摘要
高光谱遥感技术因其便捷快速、节约成本、非破坏性和准确度高的特点,被广泛应用于资源环境、军事、大气遥感和测绘等领域。本研究在化学实验(碳库分离)的基础上,结合高光谱遥感技术,利用统计分析工具SAS分析家模块将三库分离数据与光谱反射率数据进行多元逐步回归分析,筛选出敏感波段,从而建立起了一种森林土壤惰性碳和缓效性碳的快速估算模型。研究发现,无截距模型的显著性较为明显,解释能力和预测能力较好。利用光谱求平均和平滑去噪方法处理过的光谱反射率数据通过逐步回归筛选出来的敏感波段组合对森林土壤惰性碳和缓效性碳含量有较强的预测能力。可以利用高光谱技术来进行森林土壤惰性碳含量和缓效性碳含量的快速估算与监测。
The hyperspectral remote sensing technique has been widely applied to resource and environment, military affairs, atmospheric remote sensing, surveying, mapping and other fields due to its characteristics of fast detecting, low cost, non-destructive and high accuracy. Based on the three library separation data derived from laboratory analysis experiment and spectral reflectance data collected by hyperspectral remote sensing device, we made a multiple stepwise regression analysis using SAS analyst module of statistical analysis tools to find the sensitive bands, finally constructed a model for rapid estimation of inert and slow carbon content in forest soils. The results indicated that no intercept model was proved to be more significant, explanatory and predictive. The sensitive band combinations, picked out by stepwise regression method based on the spectral reflectance data processed by the methods of averaging,denoising and smoothing, was more predictive. Therefore, the hyperspectral remote sensing technique could be used to rapid estimation of the content of inert and slow carbon in forest soils.
出处
《土壤通报》
CAS
CSCD
北大核心
2015年第3期745-753,共9页
Chinese Journal of Soil Science
基金
国家重点基础研究发展计划(973)项目(2010CB9507002)
江苏高校优势学科建设工程资助项目(PAPD)
中国科学院战略性先导科技专项课题(XDA0505050703)资助
