基于高光谱的苹果花磷素含量监测模型

在室内条件下，利用ASD FieldSpec 3地物光谱仪，测定了盛花期苹果花的高光谱反射率，在分析苹果花原始光谱反射率及其一阶导数特征的基础上，分析了高光谱反射率与磷素含量间的相关关系，确定其敏感波段，构建了特征光谱参数，并建立了苹果花磷素含量的监测模型.结果表明：苹果花磷素含量与350～370 nm、670～1385 nm、1620～1760 nm波段的原始光谱反射率以及波段500～520 nm光谱反射率的一阶导数呈极显著正相关，与波段670～730
nm一阶导数呈极显著负相关；光谱参数DVI(936,676)、DVI(977,676)、NDVI(936,676)、NDVI(977,676)与苹果花磷素含量的相关性较好，其相关系数均达到0.77以上，由此建立了以4种光谱参数为自变量的磷素含量监测模型，其中，以NDVI(936,676)为自变量构建的监测模型具有最大的决定系数（R²=0.9385）、最小的均方根误差（RMSE=0.6883）和最小的相对误差（RE=7.6%），预测精度达到92.4%，为最佳监测模型.

Monitoring models for phosphorus content of apple flowers based on hyperspectrum

By using ASD FieldSpec 3 field spectrometer,the hyperspectral reflectance of apple flowers at full-blooming stage was measured under laboratory conditions,and the correlations of the hyperspectral reflectance with the phosphorus content of apple flowers were analyzed based on the original spectrum and the first derivative spectral characteristics of the flowers.The sensitive wavebands were determined,the characteristic spectral parameters were constructed,and the monitoring models for the phosphorus content of apple flowers were established accordingly.Correlation analysis showed that the phosphorus content of apple flowers was significantly positively correlated with the original spectral reflectance at 350-370 nm,670-1385 nm,and 1620-1760 nm,and the first derivative reflectance at 500-520 nm,and significantly negatively correlated with the first derivative reflectance at 670-730 nm.The spectral parameters DVI(936,676),DVI(977,676),NDVI(936,676)and NDVI(977,676)had close correlations with the phosphorus content of apple flowers,the correlation coefficients being all ＞ 0.77.By using these spectral parameters as independent variables,four monitoring models for the phosphorus content of apple flowers were established,among which,the model with NDVI(936,676)as the variable was validated as the best monitoring model,its determination coefficient was the highest(R~2 = 0.9385),root mean square error and relative error were the smallest(RMSE = 0.6883,and RE = 7.6%),and prediction accuracy reached92.4%.