生态学报2026，Vol.46Issue(9)：4742-4758,17.DOI:10.20103/j.stxb.202508042028
基于遥感与机器学习的辽河口盐沼湿地时空分异及驱动机制研究
Spatiotemporal patterns and driving mechanisms of salt marshes in the Liaohe Estuary based on remote sensing and machine learning
摘要
Abstract
This study comprehensively investigates the long-term dynamics and spatial heterogeneity of Liaohe Estuary,which is one of the most important salt marsh wetlands and located at a relatively high latitude in China.Recognizing the dynamics of salt marshes to both natural and anthropogenic pressures,we proposed an integrated technical framework that combined time-series remote sensing analysis,phenological constraint application,tidal correction,and machine learning approaches.Leveraging the computational power of the Google Earth Engine(GEE)cloud platform,multi-temporal Landsat imagery spanning the period from 1990 to 2024 was processed,enabling consistent monitoring over more than three decades.The spatio-temporal evolution of estuarine salt marshes was quantitatively examined using the Random Forest(RF)classification algorithm for accurate wetland mapping and the Digital Shoreline Analysis System(DSAS)model for assessing shoreline change rates and patterns.The results revealed distinctive dynamic phases and pronounced spatial differences in salt marsh evolution:(1)Salt marshes in the Liaohe Estuary exhibited a fluctuating trajectory,with an average annual area of 42.71km2.From 1990 to 2003,a notable degradation phase occurred,characterized by a mean annual decrease of 1.25km2.In contrast,the period from 2004 to 2024 demonstrated the signal of ecological recovery,with an average annual growth of 1.48km2.(2)Spatially,the western bank salt marshes were approximately two to three times more extensive than those on the eastern bank,following a similar decline-recovery pattern and maintaining an average area of 22.43km2.Conversely,eastern bank salt marshes displayed a fluctuating but overall upward trajectory,averaging 13.71km2 with a growth rate of 0.28km2/a.(3)Shoreline change analysis further illustrated asymmetric morphodynamical processes:the western shoreline experienced intense,spatially heterogeneous accretion-erosion dynamics,where 80.58%of transects showed accretion and 19.42%underwent erosion.By comparison,the eastern shoreline was dominated by progressive accretion,with 85.50%of transects experiencing significant siltation.In dynamic sectors,such as segment c1 on the western bank,shoreline changes were strongly associated with reclamation-restoration processes.(4)Driving mechanism analysis indicates that human activities,such as reclamation and restoration projects,were the primary forces shaping salt marsh change.Although increased upstream sediment discharge in recent years could potentially enhance wetland expansion,its effect was mitigated by relatively weak tidal energy and anthropogenic disruption of sediment transport pathways.Ecological restoration initiatives,notably"returning aquaculture ponds to wetlands",mitigated certain negative impacts caused by sea-level rise.Nevertheless,future risks were considerable,particularly due to sediment supply insufficiency,accelerating sea-level rise,and cumulative historical disturbances.This research reveals pronounced spatio-temporal heterogeneity and highlights the complex interplay between natural and human drivers in the dynamics of estuarine salt marsh.The findings provide critical scientific guidance for achieving adaptive restoration strategies and sustainable coastal management in the Liaohe Estuary and other comparable estuarine regions worldwide.关键词
盐沼变迁/机器学习/谷歌地球引擎(GEE)/围垦/退养还湿/海平面上升/辽河口Key words
salt marsh dynamics/machine learning/Google Earth Engine(GEE)/reclamation/wetland restoration/sea-level rise/Liaohe Estuary引用本文复制引用
黄栋,庞天一,戴志军..基于遥感与机器学习的辽河口盐沼湿地时空分异及驱动机制研究[J].生态学报,2026,46(9):4742-4758,17.基金项目
国家重点研发计划政府间国际科技创新合作重点专项(2023YFE0121200) （2023YFE0121200）
辽宁省教育厅科学研究一般项目(LJ212510165009) （LJ212510165009）
