - the quantity of energy signal a sensor can perceive reliably i.e. with limited noise, and on a consistent basis. For example, the Radiometric Resolution of OLI-2 Multispectral instrument onboard NASA's Landsat 9 satellite is 14 bits sensitive which means that the sensor has 2^14 i.e. 16384 potential digital values (0 to 16383) that it can use to store reflectance information in a pixel
More information on these can be accessed here. Do note that these types of resolution are not mutually-exclusive - having more of one often results in a trade-off - i.e. having less of another.
Synthetic Aperture Radar (SAR), or simply Radar Satellite Imagery, often serves as a useful alternative in workflows where Multispectral Imagery is found wanting. Besides, it can be utilized in a complementary and/or a supplementary way to validate the output derived from the processing of Multispectral Imagery as well.
Radar Satellites acquire data through an active imaging sensor which transmits microwaves on Earth and captures its reflectance, technically known as backscatter - when energy rays are reflected in the same direction as its source. Sentinel-1, a popular Earth Observation radar satellite constellation transmits pulses of C-band microwaves (1 nm - 180 nm).
One major advantage of Radar satellites over Multispectral satellites (and even Hyperspectral satellites) is that, due to the presence of an active illumination sensor, they can acquire imagery during night-time.
An important distinction for you to understand is that while the mode of imaging in Multispectral or Hyperspectral Spaceborne instruments for Earth Observation purposes is passive i.e. reflected Solar Radiation, similar acquisitions made using airborne or surface instruments such as drones, aircrafts (AVIRIS NG), geodetic equipment can involve the use of active sensors. Besides, Astronomical telescopes such as Chandra X-ray Observatory (Spaceborne instrument purposed for Outer Space research) also make use of active modes of illumination.
Another advantage is that microwaves are unaffected by cloud cover and aerosols by virtue of having a longer wavelength, which is a very useful enabler for performing accurate Earth Observation studies. Moreover, surface materials interact with microwaves in a different way than they do with the waves that constitute sunlight - a characteristic that can be exploited in a standalone, supplementary or in a complementary way through the use of Radar Satellite Imagery as I had indicated before.
A technical comparison table of the characteristics, advantages and disadvantages of Multispectral and Radar Remote Sensing can be accessed here and some of the applications involving either or both of them can be accessed here.
So if Multispectral and Radar Remote Sensing techniques have different strengths and cater to a wide variety of often-mutually-exclusive applications, then what is Hyperspectral Remote Sensing and where does its utility lie?.
The next section will be devoted to addressing this exact query that many of you may be having.
Hyperspectral Imaging
As described in the introductory course on this topic by EO College-
Imaging spectroscopy refers to imaging sensors measuring the spectrum of Solar Radiation reflected by Earth surface materials in many contiguous wavebands, on the ground as well as air or spaceborne.../...there are up to hundreds of reflectance bands that allow detection and quantification of materials based on the shape of the spectral curve.
Contrary to what you may perceive, Hyperspectral Imaging is not a new technology. In fact, the first Imaging Spectrometer (instrument which acquires hyperspectral data) became operational as far back as 1982. However, such instruments were installed onboard research aircrafts to capture images over a small region and only at select locations which would have led to only a limited number of researchers being able to access and utilize the data. Only after spectrometers were installed onboard Satellites in the 2000s that the technology and its usage became more mainstream.
As with any upcoming technology where the newer versions iron out the flaws that accompany the earlier versions as well as benefit from the growth of the overall ecosystem around the technology, so was the case with Imaging Spectroscopy. Since 2019, multiple Spaceborne Hyperspectral sensors have been launched and newer algorithms have been developed which promise to usher in a new era in the field of Earth Observation to serve mankind's quest to have a deeper understanding of the geochemical, biochemical and biophysical properties of our planet's surface and the atmosphere.
How is Hyperspectral imagery different from Multispectral imagery?
Have a look at the depiction below-
I had already delved into what Spectral Resolution is, earlier on in this post. Spaceborne Hyperspectral instruments for Earth Observation (The Environmental Mapping and Analysis Program or EnMAP for example) acquire reflectance information across multiple narrow Spectral bands (i.e. high Spectral Resolution) which are contiguous as well. Spaceborne Multispectral instruments for Earth Observation, by contrast, capture information across comparatively fewer and wider Spectral bands (i.e. lower Spectral Resolution), for non-contiguous ranges of wavelength as well.
The two charts in Figure 5 above elucidate the difference - while Multispectral imagery contains information which you can draw a Bar chart with (by virtue of it being categorical data i.e. reflection response per non-contiguous band), Hyperspectral imagery contains information which you can also draw a Histogram with (by virtue of it being continuous data i.e. reflection response across contiguous bands). The Histogram curve, in the context of Hyperspectral imaging, is known as the Spectral Signature - how an object/surface responds to the waves within a spectral range.
