Why hyperspectral is the future

Hyperspectral vs. Multispectral

We often hear the question why hyperspectral sensors should be considered over multispectral sensors as a common subject of discussion, since the latter usually are available at a significant lower price.

Furthermore, hyperspectral imaging sensors, when compared to their multispectral counterparts, are according to those arguments more difficult to handle because:

  • (a) the amount of data would be too large
  • (b) in the case of our hyperspectral snapshot technology the spatial resolution is too low and
  • (c) last but not least, multispectral imaging delivers sufficient information content which makes hyperspectral data redundant.

In order to understand why those arguments are wrong let’s take a closer look at the technology and the advantages and disadvantages of both sensor types.

Let's start

by understanding the basics of spectroscopy.

Identification of hidden features – what is spectroscopy?

Spectroscopy enables its user to identify spectral features that are not visible for common cameras or the human eye. Those features usually are directly related to the optical properties of the analysed materials.

Due to the fact that each material has a different spectral signature such data has the potential not only to separate specific materials from others, but also allowing it to make qualitative statements on the analysed object. Spectrometers originally are one-dimensional point sensors.

From point measurements to an area – spectral imaging as the next step

Hyperspectral Advantage Spectrum Comparison - Cubert GmbH - Germany

Fig. 1: Contiguous hyperspectral reflectance spectrum compared to the VNIR center wavelengths of ESA’s multispectral Sentinel-2 satellite.

Per definition imaging spectroscopy, which is also known as hyperspectral imaging, acquires simultaneous images in a high number of spectral bands, so that for each pixel of the resulting multichannel image a contiguous reflectance spectrum can be derived (e.g., GOETZ ET AL., 1985).

In order to demonstrate the huge advantage of hyperspectral data, figure 1 shows a quasi-continuous spectrum of a hyperspectral sensor and the center wavelengths of a multispectral sensor, in this case from Sentinel-2. “It is obvious that the contiguous spectrum contains much more information, e.g., on specific absorption ranges, than the multispectral dataset.” (LOCHERER, 2014)

Now bringing a hyperspectral sensor from a point to the area, it is no longer only possible to make statements on the analysed object or its quality, but also in its quantity and behaviour to its surroundings.

Spectral imaging allows to examine a spatial distribution of different materials and quality differences.

Small number of broad bands - multispectral camera technique

Multispectral sensors by contrast only have a small number of bands, and these bands usually are relatively broad.

When speaking of the spectral features of materials that are mentioned above the bands of MS sensors in most case are not sensitive enough to allow an accurate identification of those features, since these very often occur in a narrow part of the electromagnetic spectrum. These features are called absorption bands.

Of course, when calculating the most common vegetation index of all, i.e. the Normalized Difference Vegetation Index NDVI (DEERING & HARLAN, 1974), you only need one band in the Red and one in the NIR.

No matter how broad these bands are, you receive a solid result. However, you are capable of calculating the NDVI exclusively.

Hyperspectral advantage – use of narrowband indices

Since the beginning in the 1980s, hyperspectral imaging enabled the development of a whole range of narrowband indices that are used for the determination of various characteristics. Most of those indices are based on a study for a specific problem, consequently they use the whole range of different wavelengths (link to indices).

When analysing vegetation properties, those indices allow the retrieval of specific information, such as vitality status, chlorophyll content, which directly is related to nitrogen demand, water content, dry matter or leaf area index, just to name a few of those parameters being valuable for agriculture and forestry.

Furthermore, hyperspectral data is most suited for the inversion of complex physically-based radiative transfer models, which allow an even more accurate information retrieval than indices.

Since the outcome of those data mining strategies are physical parameters, in-situ measurements that relate the dimensionless value of an index to a parameter, are not necessary any longer (LOCHERER ET AL, 2015).

Low spatial resolution of
the hyperspectral snapshot camera is a disadvantage - is this true?

Simply said, no it is not. Because our technology is true snapshot, the image acquisition is quite fast.

Therefore it is no longer a time-consuming challenge to acquire several images in a wink of an eye and to mosaic those images into a high resolution hyperspectral data cube.

References

  • DEERING, D. W., & HARLAN, J. C. (1974): Monitoring the vernal advancement and retrogradation (greenwave effect) of natural vegetation (p. 371). Texas A & M University, Remote Sensing Center.
  • GOETZ, A.F.H., VANE, G., SOLOMON, J. & ROCK, B.N. (1985): Imaging spectrometry for earth remote sensing. Science, 228, 1147-1153.
  • LOCHERER, MATTHIAS (2014): Capacity of the hyperspectral satellite mission EnMAP for the multiseasonal monitoring of biophysical and biochemical land surface parameters in agriculture by transferring an analysis method for airborne image spectroscopy to the spaceborne scale. Dissertation, LMU München: Fakultät für Geowissenschaften
  • LOCHERER, M., HANK, T., DANNER, M., & MAUSER, W. (2015): Retrieval of Seasonal Leaf Area Index from Simulated EnMAP Data through Optimized LUT-Based Inversion of the PROSAIL Model. Remote Sensing, 7(8), 10321-10346.
Loading
System Comparison
realtime spectral imaging - Cubert GmbH - Hyperspectral imaging - Germany
Real-time Image Classification based on Machine Learning
Menu