Stordalen Mire Spectral Signature Library for Spectral Unmixing Using Sherwin Williams ColorSnap® Tool

Biodiversity can be an indicator of ecosystem health and a display of changes in climate over time. As Arctic environments continue to warm, permafrost peatlands will thaw at accelerated rates and become wetter (with sub-habitat shifts from intact palsa to bogs and fens). Plant species identification can provide some understanding of how these sub-habitats are changing over time. Analysis through aerial imagery from unpiloted aerial systems (UAS) can be used as a plant identification tool to measure and predict sub- habitat changes. In addition, such imagery allows for the exploration of species heterogeneity across these landscapes, which often has less than a meter of spatial patterning. Baseline field data still must be collected to create a spectral signature library of different plant species for imagery analysis. Typically, this is done with expensive, multi-band spectrometers, such as the 350-2500 nm field portable instrument used here (Spectra Vista Corp. HR-1024i). The Sherwin Williams 3-band ColorSnap® Tool is a cheaper, more portable alternative to the spectrometer with potentially similar accuracy. To test this device, more than 20 different plant species and other features such as rocks were sampled with the 3-band ColorSnap® to create a spectral signature library. Additionally, three species were scanned three times each with the ColorSnap® (30 scans each) and with the spectrometer to compare the two devices. These samples, or endmembers, represent the spectral signature of a pixel that purely represents that species. A discriminant function analysis was used to examine the overlap of endmembers and better visualize how the three-banded endmembers differentiated. Furthermore, the spectral library allows for spectral unmixing, which breaks down a UAS image into individual pixels, finds which signature is prominent in each, and displays the species associated with it. Spectral unmixing can be helpful to track sub-habitat changes over time, predict areas of high concern in the future, and measure biodiversity throughout a study area to indicate ecosystem health.