A METHODOLOGY TO STUDY BEACH MORPHODYNAMICS BASED ON SELF-ORGANIZING MAPS AND DIGITAL IMAGES

The use of digital images (DI) to study the beach morphology has become an important tool in recent years, as it is easier to obtain high resolution information for large areas and over longer periods of time. Measuring a large set of DI however is not a simple task. A methodology to classify beach DI based on Self-Organizing Maps has therefore been developed, which required several experiments to find the most adequate variables and metrics for the training process. The methodology has three parts, a preparation of DI, the SOM training and the reconstruction of evolution series. The preparation deals with the effect of tides and variations of light due to the meteorological conditions and time of day. The training of SOM is done using correlation such as metric and the processed DI. The methodology is applied to classify DI from Barceloneta Beach using the coastline position.