Coastal migration via satellite data (450k-650k)

Thanks to the public availability of satellite data (optical imagery of ESA Sentinel 2 and NASA Landsat 5, 7 & 8 with pixel resolutions of 10-30 metres and a revisit time of 1 to 2 weeks) and new analytical tools for processing big data (such as the Google Earth Engine), the EMODnet Geology team in collaboration with Deltares and TNO (Geological Survey of the Netherlands) were able to look at shoreline migration in a new way. Scripts for automated detection of the land-water boundary were used to separate land from water in annual image composites for the period 2007-2017. During this process, data points were generated for each part along the European shoreline. These points were then averaged by year and analysed for a decadal period. Visualising pan-European shoreline change means making choices, like defining a stable shoreline for example. A mean rate of 0.5 metre per year was chosen, though this rate depends on the landscape: granite cliffs for example shows less decadal dynamics compared to a sandy barrier island. The spatial resolution of the method, depending on the pixel resolution of the individual satellite images which is about 10 metres, is still limiting. Validations of abovementioned method have shown that the method is less accurate in case of bluffs, cliffs and muddy coasts, and as such further validations will need to take place. EMODnet Geology hopes that by releasing the satellite-based dataset now, coastal experts and other end users will be able to discover and communicate possibilities and limitations of automated methods for the extraction of shoreline position and quantification of annual to decadal change. To help in this process, a companion map showing shoreline migration on the basis of field data and expert is made available, thereby facilitating a first-order comparison.得益于欧洲空间局（European Space Agency, ESA）Sentinel 2、美国国家航空航天局（National Aeronautics and Space Administration, NASA）Landsat 5、7及8的公开卫星光学影像（像素分辨率10-30米，重访周期1至2周），以及谷歌地球引擎（Google Earth Engine）等大数据处理新型分析工具的支撑，欧洲海洋观测与数据网络地质工作组（EMODnet Geology）与Deltares、TNO（荷兰地质调查局，Geological Survey of the Netherlands）合作，得以以全新视角开展海岸线迁移研究。研究团队采用自动化水陆边界检测脚本，对2007至2017年的年度影像合成数据开展陆地与水体分离处理，在此过程中为欧洲海岸线各分段生成数据点，随后按年份对这些点进行平均化处理，并开展年代际尺度的分析。可视化全欧海岸线变化需做出若干设定，例如明确稳定海岸线基准。研究最终选取年均0.5米的平均变化速率作为参考，但该速率因地貌类型而异：相较于砂质堡岛，花岗岩海岸崖的年代际动态显著偏弱。该方法的空间分辨率仍受限于单幅卫星影像约10米的像素分辨率，存在一定局限。对上述方法的验证结果显示，其在陡崖、海岸崖及淤泥质海岸场景下的精度欠佳，因此仍需开展进一步验证工作。欧洲海洋观测与数据网络地质工作组（EMODnet Geology）希望通过此次发布这套基于卫星影像的数据集，助力海岸带专家及其他终端用户深入探究并分享自动化海岸线位置提取、年际至年代际变化量化方法的优势与局限。为辅助该项工作，本次同步发布了一套基于实地观测数据与专家经验生成的海岸线迁移配套地图，以支持初步的对比验证。