Coral Reef Reflectance Characteristics – Impacts of increasing water depth on spectral similarity

In a previous post we demonstrated how dendrograms can be used as effective tools for investigating the similarities and differences of reflectance spectra. Here we expand on our earlier discussion, and explore how dendrograms can be used to illustrate the decreasing separability of coral reef spectra with increasing water depth.

Coral reefs are known for their exceptional biodiversity, containing a complex array of mobile and sessile organisms. From a remote sensing perspective, however, coral reefs can be particularly challenging study areas, mostly due to the confounding effects of the overlying water column. Varying water depth and varying water properties can both contribute significant complexity to the interpretation and identification of features on the sea floor.

Given this complexity, it is therefore necessary in remote sensing to simplify coral reef ecosystems into a collection of generalized components, each representing a unique compilation of species and/or substrate types. When grouping species for analysis, and when interpreting image classification output, it is important to understand the spectral similarity – or dissimilarity – of different image features.

As an example, let’s first consider an example where reef habitat composition is represented by four fundamental components: coral, sponge, sand and submerged aquatic vegetation (SAV). As shown in our previous post, the average in situ spectra of these four components exhibit unique reflectance characteristics and can be readily differentiated at 0.1 spectral angle. However, with increasing water depth (approximated here using a semi-analytic model for clear tropical water) the separability of these components decreases. At 3 m water depth it becomes more difficult to differentiate sand from SAV, and coral from sponge; and at 10 m water depth analysis is essentially reduced to a two-component systems: sand versus coral, sponge and SAV.

Coral Components Dendrogram

Let’s now compare and contrast the above results with the same analysis applied to all of the individual spectra used to create these component averages. This includes measurements from 24 coral species, 10 sponge species, 3 SAV species and areas of sand. As shown below, many individual species (and in some cases small groups of species) can be readily differentiated when the overlying water column is not considered. However, when the effects of the water column are included, the ability to distinguish individual species diminishes significantly with increasing water depth.

Coral Species Dendrogram 0mCoral Species Dendrogram 3mCoral Species Dendrogram 10m

While the relationship between water depth and spectral similarity is to be expected, what is particularly informative from these dendrograms is the ability to discern which species group together at different depths. For example, note that coral species do not always group with other coral species, but are observed to also group with both sponges and SAV. Additionally, because spectra do not necessarily group according to type, it becomes apparent that three spectral groups can be reasonably differentiated at 10 m rather than just two groups as suggested from the analysis using just averages for each component.

Such information can be immensely valuable for guiding image analysis, as well as aiding the interpretation of results. So if you’re working on remote sensing of coral reefs, it’s worth exploring the spectral characteristics of the dominant species in your study area, and investigating how spectral similarity changes with water depth.

Related post: Assessing Spectral Similarity – Visualizing hierarchical clustering using a dendrogram

Acknowledgement: Spectral data used in the above examples were collected using a GER-1500 spectrometer by M. Lucas at the University of Puerto Rico at Mayaguez for a NASA EPSCoR sponsored research project on the biodiversity of coastal and terrestrial ecosystems.


Assessing Spectral Similarity – Visualizing hierarchical clustering using a dendrogram

When conducting remote sensing analysis, it can often be very instructive to evaluate the spectral similarity – or dissimilarity – of different image features.

Below we demonstrate the use of a dendrogram to quantitatively analyze and visually assess the similarity and hierarchical clustering of reflectance spectra.

Understanding spectral relationships can impart valuable information on: the potential spectral variability within a given scene, the capacity to differentiate certain unique features in an image, the need for clustering spectrally similar features, or the number of spectral endmembers that could be used to describe a particular area. Spectral analysis can therefore be an important first step in understanding the capabilities and limitations of different analysis methods and application objectives.

Spectra are typically obtained from field or laboratory measurements, from an existing spectral library, or derived from the image itself. Similarity between individual spectra, or between clusters of similar spectra, can be mathematically analyzed using different distance metrics, such as root-mean-square error or spectral angle. The magnitude of similarity amongst these distances can then be used as an indicator of the ability to differentiate and/or identify spectral features using different image processing techniques.

When analyzing a relatively small set of spectra, evaluating results from a similarity assessment can be easily achieved by simply examining the output directly. For example, if we consider a set of coral reef spectra representing coral, sponge, sand and submerged aquatic vegetation (SAV), analysis using spectral angle reveals measurable differences between all four spectra. This is also apparent by plotting the spectra themselves, and seeing that each feature exhibits unique characteristics.

