Development of a tidal flat recognition index based on multispectral images for mapping tidal flats

Abstract

Tidal flats worldwide are facing the dual challenges of tidal reclamation and climatic changes, with China being particularly affected. The distinctive and vital ecosystem functions and services offered by tidal flats in coastal regions necessitate the accurate and precise mapping of this specific land cover type. Current methods for mapping tidal flats rely on a large number of sample points and intricate, time-consuming classification algorithms. However, these methods are inadequate for large-scale tidal flat mapping due to their low computational efficiency and limited general applicability. This study proposes a Tidal Flat Recognition Index (TFRI) based on multispectral images. This index is designed to enhance the distinction in water content between tidal flats and other land cover types by incorporating near-infrared (NIR) bands. Four representative tidal flat areas in China (i.e., Beibu Gulf, Guangdong Province, Zhangpu County, Fujian Province, Jiaozhou Bay, Shandong Province and Seaside new area, Tianjin City) were chosen as the study areas, and sample datasets were generated through field surveys and high-resolution images obtained from Google Earth. The separability of the TFRI for tidal flats was discussed, and TFRI exhibited better ability to distinguish tidal flats. TFRI was applied to tidal flat extraction in the four study areas using Sentinel-2 images, and the extraction results were compared with existing tidal flat maps and the results of the object-oriented method. The results demonstrated that TFRI exhibited superior overall accuracy and Kappa coefficient compared to the reference tidal flat maps. Further tests indicated that TFRI was also applicable to other remotely-sensed images, including NIR bands from Gaofen-2, Landsat, and Zhuhai-1 satellites. These results suggest that TFRI is adaptable to various remotely-sensed images and various types of tidal flat extraction. Furthermore, this index exhibits remarkable potential for effective application in tidal flat mapping.