含水率对土壤有机质含量高光谱估算的影响

土壤含水率对有机质(soil organic matter,SOM)含量高光谱估算精度有很大的影响。为了探讨SOM高光谱估算中土壤含水率的影响,该文对烘干土、风干土和质量含水率为5%~40%(按5%递增)的土壤样本进行了室内高光谱测量,对光谱数据进行了反射率、反射率一阶导数和反射率倒数对数3种光谱数据变换,运用偏最小二乘回归法(partial least squares regression,PLSR)建立了相应的SOM估算模型。结果表明,风干土的SOM高光谱估算精度较好;当含水率水平小于25%时,SOM估算模型精度受含水率的影响较大,光谱数据进行反射率倒数对数变换后的模型精度最高;当含水率水平大于等于25%时,水分对土壤光谱反射率的影响要大于SOM,不适宜利用土壤光谱数据进行SOM含量高光谱估算。该研究可为大田环境不同含水率情况下光谱估算SOM提供参考。

Hyperspectral prediction of soil organic matter contents under different soil moisture contents

Abstract: Soil moisture content has great influence on the prediction accuracy of soil organic matter (SOM) content using hyperspectral data. The purpose of this study was to find the threshold of soil moisture content suitable for using hyperspectral data to predict SOM content. A total of 63 soil samples including black soil, chernozem and meadow soil were collected from crop fields in Lishu and Gongzhuling county, Jilin province and in Binxin county, Heilongjiang province. The soil samples were air-dried and sieved through a 2-mm sieve. SOM contents were measured in the laboratory. The soil samples were divided into two groups including 42 samples for calibration and 21 for validation. Reflectance of soil samples with over-dried, air-dried and 5% to 40% soil moisture contents (the interval of 5%) were measured using ASD Fieldspec Pro High Spectrometer in a dark room. Soil spectral reflectance (R) was mathematically transformed into first derivatives of reflectance (R') and the logarithm of the inverse of the reflectance (Log (1/R)). SOM content spectral prediction models were set up respectively by using partial least squares regression (PLSR) method. The method of variable importance in projection (VIP) was used to analyze which spectral ranges were important to explain SOM content under different soil moisture contents by using PLSR. The results showed that soil spectral reflectance had a larger decline with soil moisture content increasing from 5% to 25%, but the decline trend slowed down when soil moisture content increased from 25% to 40%. That means the soil moisture content with less than 25% had more obvious effect on soil spectral reflectance change than soil moisture content with higher than 25%. With the increase of soil moisture content, moisture absorption valley appeared a large tendency on bands of 1 450 and 1 900 nm. It indicated that effects of soil moisture content on soil spectral reflectance happened mainly in the near infrared wavelength range. SOM content spectral prediction model for air-dried soil samples had better accuracy. When the soil moisture content was less than 25%, the accuracy of SOM content estimation model was affected by soil moisture content largely, and the highest prediction accuracy was Log (1/R) spectral data transformation model. When the soil moisture content was or more than 25%, it was not suitable to be used for hyperspectral SOM content estimation, because SOM spectral characteristics was covered by soil moisture spectral characteristics. The VIP values of reflectance bands from 1 870 to 2 400 nm with higher than 25% soil moisture contents were less than 1. That means those wavelength had weak explanation ability of SOM content. This study can provide valuble information for SOM content spectral estimation in the crop field that has different soil moisture conditions.