An automated shoreline detection method using PlanetScope satellite imagery

Multispectral satellite images can survey the surf zone through discretizing the land-sea interface, at a known water level, to monitor recession and accretion rates along the coastline. This shoreline detection method can be enhanced by utilizing the daily return frequency of PlanetScope data, allowing a higher temporal resolution of the observed shorelines. Similar shoreline detection tools, such as CoastSat​(Doherty et al., 2022; Vos et al., 2019)​, discretize the land-sea interface by thresholding the image using a single index, such as NDWI (normalized difference water index) ​(McFeeters, 1996)​ and contouring the image at this threshold. Presented here is an alternative approach. In using several multilayer perceptrons (MLP) acting together, each pixel’s probability of being classed as land or sea is calculated. The final shoreline contour is then probabilistically defined whithout the use of manual threshold. The advantage of this method is that it allows for spatial variability within satellite bands, for regions of shadow and geographical features, to still be correctly discretized. It also allows for further use case beyond just sandy beaches, due to the implementation of multiple indices allowing identification of different classes that could be interfacing with the sea. Characteristically, apart from the usual NDWI and NDVI index, we use the RGB and IR bands as well as 24 further band relationships for a total set of 28 indices to train the MLPs. The root mean squared error (RMSE), the distance between the derived shoreline and a height contour relative to the instantaneous water-level, of this method tested at Seaford UK for cloud cover <90% is ~7m. 

NDWI=Green−IRGreen+IR">NDWI=Green−IRGreen+IRNDWI=Green−IRGreen+IR

 

NDVI=IR−RedIR+Red">NDVI=IR−RedIR+RedNDVI=IR−RedIR+Red

 

​​Doherty, Y., Harley, M. D., Vos, K., & Splinter, K. D. (2022). A Python toolkit to monitor sandy shoreline change using high-resolution PlanetScope cubesats. Environmental Modelling and Software, 157. https://doi.org/10.1016/J.ENVSOFT.2022.105512 

​McFeeters, S. K. (1996). The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features. International Journal of Remote Sensing, 17(7), 1425–1432. https://doi.org/10.1080/01431169608948714 

​Vos, K., Splinter, K. D., Harley, M. D., Simmons, J. A., & Turner, I. L. (2019). CoastSat: A Google Earth Engine-enabled Python toolkit to extract shorelines from publicly available satellite imagery. Environmental Modelling & Software, 122, 104528. https://doi.org/10.1016/J.ENVSOFT.2019.104528 

​ ​