An Improved Submerged Mangrove Recognition Index-Based Method for Mapping Mangrove Forests by Removing the Disturbance of Tidal Dynamics and S. alterniflora

Abstract

Currently, it is a great challenge for remote sensing technology to accurately map mangrove forests owing to periodic inundation. A submerged mangrove recognition index (SMRI) using two high- and low-tide images was recently proposed to remove the influence of tides and identify mangrove forests. However, when the tidal height of the selected low-tide image is not at the lowest tidal level, the corresponding SMRI does not function well, which results in mangrove forests below the low tidal height being undetected. Furthermore, Spartina alterniflora Loisel (S. alterniflora) was introduced to China in 1979 and rapidly spread to become the most serious invasive plant along the Chinese coastline. The current SMRI has failed to distinguish S. alterniflora from submerged mangrove forests because of their similar spectral signatures. In this study, an SMRI-based mangrove forest mapping method was developed using the time series of Sentinel-2 images to mitigate the two aforementioned issues. In the proposed method, quantile synthesis was applied to the time series of Sentinel-2 images to generate a lowest-tide synthetic image for creating SMRI to identify submerged mangrove forests. Unsubmerged mangrove forests were classified using a support vector machine, and a preliminary mangrove forest map was created by merging them. In addition, S. alterniflora was distinguished from the mangrove forests by analyzing their phenological differences. Finally, mangrove forest mapping was performed by masking S. alterniflora. The proposed method was applied to the entire coastline of the Guangxi Province, China. The results showed that it can reliably and accurately identify submerged mangrove forests derived from SMRI by synthesizing low- and high-tide images using quantile synthesis, and the differentiation of S. alterniflora using phenological differences results in more accurate mangrove mapping. This work helps to improve the accuracy of mangrove forest mapping using SMRI and its feasibility for coastal wetland monitoring. It also provides data for sustainable management, ecological protection, and restoration of vegetation in coastal zones.

Mangrove forest mapping from object-oriented multi-feature ensemble classification using Sentinel-2 images

“…We collected study areas (a, b) and compared the results of OMEC method with two reference maps (Figure 12), namely the 10 m resolution global mangrove forest dataset (HGMF_2020) provided by Mingming Jia (Jia et al, 2023) and the 1 m resolution China mangrove forest dataset (SMRI) provided by Qing Xia (Xia et al, 2022). Our results show similarities in terms of the mangrove areas with both HGMF_2020 and SMRI datasets in study areas (a, b).…”

Convolution Feature Inference-Based Semantic Understanding Method for Remote Sensing Images of Mangrove Forests

“…In recent years, scholars have further explored deep learning research, which provides a highly positive effect for the semantic segmentation of remote sensing images and meets the accuracy requirements of computer vision applications [13][14][15][16][17][18][19][20][21]. The fully convolutional neural network was proposed in 2015 [22].…”