NIR Spectroscopy for Large Area Assessment of Ecosystem Responses to Everglades Restoration.
Matthew Cohen, Sabine Grunwald, Mark Clark, and Ramesh Reddy. Soil and Water Science, Univ of Florida, 106 Newell Hall, Po Box 110510, Gainesville, FL 32611-0510
Restoration of the Everglades is among the world's largest and most complex ecosystem restoration projects. As implementation of the Comprehensive Everglades Restoration Program (CERP) begins, scientists and policy makers are seeking performance measures for assessing ecosystem response over time across vast areas. Numerous studies have demonstrated the robust utility of soils for indicating ecosystem degradation and recovery trajectories, but sampling and analysis constraints for large area assessment of soils are formidable hurdles to their use as routine performance measures. Here we develop models that employ diffuse reflectance spectra from a large spatial archive of Everglades soils for rapid inference of a wide array of soil performance metrics. Results suggest that spectroscopy can be reliably used as a proxy for analytical assessment of soils for many of these metrics; RPD values, which provide a useful across-parameter measure of model efficiency, were above 2.0 for most parameters, and above 1.5 for all. Further, spectra can be used to discriminate between soil and biological floc material (validation accuracy = 94%) which is of considerable utility for understanding spatial ecosystem responses to restoration activities. Finally, spectra are used to distinguish between peats deposited by different vegetative communities, which may have application for ecosystem back-casting and assessment of successional trajectories. These results coupled with the time and cost savings of spectral analysis methods and concurrent developments in field deployment of spectral sensors suggest that spectroscopy may provide a useful tool for large area assessment that can be used in service of restoration monitoring.