Spectral analysis coral reef

However, when the number of spectra is increased, and the spectral relationships become more complex, visual assessment of both the spectral signatures and similarity output becomes more difficult to interpret. For example, expanding our coral reef analysis, we now investigate spectra from 10 individual sponge species, and find the interpretation of results to be less obvious. There are clearly species exhibiting close similarities, as well as differences, but it is not immediately apparent which species can be easily differentiated and which species need to be clustered.

Spectral angle sponge species

Reflectance spectra sponge species

A useful method that can assist with this analysis is utilizing output from the distance calculations to build a dendrogram. This provides a visual representation of spectral relationships, as well as quantitative information relevant to image analysis. For example, as shown below, of the 10 sponge species in our analysis, 6 of these species can be differentiated at 0.1 spectral angle; however, there are two sets of species, (i) Chondrilla spp. and C. caribensis and (ii) A. lacunosa and I. felix, that are closely similar at 0.1 spectral angle and would need to be clustered at this level of analysis.

Dendrogram sponge species

To further illustrate the utility of using a “spectral dendrogram”, we expand our coral reef analysis to include spectra from 24 coral species, 3 SAV species and sand, in addition to the 10 sponge species. As evident from the dendrogram, spectral relationships in this analysis are significantly more complex. For example, at 0.1 spectra angle there are multiple situations where different spectral types (e.g., coral and sponge, or SAV and sponge) are closely similar and can’t be spectrally differentiated. This has important implications, and potential limitations, for subsequent spectral analysis and image classification results.

Dendrogram coral reef species

These are but a few examples illustrating the types of visualization and levels of information that can be derived from these plots. However, potential applications are as varied as your spectra, so we invite you to explore the use of dendrograms in your own spectral analysis.

Related post: Coral Reef Reflectance Characteristics – Impacts of increasing water depth on spectral similarity

Acknowledgement: Spectral data used in the above examples were collected using a GER-1500 spectrometer by M. Lucas at the University of Puerto Rico at Mayaguez for a NASA EPSCoR sponsored research project on the biodiversity of coastal and terrestrial ecosystems.

Open Access Spectral Libraries – Online resources for obtaining in situ spectral data

Coral SpectraThere are many different analysis techniques used in remote sensing, ranging from the simple to complex. In imaging spectrometry, i.e. hyperspectral remote sensing, a common technique is to utilize measured field or laboratory spectra to drive physics-based image classification and material detection algorithms. Here the measured spectra are used as representative examples of the materials and species that are assumed present in the remote sensing scene. Spectral analysis techniques can then be used to ascertain the presence, distribution and abundance of these materials and species throughout an image.

In most cases the best approach is to measure field spectra for a given study area yourself using a field-portable spectrometer; however, the time and cost associated with such fieldwork can oftentimes be prohibitive. Another alternative is to utilize spectral libraries, which contain catalogs of spectra already measured by other researchers.

Below are examples of open access spectral libraries that are readily available online:

  • The ASTER Spectral Library, hosted by the Jet Propulsion Laboratory (JPL), contains a compilation of three other libraries, the Johns Hopkins University Spectral Library, the JPL Spectral Library and the USGS Spectral Library. The ASTER library currently contains over 2400 spectra and can be ordered in its entirety via CD-ROM or users can also search, graph and download individual spectra online.
  • The SPECCHIO Spectral Library is an online database maintained by the Remote Sensing Laboratories in the Department of Geography at University of Zurich. Once users have registered with the system to create an account, the SPECCHIO library can be accessed remotely over the internet or alternatively downloaded and installed on a local system. The library is designed specifically for community data sharing, and thus users can both download existing data and upload new spectra.
  • The Vegetation Spectral Library was developed by the Systems Ecology Laboratory at the University of Texas at El Paso with support from the National Science Foundation. In addition to options to search, view and download spectra, this library also helpfully includes photographs of the actual species and materials from which the data was measured. Registered users can also help contribute data to further expand the archive.
  • The ASU Spectral Library is hosted by the Mars Space Flight Facility at Arizona State University, and contains thermal emission spectra for numerous geologic materials. While the library is designed to support research on Mars, the spectra are also applicable to research closer to home here on Earth.
  • The Jet Propulsion Laboratory is currently building the HyspIRI Ecosystem Spectral Library. This library is still in its development phase, and hence contains only a limited number of spectra at this time. Nonetheless, it is expected to grow, since the library was created as a centralized resource for the imaging spectrometry community to contribute and share spectral measurements.

It is doubtless that other spectral libraries exist and that many thousands of additional spectra have been measured for individual research projects. It is expected that more and more of this data will be available online and more uniform collection standards will be adopted, particularly as airborne and space-based hyperspectral sensors continue to become more prevalent.

Searching for other remote sensing data resources? Check out these earlier posts on resources for obtaining general remote sensing imagery as well as imaging spectrometry and lidar data.