Deep learning-driven land cover monitoring and landscape ecological health assessment: A dynamic study in coastal regions of the China–Pakistan Economic Corridor from 2000 to 2023

作者：Xu Chen, Juanle Wang, Yamin Sun, Meng Liu, Jingxuan Liu, Meer Muhammad Sajjad · 发表于：Ecological Indicators · 年份：2024 · DOI：10.1016/j.ecolind.2024.112860 · 被引用次数：5 · 研究领域：Land Use and Ecosystem Services、Remote Sensing and Land Use、Remote Sensing in Agriculture

• Integrated labels enable rapid creation of large datasets for training deep learning models. • A multi-scale convolutional neural network was used to extract 23-year land cover in the coastal area of CPEC. • The landscape ecological health of the coastal regions of CPEC from 2000 to 2023 was evaluated by VOR model. • The year 2010 showed the best ecological health status in the coastal areas of CPEC. • The coastal regions of the CPEC have consistently been “Unhealthy,” but there is a recent trend of improvement. The coastal regions of the China–Pakistan Economic Corridor (CPEC) are crucial links for the “21st Century Maritime Silk Road”. Nonetheless, this region is facing significant ecological challenges due to natural disasters and intensive human activity. To effectively monitor and assess the ecological health of these critical coastal zones, this study employed integrated labels and a deep learning model to obtain land cover data spanning from 2000 to 2023. It then constructed a vigour-organisation-resilience (VOR) model with 12 assessment indicators to evaluate the landscape ecological health of this region. The evaluation results showed distinct spatial patterns. Gwadar and Ormara’s “Bare land” areas remained “Sick,” while Karachi and Lower Indus’ “Impervious surfaces” were “Unhealthy” with minimal fluctuations. The Lower Indus region saw “Sub-healthy” expansion with increased “Crops” areas. Lasbela was “Healthy,” dominated by shrub-based “Other vegetation,” and the ...