Leaf pigments are critical indicators of plant photosynthesis,stress,and physiological conditions.Inversion of radiative transfer models(RTMs)is a promising method for robustly retrieving leaf biochem-ical traits from...Leaf pigments are critical indicators of plant photosynthesis,stress,and physiological conditions.Inversion of radiative transfer models(RTMs)is a promising method for robustly retrieving leaf biochem-ical traits from canopy observations,and adding prior information has been effective in alleviating the“ill-posed”problem,a major challenge in model inversion.Canopy structure parameters,such as leaf area index(LAI)and average leaf inclination angle(ALA),can serve as prior information for leaf pigment retrie-val.Using canopy spectra simulated from the PROSAIL model,we estimated the effects of uncertainty in LAI and ALA used as prior information for lookup table-based inversions of leaf chlorophyll(C _(ab))and car-otenoid(C_(ar)).The retrieval accuracies of the two pigments were increased by use of the priors of LAI(RMSE of C_(ab) from 7.67 to 6.32μg cm^(-2),C_(ar) from 2.41 to 2.28μg cm^(-2))and ALA(RMSE of C_(ab) from 7.67 to 5.72μg cm^(-2),C_(ar) from 2.41 to 2.23μg cm^(-2)).However,this improvement deteriorated with an increase of additive and multiplicative uncertainties,and when 40% and 20% noise was added to LAI and ALA respectively,these priors ceased to increase retrieval accuracy.Validation using an experimental winter wheat dataset also showed that compared with C_(ar),the estimation accuracy of C_(ab) increased more or deteriorated less with uncertainty in prior canopy structure.This study demonstrates possible limita-tions of using prior information in RTM inversions for retrieval of leaf biochemistry,when large uncer-tainties are present.展开更多
叶面积指数(leaf area index,LAI)作为衡量作物生长状况的关键参数,对其进行精准高效的反演对于作物监测、产量预测等活动至关重要。然而,传统经验模型在估算LAI时常存在计算负荷重、泛化能力弱等问题。为实现青贮玉米多时序LAI精准、...叶面积指数(leaf area index,LAI)作为衡量作物生长状况的关键参数,对其进行精准高效的反演对于作物监测、产量预测等活动至关重要。然而,传统经验模型在估算LAI时常存在计算负荷重、泛化能力弱等问题。为实现青贮玉米多时序LAI精准、高效估算,该研究以甘肃省民乐县的大田青贮玉米LAI为研究对象,结合Landsat-8多光谱影像与同期实地采集的LAI数据,提出了4种基于EFAST全局敏感性分析方法的机器学习混合反演模型(MLP-PROSAIL、SVR-PROSAIL、RF-PROSAIL和GBM-PROSAIL)。通过对PROSAIL模型的输入参数进行敏感性分析,以便确定参数敏感度并准确模拟输出冠层反射率光谱。进一步对Landsat-8多光谱数据进行预处理和波段变换,并采用地理配准工具结合反距离加权插值的策略减少其尺度差异。同时利用贝叶斯超参数寻优和正则化技术优化模型不同的参数类型和激活函数,得到4种改进模型用于训练LAI与光谱数据,通过5折交叉验证法和留一验证法对4种模型的反演性能进行验证并选出最优模型。优化后的模型性能均有明显提升,其中,GBM-PROSAIL模型反演性能最优,拟合精度R^(2)为0.93、均方根误差(RMSE)为0.42。MLP-PROSAIL、SVR-PROSAIL和RF-PROSAIL模型的拟合精度R^(2)依次为0.85、0.88、0.90,RMSE依次为0.80、0.69、0.51。根据GBM-PROSAIL模型绘制的研究区多时序LAI反演空间分布结果表明:不同生长期青贮玉米LAI值存在明显差异,能较好反映其生长过程。该研究提出的混合反演模型具有较高的性能及较强的鲁棒性,可为多时序、大尺度作物监测、产量预测相关研究提供方法与思路。展开更多
