Quantifying coastal cliff retreat using Structure-from-Motion photogrammetry in the Minas Basin, Nova Scotia

Abstract

Rates of retreat of cliffed coastlines are increasing around the globe, but the causes of observed increases are unclear. Lack of understanding complicates assessment of the risks to coastal infrastructure associated with rising sea levels. The frequencies and magnitudes of the coastal cliff erosion events that combine to produce overall, longer-term rates of retreat historically have been poorly constrained because observations generally were not possible on the time scales associated with major terrestrial and oceanographic forcings, which include, among others, heavy precipitation, repeated freeze-thaw cycles, and large waves associated with high water levels. To address mis-match between observational time scales and time scales of event-driven erosion, a two-part study is undertaken. In the first part, analysis of archived aerial photographs is used to produce approximately decadal estimates of coastal retreat over the past 8 decades at Thomas’ Cove Coastal Reserve in Economy, Nova Scotia. In the second part, Structure-from-Motion (SfM) analysis is used to survey actively eroding cliffs at Thomas’ Cove. On multiple visits between June 2018 and October 2019, photographs were taken with a camera mounted on a small Remotely Piloted Aircraft (sRPA). Images are used to create three-dimensional point-clouds, Digital Surface Models (DSMs) and orthomosaics with Pix4D, a commercially available SfM software package. DSMs and orthomosaics are used to map the position of the cliff top through the observation period. Calculated short-term retreat rates are compared to decadal-scale retreat rates for Thomas Cove determined from archived aerial photos and previous studies in the area. These rates indicate that present retreat rates are statistically similar to long-term averages and higher than rates observed over the past four decades. Additionally, volume change is calculated using GIS techniques and found to be consistent with values calculated in the same region by Wilson et al, 2017.