Remote sensing of Spartina anglica biomass in five French salt marshes

Abstract

Radiance measurements using a hand-held radiometer configured to match LANDSAT TM bands 3–5 were compiled from Spartina anglica plots in five geographically separate marshes in Brittany, France. Aboveground biomass of the plots was harvested following the determination of radiance. Simple regression models employing a vegetation index to predict S. anglica leaf, live and total aboveground biomass from spectral radiance values were derived from data collected in one of the marshes. Radiance measurements gathered in the other four marshes (test marshes) were substituted into the biomass estimation models. In three of the test marshes, the mean model-predicted leaf and live biomass were within 1–13 per cent of the harvest values. In the other test marsh, dead material located near the top of the canopy caused biomass to be underestimated. Total (live and dead) biomass predictions were not close to harvest values in any of the marshes. This study demonstrated that a simple regression equation relating radiance data (in the form of a vegetation index) to S. anglica biomass in one marsh can be employed to accurately predict leaf and live S. anglica biomass in other, distant, marshes.