Study of the spatial and temporal distributions of tidal flat surface sediment based on multitemporal remote sensing

Abstract

Introduction: The ecological environment of tidal flats often changes due to tidal erosion and sedimentation. The distribution of tidal flat surface sediment is a natural reflection of the changes in the external dynamic environment, the spatial and temporal distribution pattern is of great significance. Methods: In this study, the output structure of traditional convolutional neural network is combined with BP neural network. Meanwhile, four phases of Sentinel-2 multispectral images were collected and combined with field data from the Doulonggang tidal flat in Jiangsu Province, China, to construct the sediment composition inversion model. Results: The inversion accuracy was higher than 80% compared with the measured results. According to the inversion result, from 2017 to 2022, the surface sediment particle size of the tidal flat in Jiangsu varied seasonally and was coarse in summer and fine in winter. Additionally, the sediment composition tended to coarsen, showing an interannual change trend of increasing sand content and decreasing clay and silt contents. Discussion: The above change of the sedimentary environment of the tidal flat may be caused by the decrease of fine grained sediment deposition, the introduction of exotic vegetation, the global sea level rise and the influence of human activities.