Mapping beach morphodynamics remotely: A novel application tested on South African sandy shores

Sandy beaches have been identified as threatened ecosystems but despite the need to conserve them, they have been generally overlooked. Systematic conservation planning (SCP) has emerged as an efficient method of selecting areas for conservation priority. However, SCP analyses require digital shapefiles of habitat and species diversity. Mapping these attributes for beaches from field data can take years and requires exhaustive resources. This study thus sought to derive a methodology to classify and map beach morphodynamic types from satellite imagery. Since beach morphodynamics is a strong predictor of macrofauna diversity, they could be considered a good surrogate for mapping beach biodiversity. A dataset was generated for 45 microtidal beaches (of known morphodynamic type) by measuring or coding for several physical characteristics from imagery acquired from Google Earth. Conditional inference trees revealed beach width to be the only factor that significantly predicted beach morphodynamic type, giving four categories: dissipative, dissipative-intermediate, intermediate and reflective. The derived model was tested by using it to predict the morphodynamic type of 28 other beaches of known classification. Model performance was good (75% prediction accuracy) but misclassifications occurred at the three breaks between the four categories. For beaches around these breaks, consideration of surf zone characteristics in addition to beach width ameliorated the misclassifications. The final methodology yielded a 93% prediction accuracy of beach morphodynamic type. Overlaying other considerations on this classification scheme could provide additional value to the layer, such that it also describes species' spatial patterns. These could include: biogeographic regions, estuarine versus sandy beaches and short versus long beaches. The classification scheme was applied to the South African shoreline as a case study. The distribution of the beach morphodynamic types was partly influenced by geography. Most of the long, dissipative beaches are found along the west coast of the country, the south coast beaches are mostly dissipative-intermediate, and the east coast beaches range from short, estuarine pocket and embayed beaches in the former Transkei (south east), to longer intermediate and reflective beaches in KwaZulu-Natal (in the north east). Once combined with the three biogeographic regions, and distinguishing between estuarine and sandy shores, the South African coast comprised 24 different beach types. Representing shorelines in this form opens up potential for numerous spatial analyses that can not only further our understanding of sandy beach ecology at large spatial scales but also aid in deriving conservation strategies for this threatened ecosystem. (C) 2010 Elsevier Ltd. All rights reserved.