Remote sensing methods for mapping the onset and progression of spartina alterniflora marsh dieback in coastal louisiana

We investigated the use of remote sensing in mapping the occurrence and monitoring the progression of a marsh dieback phenomenon observed in coastal Louisiana. Within our investigation, we demonstrated that the early and later stages of dieback progression were generally associated with green and red edge leaf reflectance changes and trends, while the later stages of marsh impact were represented by trends in blue and red leaf reflectance. Near infrared (nir) leaf reflectance trends were inconclusive while relative changes in predicted leaf carotene and chlorophyll were indicative of stressed plants. The nir/red ratio followed blue and red reflectance trends while the nir/green ratio mimicked the green and red edge reflectance trends indicating impact onset and progression as well as indicating later stages of impact. The nir/green also provided a convenient method to determine the relative time since dieback onset and offered an operational solution that would be amenable for current operational satellite remote sensing, Following our successful linkage of leaf optical changes to marsh dieback onset and progression, we applied our newly developed methods to changes in site-specific canopy reflectance spectra. Canopy reflectance spectra (~20 m ground resolution gained from a helicopter platform) were collected from marsh sites occupied during the leaf spectral analyses and sites exhibiting visual signs of dieback. Two scales of analyses were employed. One represented whole-spectra hyperspectral sensors such as the NASA EO1 Hyperion and the other broadband spectral sensors such as the NASA EO1 Advanced Land Imager and the Landsat Enhanced Thematic Mapper (limited to about 400 nm to 1000 nm). Spectral indicators generated in the whole-spectra analysis were used to determine the percent dead and healthy marsh composition at each site. The compositions were used classified into groups of marsh sites exhibiting similar levels of dieback. Separately, blue, green, red, red-edge, and near infrared (nir) spectral bands extracted from the field spectra and nir/green and nir/red band transforms were related to marsh dieback and progression indicators calculated from classifications of the 35- mm slides collected with the canopy reflectance recordings. Both the whole spectra and broadband spectral indicators could distinguish differences in (1) healthy marsh, (2) live marsh impacted by dieback, and (3) dead marsh. Broadband indicators provided some determination of dieback progression within the impacted live marsh while whole spectra hyperspectral indicators offered increased discrimination of dieback onset and progression. I. INTRODUCTION Areas of marsh dieback, commonly termed "brown marsh," were first observed in parts of Texas and Florida and throughout coastal Louisiana in the spring of 2000. At dieback sites, the normally dense and healthy intertidal salt marshes mostly composed of Spartina alterniflora or smooth cordgrass rapidly browned, and many ultimately died. Distribution of the dieback was widespread, and in many areas of the coast it affected a majority of the salt marsh area Fig.1. The occurrence of dieback clearly suggested an aggravated and elevated change in the marsh biophysical properties. At many dieback sites not abruptly curtailed by open water or some other natural or artificial feature, marsh canopy progressed in a zonal pattern from a few plants to a denser, healthier marsh surrounding the dead marsh.