利用多光谱遥感技术定量估算野鸭湖湿地挺水植物的含水量.基于典型挺水植物的实测冠层光谱及其对应样方的叶片含水量和叶面积指数LAI数据, 首先对芦苇和香蒲的地面实测光谱进行重采样, 以模拟WorldView-2影像的光谱, 然后利用模拟光谱分别构建芦苇和香蒲任意两波段反射率组合而成的比值(SR)和归一化差值植被指数(NDVI), 通过分析植被指数与CWC(冠层含水量,Canopy Water Content)的相关关系, 选择与CWC显著相关的植被指数, 并通过单变量线性与非线性拟合的分析方法确定监测不同挺水植物群落的最佳植被指数, 建立估算模型；结合覆盖研究区的WorldView-2高分辨率多光谱影像, 对研究区的挺水植物群落CWC进行反演及制图.结果表明, 基于模拟WorldView-2影像光谱构建的比值(SR)和归一化差值植被指数(NDVI)与CWC的总体相关性较高；SR(8,3)芦苇为估算CWC芦苇的最优植被指数, 估算模型为y=0.005x+0.003, NDVI(8,3)香蒲为估算CWC香蒲的最优植被指数, 估算模型为y=2.461x2-0.313x+0.032, 通过交叉检验, CWC芦苇和CWC香蒲的预测精度分别为87.42%和82.12%, 预测精度较为理想；利用实测数据对反演的CWC空间分布图进行了验证, 通过验证, 芦苇和香蒲影像估算CWC的均方根差(RMSE)分别为0.0048和0.0052, 估算精度分别为83.56%和80.31%, 表明利用WorldView-2高分辨率多光谱影像反演湿地挺水植物群落CWC具有较高的可行性.

Quantitative estimation of emerged plant water content with multi-spectral remote sensing technique is of great significance for emerged plant physiological status and growth trend monitoring. The hyperspectral reflectance of canopy of wetland typical emerged plant(reed and cattail)was measured by Field-Spec 3 wild high-spectrum radiometer. The leaf water content and leaf area index(LAI)of corresponding samples were also measured. First of all, the ground spectral data(reed and cattail)were resampled to simulate the spectral of WorldView-2 imagery, then the simple ratio vegetation index(SR)and normalized difference vegetation index(NDVI)were constructed with arbitrary two band combination from the simulated WorldView-2 spectra, respectively. The correlation between canopy water content(CWC )and vegetation index were analyzed. The estimation models were obtained by using regression and correlation analysis for different emerged plant community. In addition, the research result of ground data was applied to WorldView-2 high resolution multispectral imagery covering the study area, and the CWC of emerged plant community was estimation in spatial scale. The results show that the SR and NDVI constructed by the simulated WorldView-2 spectra had a good overall correlation with CWC. The SR(8, 3)reedwas selected as the optimal vegetation index to estimate the CWCreed, the best models are evaluated and validated as y=0.005x+0.003. The NDVI(8, 3)cattailwas selected as the optimal vegetation index to estimate the CWCcattail, the best models were evaluated and validated as y=2.461x2-0.313x+0.032. According to two K-fold cross validation examination, these estimation models have the satisfactory prediction accuracy. The prediction accuracy of CWCreedwas 87.42% and the prediction accuracy of CWCcattailwas 82.12%. Furthermore, based on the research result of ground data, we made use of WorldView-2 high resolution multispectral imagery to map the CWC of different emerged plant community. According to the examination of measured data, the estimation RMSE of CWCreedand CWCcattailfrom imagery were 0.0048 and 0.0052, respectively. The estimation accuracy were 83.56% and 80.31%, respectively. It was demonstrated that using WorldView-2 high resolution multispectral imagery to estimate the CWC of wetland emerged plant community has a high feasibility.