基于Vis-NIR光谱的果园土壤有机质便携式检测仪研究

Research on portable detecting instrument of soil organic matter based on Vis-NIR spectra

以赣南脐橙果园土壤为研究对象,对采集到的56个土样风干、过筛,然后进行化学分析,同时使用傅里叶近红外和自主设计的便携式仪器光谱采集。采集到的光谱数据经过一阶微分、平滑、多元散射校正、归一化四种常用的预处理方法,分别应用偏最小二乘法（partial least square regress PLS）进行建立脐橙果园土壤有机质的模型。研究表明,两仪器需要不同的预处理,傅里叶仪器原始光谱建模效果较好,而便携式仪器则需要平滑预处理。傅里叶仪器建立模型的预测相关系数（RP）为0.893,预测均方根误差（RMSEP）为0.474;自行设计的仪器在1 000~1 700nm范围建立赣南脐橙果园土壤有机质含量模型的预测模型的相关系数（RP）为0.801,预测均方根误差（RMSEP）为0.779。研究表明自行设计的便携式仪器可快速用于赣南脐橙果园的土壤中有机质含量的检测。

To study the distribution of soil nutrients and build soil models of soil organic matter（SOM）that could predict the measured value,the soil samples coming from Gannan navel orange orchard were collected.The soil samples were air-dried and sieved through 0.149 mm screen holes after grinding The portable spectroradiometer of BRUKER TENSOR 37 with a full spectral wavelength of 400—2 500 nm and the portable instrument with the wavelength of 900~1700nm were used to scan the diffuse reflectance spectroscopy of soil samples,and the data validity of original spectra was averaged.Fifty-six soil samples were selected,and thirty-eight soil samples were used to build the calibration model and eighteen were used to build the prediction model.The collected spectral data was retreated by first order derivation,smoothing,multiplicative scatter correction and normalization,and the navel orange orchard soil organic matter model was established using partial least square regress（PLS）.The portable spectroradiometer of BRUKER TENSOR 37 gives the best results with the correlation coefficient（RP）of 0.893,and the root mean square error of prediction（RMSEP）of 0.474.With portable instrument,the correlation coefficient（RP）was 0.801,and the root mean square error of prediction（RMSEP）was 0.779.The portable instrument can quickly detect the organic matter content in the soil of Gannan navel orange orchard.