AUTOMATED GEOSTATISTICAL SEA-FLOOR CLASSIFICATION - PRINCIPLES, PARAMETERS, FEATURE VECTORS, AND DISCRIMINATION CRITERIA

A geostatistical method for automated seafloor classification is devel oped and applied to bathymetric data for a 150 x 100 km area at 26 deg rees N on the western flank of the Mid-Atlantic Ridge. The objective o f seafloor classification is to characterize seafloor properties quant itatively, and to use such spatial characteristics to distinguish roug hness provinces, and geologic and morphologic units automatically. The method presented here is based on the calculation of directional vari ograms as spatial structure functions. Parameters determined from filt ered variogram functions are used to compose Feature vectors, which ar e shown to be characteristic of morphologic prototypes and surface rou ghness types, and therefore facilitate a classification. Discriminatio n Criteria include spacing and strike of abyssal hill terrain, smoothn ess resulting from sediment cover, and parameters related to complexit y and morphological significance of abyssal hills and their slopes. Co mplications of automating the process concern robustness of parameter estimation, optimal window size, and subselection of data. By moving t he classification operation through the study area and color-coding pr operty classes, seafloor classification maps are obtained. The concept s of characteristic parameters, feature vectors and discrimination cri teria are illustrated in applications to bathymetric data from the wes tern flank of the Mid-Atlantic Ridge. Resultant classification maps ar e presented, with classes including roughness provinces and morphologi c units.