基于无人机高光谱遥感的草原植物冠层滞尘量估算  

Estimation of dust retention in grassland plant canopy based on UAV hyperspectral remote sensing

作　　者：闫石 赵义博 雷少刚[2] 杨星晨 李恒[2] 宫传刚 YAN Shi;ZHAO Yibo;LEI Shaogang;YANG Xingchen;LI Heng;GONG Chuangang(Shenhua BeiDIan Shengli Energy Co.,Ltd.,Xilin Hot 026015,Inner Mongolia,China;Engineering Research Center of Ministry of Educa-tion for Mine Ecological Restoration,China University of Mining and Technology,Xuzhou 221116,Jiangsu,China)

机构地区：[1]神华北电胜利能源有限公司,内蒙古锡林浩特026015 [2]中国矿业大学矿山生态修复教育部工程研究中心,江苏徐州221116

出　　处：《河南理工大学学报（自然科学版）》2022年第6期84-92,共9页Journal of Henan Polytechnic University（Natural Science）

基　　金：国家重点研发计划项目(2016YFC0501107)。

摘　　要：为了探索无人机高光谱遥感估算草原植物冠层滞尘量的可行性,以内蒙古自治区宝日希勒草甸草原为研究区,利用1/10000天平和叶面积仪测定植物冠层滞尘量,通过成像光谱仪采集机载高光谱影像,分析滞尘对冠层光谱反射率和三边参数的影响,探究滞尘量与单波段反射率和光谱指数的相关关系,建立基于光谱参数的植物冠层滞尘量估算模型。结果表明:(1)随着滞尘量增加,可见光波段反射率先升高后降低,但整体呈降低趋势,近红外波段的反射率逐渐降低;冠层滞尘对三边位置无显著影响,与黄边幅值(D_(y))、红边幅值(D_(r))、黄边面积(SD_(y))和红边面积(SD_(r))呈显著负相关;(2)冠层滞尘量与原始光谱反射率相关系数均为负值,比值指数(RI)、差值指数(DI)和归一化指数(NDI)均会不同程度影响滞尘量与光谱数据的相关性;(3)以R_(542),R_(969)(542,969 nm光谱反射率),DI_(720,969),NDI_(725,890),D_(y)和D_(r)为变量,使用随机森林(RF)构建的植物冠层滞尘量估算模型准确度较高。研究结果可为草原植物冠层滞尘量监测提供理论依据和技术支持。In order to explore the feasibility of UAV hyperspectral remote sensing to estimate the dust retention content of grassland plant canopy,Baorixile grassland in the Inner Mongolia Autonomous Region was taken as the research area.The dust content was measured with an electronic analytic balance(1/10000g scale)and a leaf area meter.The airborne hyperspectral images were obtained by using an imaging spectrometer.The influence of dust retention on spectral reflectance and trilateral parameters of the canopy was analyzed.The correlation of dust retention content with single-band reflectance and spectral index was explored,and the estimation model of plant canopy dust retention was established based on spectral parameters.The results showed that:(1)With the increase of dust retention content,the reflectance of visible wavelength increased first and then decreased,but the overall change tended to decrease;the reflectance of near-infrared wavelength decreased gradually.The canopy dust had no significant influence on the trilateral position.Still,it was significantly correlated with the yellow edge slope(D_(y)),red edge slope(D_(r)),yellow edge area(SD_(y)),and red edge area(SD_(r));(2)The correlation coefficients between the canopy dust retention content and the original spectral reflectance were negative.The ratio index(RI),difference index(DI),and normalized difference index(NDI)all enhanced the correlation between the dust retention content and spectral data in varying degrees;(3)The estimation model of plant canopy dust retention constructed through the random forest(RF)had high accuracy with the spectral reflectance of 542 and 969 nm(R_(542) and R_(969)),DI_(720,969),NDI_(725,890),D_(y),and D_(r) as variables.The distribution map of canopy dust retention based on airborne hyperspectral images can better reflect the spatial distribution.The results could provide a theoretical basis and technical support for monitoring the dust retention content of grassland plants.

关 键 词：草原植物 冠层滞尘 高光谱 光谱分析 估算模型 

分 类 号：X826[环境科学与工程—环境工程]