Remote sensing of tidal freshwater marsh elevation, channels, and vegetation structure

Abstract

I mapped three elevation classes in a Mid-Atlantic tidal freshwater marsh using QuickBird multi-spectral imagery and field measurements of elevation and channel networks. The elevation map reveals heterogeneous surfaces at a 2.4 m pixel scale. Field collected vegetation cover data differed among the three elevation classes. Species richness and the number of positive associations among species were higher in the mid- and high-marsh than the low marsh. The number of negative associations among species increased with rising elevation. Random forest classification of elevation class using species cover vectors selected only <em>Impatiens capensis</em> and <em>Nuphar luteum</em> and had an out of box predictive error of 26%. My research shows that the vegetation of freshwater tidal marshes is diverse with ill-defined boundaries between distinct communities. Yet vegetation shifts beyond the traditional low and high marsh communities could be detected, which should provide a useful tool for restoration and management.