Evaluating the use of hyperspectral imagery to calculate raster-based wetland vegetation condition indicator

Abstract

Abstract Field observations and measurements of wetland plants have traditionally been used to monitor and evaluate wetland condition; however, there has been increasing use of remote sensing applications for rapid evaluations of wetland productivity and change. Combining key aspects of field- and remote sensing-based wetland evaluation methods can provide more efficient or improved biological indices. This exploratory study set out to develop a raster-based Wetland Vegetation Condition Indicator system that used airborne hyperspectral imagery-derived data to estimate plant-community quality (via wetland classification and Coefficient of Conservatism) and vegetation biomass (estimated using the Normalized Difference Vegetation Index). The Wetland Vegetation Condition Indicator system was developed for three Lake Ontario wetland areas and compared to a field-based floristic quality index and a dominant-plant based Floristic quality indexdom. The indicator system serves as a proof-of-concept that capitalized on the spatial and spectral attributes of high-resolution imagery to quantify and characterize the quality and quantity of wetland vegetation. A Pearson correlation analysis showed moderate r values of 0.59 and 0.62 for floristic quality index and floristic quality indexdom, respectively, compared to the indicator method. The differences are potentially due to the spatial resolution of the imagery and the indicator method only accounting for the dominant plants within each assessment unit (pixel), therefore disregarding understory plants or those with low abundance. However, the multi-metric Wetland Vegetation Condition Indicator approach shows promise as an indicator of wetland condition by using remotely sensed data, which could be useful for more efficient landscape-scale assessments of wetland health, resilience, and recovery.