红树林
环境科学
大红树
恢复生态学
地理
生态学
遥感
卫星图像
生态系统
生态系统服务
环境资源管理
地理信息系统
红树林生态系统
农林复合经营
自然地理学
景观连通性
自然保护区
可持续发展
作者
Uzair Aslam Bhatti,Muhammad Asif,Bekzod Madaminov,Muzaffar Shojonov,Ramazonov Khusniddin,Yonis Gulzar,Ahmad Hasnain
标识
DOI:10.1080/10549811.2026.2699703
摘要
Mangrove communities possess exceptional capacities for soil and water conservation and ecological protection, and their successional dynamics have long commanded global attention. Guangdong Province, located along the southern coast of China, hosts some of the country’s most extensive mangrove stands. Leveraging the Google Earth Engine (GEE) platform, this study integrates Landsat-5/7/8/9 and Sentinel-2 multi-source imagery spanning 2000–2024 and incorporates multiple ecological indicators to construct a Stacking ensemble machine-learning model that blends Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Multi-Layer Perceptron (MLP). The model was applied to 32 mangrove nature reserves and six core distribution areas in Guangdong. Among the base learners, XGBoost exhibited the highest accuracy, achieving 96.98% on the validation set. The optimized Stacking model subsequently revealed that the total mangrove area in the study region expanded continuously from 3,024.72 ha in 2000 to 7,708.83 ha in 2024—a net increase of 4,684.11 ha and an average annual growth rate of 5.8%. Spatially, the mangrove landscape shifted from “scattered fragmentation” to “contiguous aggregation,” with mean patch size rising from 1.64 ha to 6.04 ha. Across the six representative core zones – Futian (Shenzhen), Qi’ao–Dangan Islands (Zhuhai), Zhanjiang, Huidong (Huizhou), Zhenhai Bay, and Cheng Village (Yangxi) – significant area increases (4.4–9.7% yr−1) and progressive patch enlargement were observed. The seaward margins of these mangroves advanced at mean rates of 7.6–15.2 m yr−1. These findings demonstrate the outstanding effectiveness of ecological restoration efforts in Guangdong’s mangroves and provide a quantifiable “Guangdong Model” for nationwide mangrove conservation and rehabilitation.
科研通智能强力驱动
Strongly Powered by AbleSci AI