Objective Beach-State Classification From Optical Sensing of Cross-Shore Dissipation Profiles

Remote sensing using terrestrial optical charge-coupled device cameras is a useful data collection method for geophysical measurement in the nearshore zone, where in situ measurement is difficult and time consuming. In particular, optical video sensing of the variability in human-visible surface refraction due to the nearshore incident wave field is becoming an established method for distal measurement of nearshore subtidal morphology. We report on the use of a low-mounted shore-normal camera for gathering data on cross-shore dissipative characteristics of a dynamic open beach. Data are analyzed for the purposes of classifying three of Wright and Shorts' intermediate classes of morphological beach state as determined by expert raters. Although these beach states are usually thought of as being distinctive in terms of their longshore bar variability, theory predicts that differences should also be observed in cross-shore dissipative characteristics. Three methods of generating features from statistical features from the archived optical data are described and compared in terms of their ability to discriminate between the beach states. Principal component scores of the percentile distributions were found to provide slightly better classification performance (i.e., 85%, while approximating the data using relatively fewer features), whereas classification using intensity distributions alone resulted in the worst performance, classifying 78% of beach states correctly. Class center moment profiles for each beach state were constructed, and results indicate that cross-shore wave dissipation becomes more disorganized as linear bars devolve into more complex transverse structures