黑土典型区有机质高光谱预测模型

土壤有机质(SOM)是鉴别土壤肥力的重要指标,是土壤肥力的物质基础,其含量预测模型研究对于土壤肥力评价、土壤碳库估算、土壤资源利用与保护具有重要意义。该文以黑龙江省黑土带典型区为例,采集区域土壤样本,基于有机质含量与土壤反射率的定量关系,对光谱反射率进行一阶微分和倒对数的处理,建立偏最小二乘法模型(PLSR)、一元线性回归模型和多元线性逐步回归模型。结果表明:(1)土壤有机质敏感波段位于650-750nm。(2)通过比较建模样本与检验样本的决定系数(R2)和均方根误差(RMSE)的大小,得到反射率和倒对数处理后的数据最优模型都为PLSR模型,一阶微分处理后的最优模型为多元线性逐步回归模型。(3)PLSR模型的建模效果优于回归模型,但其预测效果却并不理想。该研究将为改进土壤理化参数、遥感反演、土地质量评价等工作方法提供理论与技术支持。

Hyper spectral prediction model of organic matter in black soil region

The study on spatial heterogeneity of soil organic matter (SOM) is significantly important to soil fertility evaluation, soil carbon pool estimation, soil resources utilization and protection. Based on Chernomyrdin typical area in Heilongjiang province as an example, the soil samples in the area of acquisition, based on the quantitative relationship between organic matter content and soil reflectance, for first order differential spectral reflectance, logarithmic and remove abnormal numerical processing, establish a model of partial least squares (PLSR), a yuan linear regression model and multiple linear regression model. Results show that: (1) the sensitive wavelengths of organic matter in 650-750 nm. (2) by comparing the determination coefficient (R2) and the size of the root mean square error (RMSE), after processing the data of the optimal models for the PLSR model. The results can provide theoretical and technical support for improving RS retrieving of soil physic -chemical parameters, evaluating soil quality and carbon pool.