2019 CSDMS meeting-117
Geospatial prediction of subsurface wood-bearing sediments, Northern Gulf of Mexico
Recent discovery of a well-preserved drowned bald cypress forest offshore Alabama has spurred the search for analogous sites, as they provide valuable paleoclimate proxies and potential paleohuman habitats. However, drowned forests are difficult to detect when buried beneath the seabed, and degrade rapidly when exposed to the water column.
In this study, various machine learning algorithms within NRL's Global Predictive Seabed Model (GPSM) are used to geospatially predict the location of buried ancient forests offshore Mississippi. Subsurface sediment cores containing evidence of ancient forests (wood debris) are used as training and validation data, and feature layers include modern bathymetry, paleo-topographic surfaces, and seabed substrate. The resulting maps of probability of encountering wood-bearing sediments will be used to guide future data acquisition efforts.