叶面积指数(leaf area index,LAI)是反映植物冠层结构和光能利用的重要指标.随着遥感技术的不断发展,利用遥感数据获取大面积LAI已经成为监测作物生长和估产的重要手段.基于物理模型的LAI遥感反演方法经常假设作物冠层结构是均匀分布,然...叶面积指数(leaf area index,LAI)是反映植物冠层结构和光能利用的重要指标.随着遥感技术的不断发展,利用遥感数据获取大面积LAI已经成为监测作物生长和估产的重要手段.基于物理模型的LAI遥感反演方法经常假设作物冠层结构是均匀分布,然而,作为典型的垄行结构,作物冠层被公认为是介于连续植被与离散植被之间的一种过渡形式,而简单的均匀假设必然会给反演带来偏差.本文以农作物玉米为研究对象,首先重建了玉米三维冠层结构,并定量对比分析了一维辐射传输模型PROSAIL和三维辐射传输模型LESS在玉米冠层不同生长期的反射率差异,确定了玉米冠层的非均匀分布特征是引起PROSAIL模型模拟和反演误差的主要因素;然后,考虑到玉米冠层生长过程中聚集指数的变化特征,利用LESS模型定量计算了不同生育期玉米冠层结构对应的聚集指数,建立了聚集指数和有效叶面积指数(LAI_(e))之间的关系;进而,利用该关系对基于PROSAIL模型反演得到的LAI进行修正.结果表明,修正后的LAI精度有明显提高,R^(2)从0.27提高到了0.55.该方法有望提高中高分辨率遥感数据在农作物LAI反演精度.展开更多
为了探讨Landsat 8 OIL数据在LAI大范围反演方面的应用潜力,使用Landsat 8 OIL影像,通过PROSAIL辐射传输模型,采用3种波段组合(Band2-7,Band2-5,Band3-5)建立了3个模拟冠层反射率-叶面积指数(LAI)查找表,用2种代价函数(Geman and Mc Cl...为了探讨Landsat 8 OIL数据在LAI大范围反演方面的应用潜力,使用Landsat 8 OIL影像,通过PROSAIL辐射传输模型,采用3种波段组合(Band2-7,Band2-5,Band3-5)建立了3个模拟冠层反射率-叶面积指数(LAI)查找表,用2种代价函数(Geman and Mc Clure代价函数,均方根误差代价函数)实现了对玉米、土豆、森林LAI的定量反演,并用LAI-2200测量数据作为相对真值对反演精度进行评价。结果表明:(1)使用Landsat 8数据,通过PROSAIL模型反演叶面积指数的精度是可以接受的,RMSE范围为在[0.892 4,1.205 0],R2范围为[0.721 3,0.873 3]。(2)Band5(近红外),Band4(红)Band3(绿)的波段组合反演效果在3种组合中精度最高,平均RMSE=0.993 1,R2=0.787 3。(3)Geman and Mc Clure代价函数比常用的均方根误差代价函数得到了更高的反演精度,平均RMSE=0.940 5,R2=0.817 5。(4)相对最优的反演策略是Band5,Band4,Band3的波段组合结合GM代价函数,RMSE=0.892 4,R2=0.873 3。(5)存在玉米土豆的反演值普遍低于测量值,而森林的反演值普遍高于测量值的问题。展开更多
以大豆叶面积指数(Leaf area index,LAI)反演为研究目标,利用PROSAIL模型和遗传算法优化后的BP神经网络模型,分别对重组自交系(Recombinant Inbred Lines,RIL)和自然野生大豆种群的LAI进行反演。结果表明,在对人工定向培育的RIL大豆种...以大豆叶面积指数(Leaf area index,LAI)反演为研究目标,利用PROSAIL模型和遗传算法优化后的BP神经网络模型,分别对重组自交系(Recombinant Inbred Lines,RIL)和自然野生大豆种群的LAI进行反演。结果表明,在对人工定向培育的RIL大豆种群冠层叶片LAI反演中,PROSAIL模型表现出了更优异的反演能力,而对品种繁多的自然野生大豆种群LAI反演中,遗传算法优化后的BP神经网络模型表现出了更好的适用性,并且上述2种模型在始粒期(R5)时性能最佳,PROSAIL模型和遗传算法优化后的BP神经网络模型R2分别为0.89和0.85,RMSE分别为0.11和0.13,EA均为97%,典型生育期内的反演性能均优于全生育期综合反演性能。因此,针对同一农作物不同种群的表型特征反演,需要根据研究对象的特征来选择合适的模型,以便于精确的估测大豆长势情况,为农作物的规模化育种监测提供数据支持。展开更多
Biochemical components of Moso bamboo(Phyllostachys pubescens)are critical to physiological and ecological processes and play an important role in the material and energy cycles of the ecosystem.The coupled PROSPECT w...Biochemical components of Moso bamboo(Phyllostachys pubescens)are critical to physiological and ecological processes and play an important role in the material and energy cycles of the ecosystem.The coupled PROSPECT with SAIL(PROSAIL)radiative transfer model is widely used for vegetation biochemical component content inversion.However,the presence of leaf-eating pests,such as Pantana phyllostachysae Chao(PPC),weakens the performance of the model for estimating biochemical components of Moso bamboo and thus must be considered.Therefore,this study considered pest-induced stress signals associated with Sentinel-2A/B images and field data and established multiple sets of biochemical canopy reflectance look-up tables(LUTs)based on the PROSAIL framework by setting different parameter ranges according to infestation levels.Quantitative inversions of leaf area index(LAI),leaf chlorophyll content(LCC),and leaf equivalent water thickness(LEWT)were derived.The scale conversions from LCC to canopy chlorophyll content(CCC)and