Spatio-temporal changes of coastline in Jiaozhou Bay from 1987 to 2022 based on optical and SAR data

In the past 35 years, the natural coastline along Jiaozhou Bay has undergone extensive changes under the influence of human activities, and the coastal wetland area has been drastically reduced. Therefore, it is of great importance to study the spatio-temporal changes of the Jiaozhou Bay coastline, and their trends and causes, for sustainable economic development and the rational utilization of coastal resources. This paper constructed a comprehensive method for extracting the coastline information and change analysis based on long time series remote sensing data. Based on multi-spectral optical data and dual-polarization SAR data, the Normalized Difference Water Index (NDWI) and the Sentinel-1 Dual-polarized Water body Index (SDWI) combined with the Otsu threshold segmentation method were used to automatically extract the spatial distribution of coastline. The U-Net semantic segmentation model was used to classify the land cover types in the land direction of the coastline to count the coastline types. The End Point Rate (EPR) and Linear Regression Rate (LRR) were used to analyze the coastline changes, and the land reclamation was calculated according to the changing trends. The Pearson coefficient was used to study the reasons for the coastline changes. With an average time interval of 5 years, eight coastlines of Jiaozhou Bay in different years were extracted, and the coastline types were obtained. Then, the changes of the coastlines in Jiaozhou Bay from 1987 to 2022 were analyzed. The results show that: 1) Coastline type information provides important information for analyzing the coastline changes in long time series, and coastline information can be effectively extracted using multi-spectral optical data and dual-polarization SAR data. When the resolution of remote sensing data is 30m, the average error of the two types of data is better than one pixel, and the error between the data is about 1-2 pixels. 2) Based on the U-Net model, the overall accuracy of coastline classification using multi-spectral optical data and dual-polarization SAR data is 94.49% and 94.88%, respectively, with kappa coefficients of 0.9143 and 0.8949. 3) In the past 35 years, Jiaozhou Bay area has shown an obvious trend towards the ocean, with an average annual expansion of 16.723m. 4) The coastline of the Jiaozhou Bay area is dynamic. Due to the frequent human activities, the coastline has been reconstructed on a large scale, and the length of artificial coastline has increased significantly. The proportion of artificial coastline length has increased from 33.72% in 1987 to 59.33% in 2022. 5) In the past 35 years, the land reclamation area has reached 41.45km(2), of which Shibei District, Licang District, and Huangdao District are the three most frequent areas, with an area of 34.62 km(2).

Automated summary

This study constructed a comprehensive method for extracting coastline information and analyzing its changes using long time series remote sensing data. The results showed that coastline type information provides important information for analyzing long term coastline changes, and coastline information can be effectively extracted using multi-spectral optical data and dual-polarization SAR data. The coastline of Jiaozhou Bay has shown an obvious trend towards the ocean in the past 35 years, with an average annual expansion rate of 16.723m. The proportion of artificial coastline has increased significantly, reaching 59.33% in 2022. The land reclamation area in the past 35 years has reached 41.45km(2), with Shibei District, Licang District, and Huangdao District being the most frequent areas.