USING AUTOMATED IMAGE CLASSIFICATION TO LINK GEOMORPHOLOGY AND SHALLOW STRATIGRAPHY ALONG THE DELMARVA PENINSULA

A dense network of high-resolution geophysical data, bottom photographs, and samples has been used to define the surficial geology of the inner continental shelf along the Delmarva Peninsula. Three acoustically distinct bottom types are defined in composite backscatter and bathymetry data using an automated, supervised maximum likelihood image classification technique at a 20-meter-per-pixel resolution over 540 km² of seafloor. The three bottom types include high backscatter, low backscatter, and low backscatter with bedforms. Grain size statistics from 50 samples within the surveyed area suggest that the three units that were mapped based on differences in backscatter intensity and bathymetric gradient are also unique texturally and are consistent with distinct outcropping units observed in subbottom data. Subbottom data reveal a ravinement surface that is often coincident with the seafloor or locally overlain by acoustically transparent deposits that consist of 99% sand and cover about one quarter of the surveyed area. In composite imagery these areas identify as low backscatter intensity with steep bathymetric gradients and are consistent with sand wave morphology. Low backscatter units with gentle slopes are most often characterized by laminations in subbottom data. Low backscatter, low gradient areas cover nearly half of the seafloor and are on average 80% sand with mud and small amounts of shell and gravel. High backscatter seafloor has gentle slopes and occurs in depressions adjacent to shoals. High backscatter seafloor often correspond to outcropping channel fill deposits or antecedent headlands in the subsurface, and has the coarsest textural signature with an average of 70% sand with gravel and shell making up most of the remaining sediment. Zones of sediment erosion, transport, and deposition are, controlled the composition of the underlying sediments and characterize the Delmarva inner shelf seafloor. The integration of geophysical data, groundtruth datasets, and image classification techniques has produced a high-resolution, multi-faceted surficial geologic interpretation that links bottom type and texture to sedimentary environment processes along the Delmarva Peninsula.