COS 281-1 - Spectral mixture analysis for surveillance of harmful algal blooms (SMASH): A hybrid laboratory- and satellite-based approach to remote sensing of algal taxa in inland waterbodies

Background/Question/Methods

Cyanobacterial blooms are a nuisance and potential hazard to human and animal health in both freshwater and marine systems. Advances in remote sensing may enhance our ability to study and monitor these events, for example, toxic and non-toxic algal taxa could potentially be distinguished based on their spectral characteristics. The U.S. Geological Survey is developing a remote sensing approach for characterizing algal blooms in inland waters termed Spectral Mixture Analysis for Surveillance of Harmful Algal Blooms (SMASH). This framework incorporates reflectance signatures for specific algal taxa, recorded with a custom-designed microscope hyperspectral imaging system, and hyperspectral satellite imagery.

Lab-based instrumentation allows cyanobacteria and other algae to be examined in a controlled environment at high magnification and with a level of spectral detail sufficient to distinguish among genera based on their unique reflectance spectra. Laboratory-derived reflectance data were compiled into a spectral library and used as input, along with hyperspectral satellite images, to a multiple endmember spectral mixture analysis algorithm that provided estimates of the fractional abundance of each algal taxon on a per-pixel basis.

Results/Conclusions

This hyperspectral characterization yields information on the composition, occurrence, and spatial distribution of algal blooms. Comparisons between satellite images and field sampling indicated a strong correspondence between estimated endmember fractions and observed proportional biovolumes. Notably, the approach successfully detected the cyanotoxin-producing genus Microcystis at one field site (Owasco Lake, NY) and avoided misclassification of Asterionella, a genus not included in our current library, at another site (Detroit Lake, OR). Additional results from Upper Klamath Lake, OR, highlighted the presence of multiple genera, primarily Aphanizomenon, distributed throughout the lake in complex spatial patterns that are known to occur in this particular waterbody.

Ongoing work is focused on refining validation protocols for microscope operation, augmenting the spectral library to include additional taxa (such as toxin producers), streamlining the SMASH workflow, and extending the approach from lakes and reservoirs to rivers.