基于漫反射光谱的茶园土壤硝态氮检测

该文研究了充分利用土壤漫反射光谱在可见-近红外波段的有效信息，研究快速准确检测土壤硝态氮含量的新方法。试验选取89个风干土壤样本，经粉碎过直径1 mm筛孔后，使用 FieldSpec 3便携式光谱仪（光谱波长范围：400～2 500 nm），获取其漫反射光谱。检查各土样的原始光谱的有效性并进行平均，经偏最小二乘法partial least squares（PLS）聚类分析后，选取其中的63个样本构成校正集建立模型，10个样本构成预测集进行模型验证。通过一阶微分与滑动平均滤波相结合的预处理方法，用15个主成分建立的主成分+神经网络模型为最好，其校正模型的回判相关系数为0.9908，均方根误差（RMSEC）为1.4528，预测模型的相关系数为0.7179。研究结果表明，利用可见-近红外光谱技术可以准确地检测茶园土壤硝态氮含量。

Soil nitrate nitrogen sensing for tea garden based on diffuse reflectance spectroscopy

Authors researched the method of soil nitrate nitrogen content determination with diffuse reflectance spectroscopy for its rapidness and accuracy. The portable spectroradiometer, FieldSpec 3 with a full spectral wavelength of 400-2500 nm, was used to scan diffuse reflectance spectra of soil samples. Eighty-nine soil samples were selected according to different soil fertility, depth and sites, which covered wide range of nitrate nitrogen content. Soil samples were air-dried and sieved through 1-mm screen holes after grinding. Data validity of original spectra was checked and averaged. Sixty-three samples were used to establish the calibration model with the methods of standing-wave ratio after the cluster analysis by partial least squares. Ten samples were used to establish the prediction set and the calibration model was validated. After being preprocessed by the combination of first-order derivative and moving average filter, the calibration model with fifteen principal component factors was regarded as the best with the algorithm of principal constituent analysis and artificial neural networks, and the correlation coefficient of the calibration model was 0.9908. The root mean square error of calibration was 1.4528. The correlation coefficient between predicted values and real values was 0.7179. The results show that soil nitrate nitrogen content can be determined precisely with visible-near infrared spectra.