Revalue historical data to overcome the persistent lack of long-term datasets in coastal geomorphology: examples from Northern Ireland.

Abstract

Research studies and numerical models in coastal geomorphology often rely on short to medium-term datasets (months to years), normally depending on available financial resources to enrol employees or adopt up-to-date surveying technologies. Lately, satellite images have largely benefit long-term analyses of coastal change, extending the temporal investigation up to few decades, but limitations related to temporal coverage and spatial resolution remain. An unexploited source of information is represented by historical data, but modern coastal geomorphologists do not take advantage of them due to their poor reliability and accuracy, especially if compared to modern technologies. The keynote will explore a new chance to build long-term datasets of coastal change through the revalue of historical data with modern techniques of Structure-from-Motion (SfM), GIS analyses and reliable estimation of spatial errors. Reliable geographic data were extracted from different sources available from historical archives (i.e., aerial photos, maps, bathymetric charts, topographic data pre-DGNSS era, military reports, newspapers, etc.) for the coastal site of Dundrum Bay in the Irish Sea (Northern Ireland, UK). Historical maps and aerial photographs were analysed to extrapolate 23 shorelines and quantify almost 200 years of shoreline change (1833 to 2020). Vertical aerial photographs were processed thorough SfM techniques to build 3D models and calculate volumetric change of the extensive local coastal dune for the last six decades (1963 to 2020). Geomorphic changes were discussed considering local forcing parameters (wave, wind, water level) from historical or hindcasted datasets available (1901 to 2020). Advantages and limitations of each ‘revaluation’ technique of the historical data were considered. Several potential errors are associated with historical coastal data; however, their advantage is to provide a quantifiable record when other (more accurate) data sources are not available. In the pre-satellite era (pre-1970’s), historical geographic data are the only available source, and they have also the advantage to be coeval with the increase of human pressure on the coast (post Industrial Revolution and post WWII). Long-term datasets are crucial for decision makers, often divided between warnings from the scientific community and the will to give effective and immediate response to concerns from the coastal communities. Adaptation strategies for coastal environments are often conceived without a long-term perspective which is partially caused by the lack of reliable long-term data. In a current situation of rapidly improving machine learning techniques, reliable long-term datasets are crucial to feed algorithms and better refine models of future coastal behaviour, especially in a context of sea level rise and climate change.