论文部分内容阅读
通过系统分析我国中、东部地区5种不同类型土壤风干样本的有机质含量与近红外(1000～2500nm)光谱反射率和一阶导数两波段构成的比值、差值及归一化指数之间的关系,构建了适合土壤有机质含量估测的光谱参数及定量反演模型.结果表明:用多元散射校正及Savitzky-Golay平滑方法对原始光谱反射率进行预处理后,其两波段组成的光谱参数与土壤有机质含量的相关性明显优于原始光谱反射率组成的光谱参数,而由上述预处理后的反射率一阶导数的两波段构成的光谱参数介于二者之间;不同类型光谱参数构成形式中,以差值指数的预测性最好,其次为比值和归一化指数;与土壤有机质含量相关程度最高的光谱参数,是以近红外合频区1883和2065nm2个波段的反射率经多元散射校正和Savitzky-Golay平滑后构建而成的差值指数DI(CR1883,CR2065),两者呈极显著的直线相关.经不同类型土壤的观测资料检验,模型的决定系数为0.837,均方根误差为4.06;与偏最小二乘法的全谱建模结果相比,尽管DI(CR1883,CR2065)的预测精度略低于后者,但该指数只使用了2个波段的反射率,且所建模型比较简单,能为便携式监测仪的研制提供更有效的信息,可作为一种良好的土壤有机质估测光谱参数.
Through the systematic analysis of the relationship between the organic matter content of the air-dried samples of five different types of soils in the eastern and central regions and the spectral reflectance and the first-order two-band spectral composition of the near infrared (1000-2500nm), the difference and the normalized index , The spectral parameters and the quantitative inversion models suitable for the estimation of soil organic matter content were constructed.The results showed that the spectral parameters of the two bands after the pretreatment of the original spectral reflectance by the multivariate scatter correction and the Savitzky- The correlation of organic matter content is obviously better than that of the original spectral reflectance, while the spectral parameters of the two bands of the first-order reflectance pre-treatment reflect the spectral parameters between the two; , With the best predictor of difference index, followed by the ratio and normalized index; the spectral parameters with the highest correlation with soil organic matter content were corrected by multivariate scatter in the two bands of 1883 and 2065nm After the Savitzky-Golay smoothing, the difference index DI (CR1883, CR2065) is constructed, and the two are extremely significant linear correlations.The different types of soil The test results show that the coefficient of determination of the model is 0.837 and the root mean square error is 4.06. Although the prediction accuracy of DI (CR1883, CR2065) is slightly lower than that of partial least squares method, The index only uses the reflectivity of two bands, and the model is relatively simple. It can provide more effective information for the development of portable monitor and can be used as a good spectral parameter of soil organic matter estimation.