小波分析在大豆叶绿素含量高光谱反演中的应用

Application of Wavelet Transformation in In-situ Measured Hyperspectral Data for Soybean LAI Estimation

实测了不同水肥耦合作用下,大豆冠层高光谱反射率与叶绿素含量数据,并对光谱反射率、微分光谱与叶绿素含量进行了相关分析;采用叶绿素A与叶绿素B诊断波段构建了特定植被指数,对叶绿素A、叶绿素B进行了回归分析;采用小波分析对采集的光谱反射率数据进行了能量系数提取,并以小波能量系数作为自变量进行了单变量与多变量回归分析,对叶绿素含量进行估算。经分析发现,叶绿素A、B与光谱反射率在可见光与近红外波段的相关系数的变化趋势基本一致——在可见光谱波段呈负相关,近红外波段呈正相关,红边处相关系数由负变正。特定色素植被指数可以提高大豆叶绿素估算精度(R2>0.73);小波能量系数回归模型可以进一步提高大豆叶绿素含量的估算水平,以一个特定小波能量系数作为自变量的回归模型,叶绿素A其确定性系数R2为0.76,叶绿素B为0.78;以4变量与9变量回归分析结果表明:叶绿素A实测值与预测值的线性回归确定性系数R2分别大于0.85、0.89;叶绿素B实测值与预测值的线性回归确定性系数R2分别为0.86、0.90。

Soybean canopy reflectance data collected with ASD spectroradiometers （350-1050nm） which were cultivated in water-fertilizer coupled control conditions, and chlorophyll-A and chlorophyll-B content data were collected simultaneously. First, correlation between reflectance, derivative reflectance against chl-A and chl-B were analyzed. Secondly, RVI, RARSa and PSSRb regressed against chl-A and chl-B. Finally, wavelet energy coefficients of spectral reflectance were extracted, and then those energy coefficients regress against chl-A, chl-B with different method. It was found that soybean canopy reflectance showed a negative relation with chl-A and chl-B, while it showed a positive relation with chl- A and chl-B in near infrared region. Reflectance derivative has an intimate relation with chl-A and chl-B in blue, green and red edge spectral region, and got maximum correlation coefficient in red edge region. Chlorophyll specified absorption vegetation index have intimate relation with chl-A and chl-B, with regression determination coefficient R2 greater than 0.736. Regression model established with single wavelet energy coefficient obtained and determination coefficient R2 greater than 0.76 and 0.78 for chl-A and chl-B respectively. Step wise regression with 4 and 9 wavelet energy coefficients were also done, the result showed that the relation between regression model, with 4 and 9 independents, predicted chl-A and measured chl-A with a determination coefficient R2 of 0.85 and 0.89 respectively, however, for chl-B, the model predicted chl-B and measured chl-B with a determination coetticient ItLot O.~O an^l o.pu respectively. By above analysis, it indicated that wavelet transform can be applied to ,,9-s,,？L, collected hyperspectral data processing and model establishing with quite accurate model prediction, and in the future, wavelet transform still should be applied to hyperspectral data for other vegetation biophysical and biochemical parameters inversion.