基于PROSAIL模型的高寒冬季枯草关键参数阈值率定  

Threshold Calibration of Key Parameters of Withered Grass Based on PROSAIL Model in Qinghai-Tibet Plateau

作　　者：梁好 徐维新 段旭辉 张娟[2] 代娜 肖强智 王淇玉 LIANG Hao;XU Wei-xin;DUAN Xu-hui;ZHANG Juan;DAI Na;XIAO Qiang-zhi;WANG Qi-yu(School of Resources and Environment,Chengdu University of Information Technology,Chengdu 610225,China;Qinghai Institute of Meteorological Science,Qinghai Meteorological Bureau,Xining 810001,China)

机构地区：[1]成都信息工程大学资源环境学院,四川成都610225 [2]青海省气象科学研究所,青海西宁810001

出　　处：《光谱学与光谱分析》2022年第4期1144-1149,共6页Spectroscopy and Spectral Analysis

基　　金：青海省科技计划项目(2019-HZ-820);国家自然科学基金项目(41861049,41971328);四川省科技计划重点研发项目(2019YFS0465)资助。

摘　　要：冬季牧草即枯草的存量是生态补偿计算与畜牧生产科学管理的关键基础,而对青藏高原枯草关键参数认识的不足,直接限制了高寒冬季枯草监测研究与应用发展。PROSAIL是一种光学辐射传输模型,它可以定量描述植被参数与冠层反射率的关系。利用最新版本PROSAIL模型,结合野外实测的枯草光谱及叶面积、叶绿素等10个性状参数数据,模拟生成了15 000组潜在的枯草光谱数据序列。通过冬、夏实测枯草与绿草样方的反射光谱特征分析,揭示了枯草在可见光波段与近红外波段与绿草的显著差异性,描述了青藏高原冬季枯草在400~1 300 nm波段近似线性的独特光谱分布特征。在此基础上,提出了以红光与绿光波段差值为依据的鲜/枯草光谱区分方法,并据此实现了15 000组模拟光谱中枯草光谱的初级与二级筛选,建立了枯草模拟光谱数据序列集。该模拟光谱数据序列集与实测光谱在400~2 500 nm全波段明显相关,所有模拟谱线R^(2)均在0.904~0.994之间,表明该模型能够很好地模拟高寒冬季枯草的反射率光谱。进一步采用EFAST方法,对枯草模拟光谱数据序列进行全局敏感性分析,识别出棕色素、类胡萝卜素、花青素、叶片结构、热点5个对枯草光谱变化不敏感的参数,并在此基础上优化枯草敏感参数阈值区间。最终,以99%置信区间为标准、余弦距离为评价函数,OFAT方式再次运行模型,界定了枯草敏感的参数阈值:叶面积指数阈值区间为0.2~0.89、叶绿素含量为0~1.29μg·cm^(-2)、平均叶倾角为11°~90°、等效水厚度为0.000 1~0.005 cm、干物质含量为0.008~0.05 g·cm^(-2)。通过对10个枯草性状参数及其取值区间的率定,提出了枯草光谱关键参数数值区间参考表,为提高对高寒冬季枯草性状特征的科学认识及探究遥感反演应用技术方法提供理论依据与基础数据。Grassland,as an important part of the ecosystem in the Qinghai-Tibet Plateau,plays an ecological indicator role.However,during the non-growing season,it generally didn’t monitor or observe alpine grass in winter.It could be a great gap to develop the methods of grassland monitoring and its application in winter.PROSAIL,a physical radiation model,can quantitatively describe the relationship between various vegetation parameters and canopy reflectance spectra.In this study,the latest version of the PROSAIL-D model and ground observed data were applied to explore the thresholds of critical range for 10 parameters of withered grass affected by reflectance spectrum.Based on the reflectance spectrums and the corresponding character’s parameters of withered grass that were obtained in the field,15 000 possible withered grass spectrums were simulated by the PROSAIL model.Compared to the difference of reflectance spectra between withered grass and green grass observed in winter and summer respectively,it is found that a clear difference displayed on visible and near-infrared bands and with a significant linear in 400~1 300 nm spectral range for withered grass in winter in Qinghai-Tibet Plateau.On that basis,we proposed a method to distinguish the withered and green grass using the difference between red and green spectral reflections.It can be considered as a withered grass spectrum while the difference is greater than 0.Furthermore,a dataset of potential withered grass spectrum was established by two-steps identification from 15 000 possible spectrums based on the methods described above.The potential withered grass spectrums are correlated closely to the observed spectrums with a whole range of 400~2 500 nm,and the R^(2) of all the simulated spectrum lines was between 0.904 and 0.994.By EFAST method and global sensitivity analysis,the brown pigment,carotenoid,anthocyanin,leaf structure and hot spot were identified as non-sensitive factors that respond to the withered grass spectrum.Finally,PROSAIL model was run agai

关 键 词：枯草 PROSAIL模型 参数阈值 敏感性分析 青藏高原 

分 类 号：O657.3[理学—分析化学]