Unraveling Multi-Scale dynamics of estuarine wetland vegetation using the multi-resolution analysis wavelet transform and the Landsat time-series

Abstract

Estuarine wetland vegetation (EWV) serves as a critical indicator of ecosystem health but is increasingly threatened by human activities and climate change. Though remote sensing (RS) data provide a holistic view of vegetation change with sufficient spatial resolution, temporal changes in EWV are still ambiguous because of the twisted signals between phenological dynamics and multi-scale external disturbances. To address these knowledge gaps, a method integrated with multi-resolution analysis wavelet transform (MRA-WT) and time-series RS images were utilized in this study to uncover detailed multiple-scale EWV dynamics from the perspectives of vegetation phenology and evolution trends. Our results showed that: 1) The MRA-WT effectively decomposes NDVI at multiple scales, revealing intra- and inter-annual dynamics of EWV in complex environments. 2) In the Jiuduansha (JDS) wetland, the phenological characteristics and evolution processes of EWV are determined by vegetation type and geographical location. 3) The EWV in the JDS wetland remained stable with gradual growth from 2000 to 2021, despite the impacts of sediment reduction, estuarine projects, and invasive species. However, a declining trend post-2018 serves as a warning sign of potential ecosystem issues. Moreover, higher spatial and temporal resolutions enable more precise tracking of fine-scale details in the dynamics of EWV with the proposed method.

Keywords

• Estuarine wetland
• Vegetation dynamics
• Multi-resolution analysis
• Remote sensing
• Time-series
• Human activity