LEWT to canopy equivalent water thickness(CEWT)were calculated.The results showed that LAI,CCC,and CEWT were inversely related with PPC-induced stress.When applying multiple LUTs,the p-values were<0.01;the R2 values for LAI,CCC,and CEWT were 0.71,0.68,and 0.65 with root mean square error(RMSE)(normalized RMSE,NRMSE)values of 0.38(0.16),17.56μg cm-2(0.20),and 0.02 cm(0.51),respectively.Compared to the values obtained for the traditional PROSAIL model,for October,R2 values increased by 0.05 and 0.10 and NRMSE decreased by 0.09 and 0.02 for CCC and CEWT,respectively and RMSE decreased by 0.35μg cm-2 for CCC.The feasibility of the inverse strategy for integrating pest-induced stress factors into the PROSAIL model,while establishing multiple LUTs under different pest-induced damage levels,was successfully demonstrated and can potentially enhance future vegetation parameter inversion and monitoring of bamboo forest health and ecosystems.展开更多
基金supported by the National Natural Science Foundation of China (41975044)the Open Research Fund of the State Laboratory of Information Engineering in Surveying,Mapping,Remote Sensing,Wuhan University (20R02)+2 种基金the Fundamental Research Funds for the Central Universities,China University of Geosciences (Wuhan)(111-G1323520290)funded by SNSA (Dnr 96/16)the EU-Aid funded CASSECS Project。
文摘Leaf pigments are critical indicators of plant photosynthesis,stress,and physiological conditions.Inversion of radiative transfer models(RTMs)is a promising method for robustly retrieving leaf biochem-ical traits from canopy observations,and adding prior information has been effective in alleviating the“ill-posed”problem,a major challenge in model inversion.Canopy structure parameters,such as leaf area index(LAI)and average leaf inclination angle(ALA),can serve as prior information for leaf pigment retrie-val.Using canopy spectra simulated from the PROSAIL model,we estimated the effects of uncertainty in LAI and ALA used as prior information for lookup table-based inversions of leaf chlorophyll(C _(ab))and car-otenoid(C_(ar)).The retrieval accuracies of the two pigments were increased by use of the priors of LAI(RMSE of C_(ab) from 7.67 to 6.32μg cm^(-2),C_(ar) from 2.41 to 2.28μg cm^(-2))and ALA(RMSE of C_(ab) from 7.67 to 5.72μg cm^(-2),C_(ar) from 2.41 to 2.23μg cm^(-2)).However,this improvement deteriorated with an increase of additive and multiplicative uncertainties,and when 40% and 20% noise was added to LAI and ALA respectively,these priors ceased to increase retrieval accuracy.Validation using an experimental winter wheat dataset also showed that compared with C_(ar),the estimation accuracy of C_(ab) increased more or deteriorated less with uncertainty in prior canopy structure.This study demonstrates possible limita-tions of using prior information in RTM inversions for retrieval of leaf biochemistry,when large uncer-tainties are present.
