Unsupervised monitoring of vegetation in a surface coal mining region based on NDVI time series

• Published in: Environmental science and pollution research international Vol. 29; no. 18; pp. 26539 - 26548
• Main Authors: Yang, Zhen, Shen, Yingying, Li, Jing, Jiang, Huawei, Zhao, Like
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.04.2022; Springer Nature B.V
• Subjects: Aquatic Pollution; Atmospheric Protection/Air Quality Control/Air Pollution; Center of gravity; Centroids; Change detection; China; Clustering; Coal; Coal Mining; Earth and Environmental Science; Ecotoxicology; Energy sources; Environment; Environmental Chemistry; Environmental Health; Environmental Monitoring; Environmental science; Landsat; Landsat satellites; Monitoring; Normalized difference vegetative index; Parameters; Remote sensing; Research Article; Satellite Imagery; Time Factors; Time series; Training; Vegetation; Waste Water Technology; Water Management; Water Pollution Control; China; Shape-based clustering; Surface coal mining; NDVI time series; Remote sensing; Landsat; Vegetation
• ISSN: 0944-1344; 1614-7499
• DOI: 10.1007/s11356-021-17696-9

Surface coal mining causes vegetation disturbance while providing an energy source. Thus, much attention is given to monitoring the vegetation of surface coal mining regions. Multitemporal satellite imagery, such as Landsat time-series imagery, is an operational environment monitoring service widely used to access vegetation traits and ensure vegetation surveillance across large areas. However, most of the previous studies have been conducted with change detection models or threshold-based methods that require multiple parameter settings or sample training. In this paper, we tried to analyze the change traits of vegetation in surface coal mining regions using shape-based clustering based on Normalized Difference Vegetation Index (NDVI) time series without multiple parameter settings and sample training. The shape-based clustering used in this paper applied shape-based distance (SBD) to obtain the distance between time series and used Dynamic Time Warping Barycenter Averaging (DBA) to generate cluster centroids. We applied the method to a stack of 19 NDVI images from 2000 to 2018 for a surface coal mining region located in North China. The results showed that the shape-based clustering used in this paper was appropriate for monitoring vegetation change in the region and achieved 79.0% overall accuracy in detecting disturbance-recovery trajectory types.