引用本文:林 卡,李德成,刘 峰,张甘霖.基于可见-近红外反射光谱的土壤碳酸钙含量与反演效果关系研究[J].土壤学报,2018,55(2):304-312.
LIN Ka,LI Decheng,LIU Feng,ZHANG Ganlin.Study on Relationship between Soil Calcium Carbonate Content and Inversion Effect Based on Visible Near-infrared Reflectance Spectra[J].Acta Pedologica Sinica,2018,55(2):304-312
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 558次   下载 944 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于可见-近红外反射光谱的土壤碳酸钙含量与反演效果关系研究
林 卡, 李德成, 刘 峰, 张甘霖
土壤与农业可持续发展国家重点实验室(中国科学院南京土壤研究所)
摘要:
土壤可见-近红外反射光谱中包含了大量的土壤属性信息,研究人员根据土壤属性信息在光谱上的特征,对土壤属性进行定量反演。是否属性值越高,反演精度越高?目前对于属性含量与反演效果的定量关系尚不清楚。采集了我国西北地区黑河流域69个代表性干旱土剖面(292个发生层土样),以气量法测定其碳酸钙含量,使用Cary 5000分光光度计测定其可见-近红外光谱反射率,以样本量和离散度(变异系数)作为数据集划分标准,分别建立了11个相同样本量子集(A)和5个相近离散度子集(B),应用偏最小二乘回归(PLSR)算法对各子集进行土壤碳酸钙含量反演,以此探究碳酸钙含量与反演效果的定量关系。结果表明,碳酸钙可增加可见-近红外波段的光谱反射率,但利用可见近红外光谱反演土壤碳酸钙含量,其反演效果与碳酸钙含量关系不显著。
关键词:  可见-近红外反射光谱  碳酸钙含量  反演效果
DOI:10.11766/trxb201707170240
分类号:
基金项目:国家自然科学基金项目(41371224,41130530)
Study on Relationship between Soil Calcium Carbonate Content and Inversion Effect Based on Visible Near-infrared Reflectance Spectra
LIN Ka, LI Decheng, LIU Feng, ZHANG Ganlin
State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences
Abstract:
【Objective】 Soil visible near-infrared reflectance spectra contains large volumes of information on soil physical and chemical properties, which implies that it is feasible to use soil spectra to invert soil properties quantitatively. Is it the higher the property value, the higher the inversion accuracy? However, at present, it is still unclear how to relate quantitatively effects of inversions to soil property contents. 【Method】 Therefore, this study selected soil calcium carbonate content as the target attribute for exploration of quantitative relationship between spectral inversion effect and calcium carbonate content. A total of 292 soil samples were collected out of the genetic horizons of 69 typical Aridosols profiles in the Heihe River Basin, Northwest China, for analysis of calcium carbonate contents with the gasometric method and acquisition of visible near-infrared reflectance spectra with a Cary5000 spectrophotometer. Based on the characteristics of the distribution of calcium carbonate content in the typical study area, 11 identical sample size subsets (A) and 5 similar dispersion subsets (B) were established with sample size and data dispersion (coefficient of variation) as the criteria for dataset partitioning, and the partial least-squares regression (PLSR) method was used to invert calcium carbonate content from the spectral curves.【Result】 Results show that calcium carbonate in the Aridosols of the Heihe River Basin varied in the range of 4.86 g kg-1 ~ 236.03 g kg-1 in content with an average of 103.07 g kg-1. Soil samples with calcium carbonate content varying in the range of 30 ~ 60 g kg-1 and of 120 ~ 150 g kg-1, were in dominancy, accounting for 21.4% and 32.6% of the total, respectively. As a whole, the soil is high in calcium carbonate content, which is consistent with the characteristics of Aridosols being rich in calcium carbonate. With the PLSR, modeling was performed for prediction of calcium carbonate contents of the soil samples in the 11 A subsets. RPD of the validation set of each subset ranged between 0.92 and 1.04, fluctuating around 1 with no obvious features of variation, which indicates that calcium carbonate content does not have much impact on prediction or inversion of soil calcium carbonate content, using visible near-infrared reflectance spectra. Modeling was also done for prediction of calcium carbonate content in 5 B subgroups, with a similar result. 【Conclusion】 Therefore, soil calcium carbonate content is not the main factor affecting the prediction using spectra, which is inconsistent with the qualitative knowledge the researchers already have in mind. Calcium carbonate can enhance spectral reflectance of visible near-infrared bands, but the effect is not so significantly reflected in using the visible near-infrared spectral reflectance to inverse soil calcium carbonate content. Therefore, it seems unnecessary to divide calcium carbonate samples by content of soil calcium carbonate when using spectra to predict calcium carbonate contents. Whether the conclusion is applicable to other soil properties needs to be further verified, and how to improve accuracy of the prediction of target attribute will be the focal point of the next phase of the study.
Key words:  Visible near-infrared reflection spectra  Calcium carbonate content  Inversion effect