Refocusing the Lens: A New Approach to Biodiversity Mapping from SBG-Like Data

Recommendations for the anticipated NASA Surface Biology and Geology (SBG) investigation include Earth observing imaging spectroscopy at 10 nm spectral resolution and 30 m spatial resolution. Mixed pixels will be common in SBG data because most terrestrial organisms (and their shadows) are shorter than 30 m. Consequently, variability due to mixed pixel content can overwhelm subtle biodiversity-relevant signals related to biophysical properties of interest, e.g., leaf nitrogen, chlorophyll, and cellulose. How can we best disentangle the signals of interest to isolate the ecologically relevant signal?

Here, we combine conventional canopy chemistry inversions with geographic mixture models to measure ecosystem structure, composition, and function. We illustrate the approach using airborne imagery from the National Ecological Observatory Network (NEON) Airborne Observational Platform (AOP), and the spaceborne DLR Earth Sensing Imaging Spectrometer (DESIS). When compared against field measurements of plant species richness, the new approach outperforms traditional methods by honing in on signals associated with foliar chemistry and structure, yielding novel insights into the underlying ecology of the landscapes under consideration. We propose a generalized model that would allow for automated implementation at scale. Primary benefits are: 1) physical basis, 2) computational ease, and 3) transparency. This approach can also be used in conjunction with other sensing modalities like LiDAR, thermal, and SIF to provide a comprehensive characterization of terrestrial ecosystem properties through multisensor fusion and test fundamental ecological hypotheses related to niche limitation and habitat complexity vs. species diversity relationships.