论文部分内容阅读
为研究土壤水分的高光谱特征,利用美国ASD公司的FieldSpec 4 HR型地物光谱仪(光谱范围为350~2500nm)测定获取不同含水量土壤样品的高光谱数据,并通过实验分析其水分含量。分别以土壤反射率(R)、反射率的对倒数(Log(1/R))及包络线去除的反射率(Rcr)三个光谱特征进行土壤水分定量反演研究,通过分析各高光谱特征与土壤水分的相关性,遴选特征波段;利用偏最小二乘回归方法(PLSR)建立土壤水分的高光谱定量反演模型,最后利用检验样品进行精度检验和比较。研究结果表明,利用经包络线去除后光谱反射率进行定量反演的预测精度最高,预测值与实测值相关性良好(r~2=0.8983**),趋势线接近于y=x,这为土壤水分的大范围快速监测提供了有效手段。
In order to study the hyperspectral characteristics of soil moisture, the hyperspectral data of soil samples with different water contents were determined by the FieldSpec 4 HR type spectrometer (spectral range 350 ~ 2500 nm) from ASD Company of America and their moisture content was analyzed experimentally. The quantitative analysis of soil moisture was carried out using three spectral characteristics of soil reflectance (R), log reciprocal of reflectance (Log (1 / R)) and reflectance (Rcr) Characteristics and soil moisture, the selection of characteristic bands; using partial least-squares regression (PLSR) to establish soil moisture hyperspectral quantitative inversion model, and finally the use of test samples for accuracy testing and comparison. The results show that the prediction accuracy of quantitative inversion using the spectral reflectance after removing the envelope is the highest, the correlation between the predicted value and the measured value is good (r ~ 2 = 0.8983 **), and the trend line is close to y = x It provides an effective method for rapid monitoring of soil moisture on a large scale.