土壤氮素室内外高光谱响应机理与预测模型研究

Soil nitrogen-supplying capacities are determined by the content of total nitrogen (TN) and available nitrogen, which are important basis of precise fertilization. Spectral technique provides an non-pollution, cost and time effective, and less laborious way for fast getting soil nitrogen information. Moisture and particle size are the main limitations for the prediction accuracy of nitrogen in lab and field. This project focuses on removing the effect of soil moisture and particle size from reflectance spectra for nitrogen predicting. Firstly, to explore the spectral mechanism of total nitrogen and available nitrogen and to improve the nitrogen spectral prediction accuracy, the inspective and combinative effect of soil moisture and size are analyzed by orthogonal experiment in lab standard condition. Then further compare soil testing methods to obtain optimal measuring technique and related parameters, and explore the optimal algorithm of reducing or eliminating the effect of moisture and size under the field condition. Based on knowing the common effect factors of different soil spectra, the soil types with spatial difference are classified, and the nitrogen spectral prediction models are established according to classifications. The resluts will Provide an important basis for measuring directly the fresh soil samples in lab and detecting the real-time nitrogen information in situ, and nitrogen spectral prediction for different soil types.

土壤中全氮（TN）和速效氮含量决定着土壤的供氮能力，是精准施肥的重要依据。高光谱技术为快速获取土壤氮素信息提供了一条省时、省力、无污染并且测试成本低的有效途径。土壤水分和粒径的影响一直是制约着实验室和野外土壤养分光谱快速、精准预测的关键因素。本项目以降低或消除土壤水分和粒径对氮素光谱的干扰为研究重点。首先在室内标准测定条件下通过正交实验来分析水分、粒径和氮素独立和综合效应光谱响应机理，探索不同形态氮素光谱探测精度提高的方法；再进一步通过比较田间原状土壤不同测试方法，明确野外条件下土壤光谱测试技术和相关参数，探索降低或消除水分和粒径干扰、提高氮素光谱预测精度的最优算法。在解决土壤光谱共性影响因子基础上，对具有空间差异的土壤样本进行聚类，建立不同分类土壤氮素含量的光谱预测模型。本项目为实验室新鲜土样的直接测定、田间原位土壤氮素实时获取以及不同土壤类型氮素光谱预测提供了重要依据。

氮素作为矿质营养之首与作物产量和品质关系最为密切，高光谱技术可以实现快速、准确估测土壤氮素含量水平，推动土壤信息化管理进程。本课题围绕不同土壤类型高光谱监测的共性影响因子及其光谱的差异性进行了探索，取得的主要结果如下：.1）在明确土壤样品室内外高光谱测试技术规范的基础上，首先研究了可控条件人为处理的样品和新鲜原土样光谱响应，解析土壤含水量、粒径对土壤氮素光谱的响应规律和贡献。通过人为控制获得不同粒级和不同含水量的土壤样品，采集不同土样的光谱特征并进行比较，按粒径等级利用最小二乘法（PLSR）建立农田土壤含水量的光谱定量预测模型。.2）针对不同类型、不同来源的土壤，比较了不同土壤类型光谱反射率曲线特征变化，并进行了基于光谱特征的土壤类型分类，该结果对于土壤调查及分类提供了重要的参考价值。建立了土壤氮素、有机质光谱预测模型，模型在有机质较丰富的东北黑土上预测达到了较高的精度。通过比较不同土壤类型有机质光谱预测模型及其同质性分析，确定了土壤理化性质相似的可采用共同模型，但预测精度下降。