The Data Cube in Figure 6 depicts hundreds of stacked Hyperspectral images of the same region acquired at the same time, albeit with varying reflectance response to the numerous contiguous bands of wavelengths within Solar Radiation that are acquired by the instrument - this a spectral fingerprint of the entire region.
Thus, higher spectral resolution and reflectance acquisition across a continuous spectral range (spectral signature) are the standout features that distinguish Spaceborne Hyperspectral Imaging from Spaceborne Multispectral Imaging for Earth Observation purposes.
Let me elaborate where the utility of using Hyperspectral Imaging is using a simple analogy. Imagine a transparent container stored at a cryogenic temperature (−150°C) which contains five substances in a solid state within. Upon gradually heating the container to a piping hot 150°C, you note down the exact temperature when each substance within undergoes a change in its state of matter. Upon charting your findings (Figure 7), what you have essentially come up with is the Temperature Signature of the five substances.
Which of the five substances is H2O (Water)?
Very simple - only Substance C as H2O exists as Ice below 0 degrees Celsius, as Water between 0-100 degrees and as Water Vapour above 100 degrees. The other four substances do not have the same temperature signature as what is known of H2O and hence, one can easily rule all of them out.
Now imagine that you maintain a database of Temperature Signatures, i.e. response to a temperature range, of ten important substances that are of interest to you for your study. Using this available information, you will be able to identify which, if any, of the remaining four substances in the container are the ones important to you. This is exactly similar to how Hyperspectral Imaging is utilized to good effect - instead of temperature signature, we compare the spectral/energy signature acquired by the spaceborne instrument to a database of verified spectral signatures in order to detect a particular surface feature, delineate between surface features, or classify surfaces.
For example, in Figure 8 below, observe the spectral response of healthy, stressed and dry vegetation to Solar Radiation at lower wavelengths (between 400-800 nm i.e. the Visible Light portion). There is a distinct difference in the reflectance response of the three types of surfaces and this can be strongly attributed to the quantity of pigments within - healthy vegetation appear greener due to the overshadowing presence of chlorophyll, the green-colored pigment which helps plant create food through photosynthesis, which absorbs the red and blue wavelengths of visible light, reflecting only the green, hence the overall low reflectance to the visible light wavelengths.
If the plants become stressed due to adverse conditions, they start to lose their ability to develop new chlorophyll, and the proportion of other pale-colored pigments within increases which gives the vegetation a yellow-hue and this is also when the rate of visible light reflectance peaks.
However, the reflectance becomes very low again when the plants wither and die, let's say due to drought. This is because all the pigments, be it chlorophyll or carotenoids, cease to develop and only the tannins are left behind which give the leaves its woody color and which reflects only the red wavelength of visible light.
Thus, you can appreciate how intricately the types of vegetation respond to the visible light portion of solar radiation - the varying spectral signatures of these surfaces helps us to glean profound insights.
Similarly, there is a distinct variability in reflectance response of these three types of surfaces at higher wavelengths (between 1400-2400 nm i.e. intermediate infrared portion of solar radiation). This can be strongly attributed to the extent of moisture content within. However, unlike pigments, the variation is much more linear - this is because the extent of reflectance is directly proportional to the quantity of moisture - water absorbs infrared rays and is quintessential for the health of the vegetation as well. Lesser the quantity of moisture, higher the extent of infrared reflectance there would be which would signal that the vegetation is stressed or dry.
Thus, I hope the utility of having continuous energy readings across a spectral range is evident to you - through this information acquired by a hyperspectral imaging sensor, one can detect, delineate or classify vegetated surfaces and numerous other types of surfaces. The strengths of Hyperspectral imaging also serves to highlight the weakness of Multispectral imaging - consider the spectral range of Sentinel-2 satellite's Multispectral Imager (MSI) instrument as depicted in Figure 9 below-
While it would be possible to demarcate healthy vegetation from stressed or dry vegetation using the information aggregated within the spectral bands of Sentinel-2, however, it would be challenging to distinguish whether the vegetation is highly-stressed or dry due to a) non-contiguous bands i.e. gaps in data and b) broader bands particularly at higher wavelengths which would normalize rather than highlight the variable reflectance response of these two surfaces.
Remind yourself that how a surface responds to energy is determined not by an individual factor in isolation but by the surface roughness, geometric and biogeochemical properties of the surface material as a whole (besides atmospheric influences and the characteristics of the transmitted wavelength itself). More of one and less of the other would alter the Spectral Signature of the surface under consideration. Video 1 below depicts an iterative spectral signature of vegetation by altering using one geometric influencer (leaf area index) and two biogeochemical influencers (chlorophyll, leaf water content). Such intricacies in reflectance responses are precisely why using Multispectral imagery for finer characterization of surfaces tends to be challenging and which is where the superiority of Hyperspectral imaging comes to the fore.
