Canada’s coastal dynamics from multi-decadal Landsat

An existing national-scale Landsat dynamic surface water dataset over Canada is leveraged to map coastlines in 5-year periods centred on 1987, 2003 and 2019. The spatial and temporal consistency in shoreline locations mapped due to tide level variation is verified using simulations with tide gauge data, and high-resolution WorldView-2 scenes are used to benchmark shoreline positional accuracy. Shoreline change is mapped between periods, and erosion, accretion, and stability are calculated for Canadian Arctic and southern coastal regions. Simulations indicate that the variance in mapped shoreline position due to random sampling of clear-sky Landsat pixels is similar or less than that of tide modelling used in other studies. The shoreline positional accuracy is within one Landsat pixel with a consistent seaward bias, which is comparable to benchmark results for Landsat shoreline extraction algorithms over a micro-mesotidal site with a wide intertidal zone. Between 1987 and 2003, accretion dominated in the Arctic and erosion in southern Canada. While erosion continued to dominate between 2003 and 2019 in southern regions, the Arctic region switched from being accretional to net erosional, possibly due to climatic effects overtaking isostatic rebound. A comparison of the derived coastal dynamics and the CanCoast Sensitivity Index (CSI), which represents an aggregate measure of coastal physical susceptibility to climate change, shows strong monotonic relationships in both Arctic and southern study regions, confirming the consistency between both datasets.