• 关键词：无人机  多光谱  总磷  悬浮物浓度  浊度  遥感  反演

• 基金项目：国家自然科学基金两化融合项目（No.U1809208）；浙江省重点研发计划项目（No.2015C03008）；浙江省自然科学基金和青山湖科技城联合基金（No.LQY18C160002）

• 刘彦君
• 1. 浙江农林大学信息工程学院, 杭州 311300;2. 浙江省林业智能监测与信息技术研究重点实验室, 杭州 311300;3. 林业感知技术与智能装备国家林业局重点实验室, 杭州 311300

• 夏凯
• 1. 浙江农林大学信息工程学院, 杭州 311300;2. 浙江省林业智能监测与信息技术研究重点实验室, 杭州 311300;3. 林业感知技术与智能装备国家林业局重点实验室, 杭州 311300

• 冯海林
• 1. 浙江农林大学信息工程学院, 杭州 311300;2. 浙江省林业智能监测与信息技术研究重点实验室, 杭州 311300;3. 林业感知技术与智能装备国家林业局重点实验室, 杭州 311300

• 方益明
• 1. 浙江农林大学信息工程学院, 杭州 311300;2. 浙江省林业智能监测与信息技术研究重点实验室, 杭州 311300;3. 林业感知技术与智能装备国家林业局重点实验室, 杭州 311300

• 摘要：总磷（TP）、悬浮物浓度（SS）、浊度（TUB）3种水质参数可以直接通过遥感反演得到，常用于评价区域水环境的污染状况.以浙江农林大学东湖为研究对像，使用无人机携带多光谱传感器（Mica Sense Red Edge）获取多光谱影像，进而提取16个光谱参数，分别构建东湖水域TP、SS、TUB的反演模型.结果表明：光谱参数V5（NIR 0.770~0.890 μm）与TP、SS相关性显著（r分别为0.470、-0.537，p<0.05），V4（0.670~0.760 μm）与TUB相关性显著（r=0.486，p<0.05）.在建立的TP反演模型中，指数函数模型精度最高，决定系数R²为0.7829；在建立的SS、TUB反演模型中，多项式函数模型精度最高，决定系数R²分别为0.7503、0.7334.经检验，TP、SS、TUB模型估测值与实测值线性拟合曲线的决定系数R²分别为0.7374、0.8978、0.6726，满足水质要素反演的精度要求.最后利用建立的模型，结合多光谱影像数据，建立了东湖水域各参数的空间分布图，实现了水质参数的可视化，可为小微水域的污染防治提供技术支撑.

• Abstract：Total phosphorus (TP), suspended sediments concentration (SS), and turbidity (TUB) are important parameters that describe the water quality and have been proven to be promising tool for the regional water environment evaluation. This paper takes the Donghu Lake of Zhejiang Agriculture and Forestry University as the research object, obtains the multi-spectral images by using the multi-spectral sensor (Mica Sense Red Edge) carried by UAV, and then extracts 16 spectral parameters, which are used to construct the inversion models of TP, SS and TUB in the Donghu lake waters. The results show that the spectral parameters V5 (NIR 0.770~0.890 μm) are highly correlated with TP and SS, which are 0.47 and -0.537, respectively, V4(0.670~0.760 μm)is highly correlated with TUB, which is 0.486. For all the three parameters, the significant differences were smaller than 0.05. In the established TP inversion model, the exponential function model has the highest precision, the determination coefficient R² is 0.7829; in the SS and TUB model, the polynomial function model has the highest precision, the determination coefficient R² are 0.7503 and 0.7334 respectively. The test results show that the determination coefficients R² of the estimated and measured values of linear fitting curves of TP, SS and TUB are 0.7374, 0.8978 and 0.6726 respectively, which meet the accuracy requirements of water quality factor inversion. At last, by using the established models, combined with multi-spectral image data, the spatial distribution map of the parameters concentration in the Donghu Lake waters has been established, and the visualization of water quality parameters has been realized, which provides technical support for the prevention and control of pollution in small and micro waters.

• Key words：UAV  multispectral  total phosphorus  suspended sediments concentration  turbidity  remote sensing  inversion

• 
  摘要点击次数： 1106 全文下载次数： 1462