Scholarlink

<p>Lakeside wetlands are increasingly exposed to overlapping pressures from eutrophication and land conversion, yet quantitative evidence linking restoration to both carbon (C) storage recovery and water-quality improvement remains limited. Here we use an integrated analytical framework combining high‐resolution remote sensing, process‐based C accounting, and machine‐learning forecasting to assess a 15‐year restoration trajectory of the Shibalianwei Wetland in China’s Chaohu Lake Basin. This approach integrates ecosystem service modeling with machine-learning predictive analytics to capture long-term, non-linear restoration dynamics beyond traditional short-term monitoring. Supervised decision‐tree classification (overall accuracy = 91% and Kappa = 0.88 across all assessment years) and InVEST modeling of four C pools revealed that total C density remained stable from 2010 to 2017 (48.69 to 48.7 t ha⁻¹) but nearly doubled by 2024 (90.2 t ha⁻¹), coinciding with vegetation and water expansion and declines in built‐up and cultivated land. XGBoost (R<sup>2</sup> = 0.92, RMSE = 0.002) modeling identified NDVI and land‐cover transitions as dominant predictors of post‐restoration gains, while correlation analyses showed that biomass recovery paralleled reductions in total nitrogen (N), phosphorus (P), and organic load. Our findings confirm that anthropogenic land-use transitions, rather than climatic variability, were the primary drivers of these synergistic C and water‐quality benefits. Together these results demonstrate that large‐scale wetland rehabilitation can deliver synergistic ecological outcomes. The integrated framework provides a transferable blueprint for monitoring ecosystem multifunctionality and supports global efforts to reconcile C neutrality with aquatic ecosystem resilience.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Synergistic Climate and Water-Quality Benefits from Wetland Restoration: A Long-Term Assessment Using Integrated Remote Sensing and Machine Learning

  • 期刊论文

  • 发表日期：

    2026年8月1日

查看全文
  • Sajida Sajida,
  • Zifang Wang,
  • Changming Yang

摘要

Lakeside wetlands are increasingly exposed to overlapping pressures from eutrophication and land conversion, yet quantitative evidence linking restoration to both carbon (C) storage recovery and water-quality improvement remains limited. Here we use an integrated analytical framework combining high‐resolution remote sensing, process‐based C accounting, and machine‐learning forecasting to assess a 15‐year restoration trajectory of the Shibalianwei Wetland in China’s Chaohu Lake Basin. This approach integrates ecosystem service modeling with machine-learning predictive analytics to capture long-term, non-linear restoration dynamics beyond traditional short-term monitoring. Supervised decision‐tree classification (overall accuracy = 91% and Kappa = 0.88 across all assessment years) and InVEST modeling of four C pools revealed that total C density remained stable from 2010 to 2017 (48.69 to 48.7 t ha⁻¹) but nearly doubled by 2024 (90.2 t ha⁻¹), coinciding with vegetation and water expansion and declines in built‐up and cultivated land. XGBoost (R2 = 0.92, RMSE = 0.002) modeling identified NDVI and land‐cover transitions as dominant predictors of post‐restoration gains, while correlation analyses showed that biomass recovery paralleled reductions in total nitrogen (N), phosphorus (P), and organic load. Our findings confirm that anthropogenic land-use transitions, rather than climatic variability, were the primary drivers of these synergistic C and water‐quality benefits. Together these results demonstrate that large‐scale wetland rehabilitation can deliver synergistic ecological outcomes. The integrated framework provides a transferable blueprint for monitoring ecosystem multifunctionality and supports global efforts to reconcile C neutrality with aquatic ecosystem resilience.