Similarly, refer Figure 10 which depicts the spectral signature of open water and coastal water. Sure, one can use Multispectral Imagery (the narrow B1 & B3 bands in Sentinel 2 MSI for example) to distinguish whether a pixel over an ocean is open or coastal, however it would be complicated as the difference in reflectance is tiny. With Hyperspectral imaging, the task would become simpler through the derivation of spectral signature.
Everything sounds so rosy about Hyperspectral Imaging. Do we really need Multispectral Imaging?
Yes we do. More information does not always imply good information, it can become a double-edged sword. Because the Spectral bands in Hyperspectral images are narrow and contiguous, researchers often encounter interference from the neighbouring bands while processing select bands for a workflow. This is akin to cross-talk in telecommunications and is problematic as it hampers the accuracy of delineation. Filtering the unwanted noise away isn't simple either - it necessitates the use of specialized and computationally-intensive corrective techniques. In contrast, the non-contiguous nature of Spectral bands within Multispectral images ensures that data leakage doesn't occur. Moreover, the cost of acquiring Hyperspectral data and the expertise required to process it is much higher compared to that of Multispectral imaging as of today.
Some of the drawbacks of Spaceborne Multispectral imaging for Earth Observation are also applicable to Spaceborne Hyperspectral imaging for Earth Observation - acquisitions cannot be made at night-time and are also susceptible to atmospheric influences during the day.
Hyperspectral data can be acquired using different techniques, the choice of which depends on aspects such as the biogeochemical characteristics of the object or surface, sensitivity to external influences, desired resolution, cost constraints, and coverage requirements.
Some acquisitions are best captured using Ground-based sensors in a field or a laboratory environment - especially if the object or surface under consideration is highly sensitive to external influences or where there is a need for very high spatial resolution.
Have a look at the two videos below - you'll realize that acquiring hyperspectral data using ground-based techniques requires considerable expertise.
Another laboratory-based Hyperspectral imagery acquisition video can be viewed here.
On the other hand, Spaceborne sensors are preferred when one desires wider and faster coverage, have tighter cost budgets, have lesser need for very high spatial resolution, and to study surfaces that are less sensitive to external influences.
Airborne sensors, where the sensor system is installed on a research aircraft or a drone - are used extensively too, and are particularly effective for workflows which require more and faster coverage than a on-ground acquisition and/or when there is a need for a higher spatial resolution than what a spaceborne sensor is able to acquire.
Besides the mode of acquisition, there are different scanning technologies to choose from as well. Refer the video below to know more-
Some research studies may require Hyperspectral data acquired using different modes and techniques in order to derive or validate the output.
The scope of utilizing Hyperspectral Imaging in scientific research is vast - it is already being used extensively in Agricultural Applications, Soil Applications, Mining Applications, Coastal Applications, Hazard Applications, Archaeological Applications and Military Applications, among several others (explore some Hyperspectral missions and applications here).
I was pleasantly surprised to know that what was originally designed for Remote Sensing has even found applications in Crime Scene detection, Forensic Medicine and the Biomedical sector, the latter where Imaging Spectroscopy is deployed as a non-invasive technique to distinguish cancerous tissues from healthy ones through the use of wavelengths that are able to penetrate the outer skin of humans.
Now that you've become familiar with the concept of Hyperspectral Imaging and its utility, I'll leave you with a video which demonstrates its practical use for Land Cover classification. You'll get hands-on insights on how to process Hyperspectral Images on an open-source software and how to interpret the output-
Video Time Stamps
00:04 - Video Details
00:19 - Getting familiar with the Datasets
01:09 - Exploring Airborne & Spaceborne Hyperspectral Imagery
03:23 - Visualizing the Spectral Signature
04:50 - Visualizing the Spectral Library
06:05 - Using Regression Analysis to generate Land-Cover Map
Thanks for reading this post. Feel free to share your feedback here.
ABOUT US
Intelloc Mapping Services, Kolkata | Mapmyops.com offers Mapping services that can be integrated with Operations Planning, Design and Audit workflows. These include but are not limited to Drone Services, Subsurface Mapping Services, Location Analytics & App Development, Supply Chain Services, Remote Sensing Services and Wastewater Treatment. The services can be rendered pan-India and will aid your organization to meet its stated objectives pertaining to Operational Excellence, Sustainability and Growth.
Broadly, the firm's area of expertise can be split into two categories - Geographic Mapping and Operations Mapping. The Infographic below highlights our capabilities-
Our Mapping for Operations-themed workflow demonstrations can be accessed from the firm's Website / YouTube Channel and an overview can be obtained from this brochure. Happy to address queries and respond to documented requirements. Custom Demonstration, Training & Trials are facilitated only on a paid-basis. Looking forward to being of service.
Regards,
