Correcting wave reflection estimates in the coastal zone

Abstract

The impact of random noise on an existing two-dimensional method for separating incident and reflected wave spectra using an array of wave gauges is investigated using simulated time series with known wave amplitudes, reflection coefficients, and signal-to-noise ratios. Both the incident and reflected spectra are overestimated by a quantity that can exceed 100% for signal-to-noise ratios less than 1. Consequently, estimated reflection coefficients are also overestimated with larger errors occurring when the known reflection is low. Coherence decreases systematically with increasing noise and this trend is used to develop a mathematical function to correct for the observed bias and provide 95% confidence intervals for incident and reflected spectra and reflection coefficients. The correction technique is shown to be very effective in reducing error by up to ~90%. Field data from a natural beach are used to demonstrate the application of these results; corrected values suggest that reflection coefficients are frequently overestimated by over 50%.