论文部分内容阅读
为提高近红外传感器测量的准确度,进一步理解不同标定模型对土壤含水率测量精度的影响。利用土壤表面的近红外反射光强来预测土壤含水率,通过归一化处理将反射光强转化为相对吸收深度和相对反射率,采用2种标定方法,分别建立土壤含水率与相对吸收深度之间及土壤含水率与相对反射率之间的线性模型与非线性模型。选取我国东北地区的黑土进行标定,并用独立的试验数据对模型进行检验。结果表明,吸收深度法的线性和非线性模型的预测值和实测值符合度较好。反射率法的线性模型和非线性模型对土壤的含水率预测均方根误差(RMSE)分别为2.89%和2.95%,相对吸收深度法非线性模型的RMSE值明显大于其他3种模型,预测准确度最低。说明不同标定方法会影响土壤含水率的预测结果。4种模型的预测精度能够满足测量要求。
In order to improve the accuracy of NIR sensor measurement, we further understand the influence of different calibration models on the measurement accuracy of soil moisture content. The soil water content is predicted by the intensity of near-infrared reflection of the soil surface, and the light intensity is transformed into the relative absorption depth and the relative reflectivity by normalization. Two methods are used to establish the soil moisture content and the relative absorption depth Linear model and nonlinear model between soil water content and relative reflectance. Select the black soil in northeast China for calibration, and test the model with independent test data. The results show that the predictions and the measured values of the linear and nonlinear model of the absorption depth method are in good agreement. The root mean square error of prediction (RMSE) was 2.89% and 2.95% respectively for the linear model and the non-linear model of the reflectance method, and the RMSE value of the non-linear model of the relative absorption depth method was significantly greater than that of the other three models The lowest degree. Describe the different calibration methods will affect the prediction of soil moisture content. The accuracy of the four models can meet the measurement requirements.