| 摘要: |
| 浮叶/挺水植被是重要的湖泊水生植被类群,其面积/覆盖度是湖泊生态健康评估及固碳潜力核算的重要参数,大面积、精确获取湖泊中浮叶/挺水植被面积/覆盖度及其变化信息,对湖泊生态修复及碳汇核算至关重要。卫星遥感是获取湖泊浮叶/挺水植被面积/覆盖度最有效的手段。然而,传统的卫星监测方法只能获取卫星像元内是否存在水生植被,无法定量估算像元内植被的覆盖度,从而无法定量、精准地获取湖泊中浮叶/挺水植被的面积/覆盖度。围绕该问题,本研究利用无人机、Sentinel-2 MSI和Landsat 8 OLI遥感数据,基于XGBoost建模方法,通过逐步升尺度的思路,分别构建了基于Sentinel-2 MSI和Landsat 8 OLI像元尺度的浮叶/挺水植被覆盖度定量估算模型,并成功地应用于四大淡水湖泊。结果表明：基于Sentinel和Landsat的估算模型测试集R2分别是0.95和0.97,RMSE分别是7.85%和4.80%,MAE分别是5.35%和3.35%。1990-2022年,鄱阳湖和洞庭湖呈显著的增加趋势(p ＜ 0.01),太湖呈先增后减的趋势(p ＜ 0.01),洪泽湖有不显著的增加趋势(p = 0.59)。模型在四大淡水湖的长时序应用,证明了模型的稳健性和应用潜力,预期能为湖泊生态系统碳汇核算和固碳潜力评估提供方法和数据支撑。 |
| 关键词: 湖泊 浮叶/挺水植被 覆盖度 卫星像元尺度 遥感 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金面上项目“湖泊沉水植被区水体二氧化碳分压（pCO2）遥感估算研究”（编号：42271377）；云南省省市一体化专项 |
|
| Estimation model and application of satellite pixel-scale floating/emergent aquatic vegetation coverage in lakes |
|
qinhaitao,luojuhua,xuying,Zhangchunyu,xuyatian,Mengdi,Hefeng,Lulu
|
|
Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences
|
| Abstract: |
| Floating/emergent aquatic vegetation is an important aquatic vegetation group in lakes, and its area/coverage is an important parameter for lake ecological health assessment and carbon sequestration potential accounting. Accurately obtaining the area/coverage of floating/emergent aquatic vegetation over large lake areas and understanding their changes is crucial for lake ecological restoration and carbon sink accounting. Satellite remote sensing is the most effective means to obtain the area/cover of floating/emergent aquatic vegetation in lakes. However, traditional satellite monitoring methods can only obtain the presence or absence of aquatic vegetation within satellite pixels, and cannot quantitatively estimate the coverage of aquatic vegetation in the pixels Consequently, it is impossible to quantitatively and accurately obtain the area/coverage of floating/emergent aquatic vegetation in lakes. To address this issue, we utilized UAV, Sentinel-2 MSI, and Landsat 8 OLI remote sensing data. Using the XGBoost modeling method, we developed quantitative estimation models for floating and emergent aquatic vegetation coverage at the Sentinel-2 MSI and Landsat 8 OLI pixel scales through a stepwise upscaling approach, and successfully applied it to China’s four largest freshwater lakes. The results showed that the test sets of the two estimation models based on Sentinel and Landsat images had R2 of 0.95 and 0.97, RMSE of 7.85% and 4.80%, and MAE of 5.35% and 3.35%, respectively. From 1990 to 2022, Lake Dongting and Lake Poyang showed highly significant increasing trends (p ＜ 0.01), Lake Taihu showed an increasing and then decreasing trend (p ＜ 0.01), and Lake Hongze had a non-significant increasing trend (p = 0.59). The long-term application of the model in the four largest freshwater lakes proved the robustness and application potential of the model, which is expected to provide methodological and data support for the accounting of carbon sinks in lake ecosystems and the assessment of carbon sequestration potential. |
| Key words: Lakes Floating/emergent aquatic vegetation Coverage Satellite pixel scale Remote sensing |
