Hyperspectral Bioindicators for Remote Detection of Environmental Contaminants

Nuclear facilities release large amounts of radioactive and non-radioactive contaminants into their local environments. These include radioactive wastes, toxic concentrations of metals, and petroleum products. It is widely understood that such contaminants can jeopardize human health, compromise the food supply, and damage ecosystem functions. However, these risks are poorly quantified because of technological constraints that limit the scope of environmental monitoring. Improving our capacity to detect contamination over large spatio-temporal scales is essential to monitor and potentially provide early warning for such hazards.

Hyperspectral remote sensing (imaging spectroscopy) provides optical imagery across a continuous range of wavelengths. Such contiguous narrowband (~3-20 nm) data offer greater diagnostic capabilities than multispectral imagery due to the presence of distinct absorption features associated with chemical constituents. Spectroscopy is already widely used to measure agricultural productivity and plant functional traits.

When plants are exposed to various contaminants, they may respond with detectable physiological or chemical changes, such as increased defense compounds, or a reallocation of pigment pools. Identifying hyperspectral features that are associated with exposure to contaminants could leverage local vegetation as passive, low-cost bioindicators of pollution. Operationalizing this technology requires that we: (i) quantify the physiological and chemical changes that contaminants induce in plants; (ii) differentiate between stress responses induced by contaminants versus other environmental stressors; (iii) assess the interactive effects between multiple environmental stressors; (iv) characterize species-specific interactions with contaminants.

I report results from a field experiment using fescue conducted in conjunction with US National Lab partners. We show distinctive spectral responses in exposed vs unexposed plants, and also that different stressors (drought, chromium, copper) exhibit different spectral responses.