The definition of a shoreline is the boundary between sea and land area. It provides important information on maps, and therefore defining an accurate shoreline location, is important. Studies of shoreline mapping have increased especially during the 2010’s. There are many different methods, and the use of remote sensing images has been popular in studies. The aim of this thesis is to examine the usage of satellite and aerial imagery for extracting shoreline in the Archipelago Sea. The study area is in the southern Archipelago Sea, and the land areas are rock islets. This thesis focuses on shoreline extraction for the purposes of nautical charts.

The used datasets are Sentinel-2 imagery (10 m spatial resolution) and aerial imagery (0,5 m spatial resolution) by the National land survey of Finland (NLS). When mapping the shoreline, sea levels need to be taken into consideration. This affected the selection of Sentinel-2 image used. I chose water level data collected from the Föglö measuring station, located in Åland. I chose the Sentinel-2B image taken on the 23rd of April in 2020, when the mean water was measured at -9 millimeters.

The three-band aerial image classification was done with random forest (RF) algorithm. The classification results were converted from raster to polygon, from which the shoreline was extracted. For the satellite data, a normalized water index (NDWI) was calculated, and the shoreline was extracted with the threshold value of 0. To examine general errors in the shoreline position, the results were compared to a manually digitized reference shoreline. Shorelines extracted from both aerial and satellite images were geometrically intact and the shoreline was formed for all the islets across the study area. The average distance to the reference shoreline was 8,25 meters in the aerial image results, with the maximum error of 47,56 meters. The equivalent distances for satellite image results were 20,42 meters and 98,09 meters. Both results were also compared to the shoreline used in the Finnish nautical charts, provided by NLS. The aerial image results were closer to NLS shoreline than the satellite image results. The shoreline provided by NLS was also compared to the reference shoreline, which showed that there are some deviant values in the NLS shoreline. The average error distance between the NLS shoreline and the reference shoreline was 5,75 meters, with deviant values up to 74,03 meters. The distance error was smaller between the NLS shoreline, and the reference shoreline, compared to the shorelines made in this thesis. However, the deviant values indicate that there are some errors in the location of the NLS shoreline.

The results showed that the location of shorelines made by NLS is more accurate than the shorelines made in the thesis. However, the results made from aerial images could be used for locating and fixing the errors in the shorelines made by NLS. The results of satellite image classification could theoretically be used in small-scale nautical charts, but the precision of the shoreline is not good enough for large-scale nautical charts. The results of aerial classification could theoretically be used in nautical charts.