Mapping tidal flat topography using time-series Sentinel-2 images and ICESat-2 data: A case study in Cixi City

TL;DR

This study combines Sentinel-2 time-series imagery and ICESat-2 data to accurately map and model tidal flat topography and elevation in Cixi City, demonstrating high accuracy and the importance of temporal resolution.

Contribution

It introduces a novel method integrating Sentinel-2 and ICESat-2 data for tidal flat elevation inversion, improving accuracy in dynamic coastal environments.

Findings

Strong correlation (R2=0.85) between exposure frequency and elevation.

Tidal flat area in Cixi City is 115.81 km² with 95.36% accuracy.

Elevation ranges from -0.42 to 2.73 meters with MAE of 0.24 meters.

Abstract

Tidal flat topography provides crucial insights for understanding tidal flats and their dynamic evolution. However, the wide-ranging and rapidly changing nature of tidal flats, which are periodically submerged in shallow water, pose challenges for many current monitoring methods in terms of both efficiency and precision. In this study, we considered the dynamic process of tidal flat submergence and utilized time-series Sentinel-2 images on Google Earth Engine (GEE) to calculate the tidal flat exposure frequency. This information was used to determine the spatial extent of the tidal flats, and subsequently, by employing ICESat-2 data, we established a 1D-linear regression model based on elevation and frequency values, which realizes the inversion of the tidal flat elevation within Cixi City. The study shows the following: (1) the tidal flat exposure frequency and ICESat-2 elevation data exhibit a strong positive correlation (R2=0.85); (2) the tidal flat area within Cixi City is 115.81 km2, and the overall accuracy is 95.36%; and (3) the elevation range of the tidal flats in the study area is between -0.42 and 2.73 m, and the mean absolute error (MAE) is 0.24 m. Additionally, we consider that the temporal resolution of remote sensing imagery plays a crucial role in determining the accuracy of the elevation inversion, and we found that higher tidal flats exhibit better inversion accuracy than lower tidal flats.