Video-based nearshore bathymetry estimation for Rip current forecasting on a macrotidal beach

Abstract

Rip currents are seaward directed flows of water that traverse the surf zone, and in many cases pose potentially dangerous conditions for swimmers and beach users. The present study is carried out focusing on Perranporth beach, UK, which is known to present quite strong rip current events that have been the culprits of large numbers of incidents or rescues (in some cases more than 65 in a single day) in the past (Scott et al., 2009). The main objective of this work is related to the forecasting of rip currents, particularly in helping to develop, calibrate, and validate an operational swimmer safety model for Perranporth beach, UK. This main objective is divided into several specific objectives, which look into answering both general scientific questions and specific details for the implementation of the model at the study site. The tool used for the estimation of bathymetries is the data-model assimilation scheme Beach Wizard, which uses video-derived information to adjust an initial bathymetry into an updated bathymetry. The Beach Wizard model is run with the numerical model XBeach, assimilating the hydrodynamic model output with video-derived measures of wave dissipation (imaps). Specific model formulations and parameters of Beach Wizard are varied and tested as to determine the effects of each, as well as to obtain the best working model for Perranporth, UK. Two systematic estimation errors that were originally pointed out by van Dongeren et al. (2008) are looked into, resulting in the implementation of a variation in the bed updating process, by introducing a time-dependent bed change factor. The changes applied to the model formulations and the adjustments of default parameters resulted in accurate estimations of bathymetries over a three month period (April-June 2011) during which video derived input and five measured bathymetries are available at Perranporth. The importance of video derived input (timex images) is analyzed in detail, by defining several criteria concerning the quality, quantity, and temporal regularity of images used as input. The quality criteria is divided into two separate indicators, in this manner differentiating between actual image quality (if the camera lens is clean or dirty, or if there is foggy conditions, etc.) and dissipation map (imap) quality (if the dissipation patterns are good or not). With this, it was found that imap quality stands over image quality, meaning that it is more important to have favorable wave breaking conditions than perfect image resolution. As well as this, image regularity was found to be important in maintaining the models predictive ability, due to large temporal gaps between consecutive images increasing the bed uncertainty significantly (resulting in very large, and possibly erroneous bed changes if the corresponding imap is not perfect). The final criteria looked at was image quantity, which turned out not to be of great importance as long as the image regularity is maintained; presenting even improved results for a case using a smaller total amount of input with the same temporal spacing. The estimated bathymetries computed with Beach Wizard were used to model nearshore flows in a non-stationary hydrodynamic XBeach model. It was found that very promising similarities are seen between said currents and the currents modeled over measured bathymetries. The use of estimated bathymetries show clear improvements over using an outdated measured bathymetry for the prediction of rip currents, essentially validating the use of the Beach Wizard tool in the eventual implementation of an operational rip current forecasting system at Perranporth beach.