文摘叶面积指数(leaf area index,LAI)作为衡量作物生长状况的关键参数,对其进行精准高效的反演对于作物监测、产量预测等活动至关重要。然而,传统经验模型在估算LAI时常存在计算负荷重、泛化能力弱等问题。为实现青贮玉米多时序LAI精准、高效估算,该研究以甘肃省民乐县的大田青贮玉米LAI为研究对象,结合Landsat-8多光谱影像与同期实地采集的LAI数据,提出了4种基于EFAST全局敏感性分析方法的机器学习混合反演模型(MLP-PROSAIL、SVR-PROSAIL、RF-PROSAIL和GBM-PROSAIL)。通过对PROSAIL模型的输入参数进行敏感性分析,以便确定参数敏感度并准确模拟输出冠层反射率光谱。进一步对Landsat-8多光谱数据进行预处理和波段变换,并采用地理配准工具结合反距离加权插值的策略减少其尺度差异。同时利用贝叶斯超参数寻优和正则化技术优化模型不同的参数类型和激活函数,得到4种改进模型用于训练LAI与光谱数据,通过5折交叉验证法和留一验证法对4种模型的反演性能进行验证并选出最优模型。优化后的模型性能均有明显提升,其中,GBM-PROSAIL模型反演性能最优,拟合精度R^(2)为0.93、均方根误差(RMSE)为0.42。MLP-PROSAIL、SVR-PROSAIL和RF-PROSAIL模型的拟合精度R^(2)依次为0.85、0.88、0.90,RMSE依次为0.80、0.69、0.51。根据GBM-PROSAIL模型绘制的研究区多时序LAI反演空间分布结果表明:不同生长期青贮玉米LAI值存在明显差异,能较好反映其生长过程。该研究提出的混合反演模型具有较高的性能及较强的鲁棒性,可为多时序、大尺度作物监测、产量预测相关研究提供方法与思路。
文摘叶面积指数(leaf area index,LAI)是反映植物冠层结构和光能利用的重要指标.随着遥感技术的不断发展,利用遥感数据获取大面积LAI已经成为监测作物生长和估产的重要手段.基于物理模型的LAI遥感反演方法经常假设作物冠层结构是均匀分布,然而,作为典型的垄行结构,作物冠层被公认为是介于连续植被与离散植被之间的一种过渡形式,而简单的均匀假设必然会给反演带来偏差.本文以农作物玉米为研究对象,首先重建了玉米三维冠层结构,并定量对比分析了一维辐射传输模型PROSAIL和三维辐射传输模型LESS在玉米冠层不同生长期的反射率差异,确定了玉米冠层的非均匀分布特征是引起PROSAIL模型模拟和反演误差的主要因素;然后,考虑到玉米冠层生长过程中聚集指数的变化特征,利用LESS模型定量计算了不同生育期玉米冠层结构对应的聚集指数,建立了聚集指数和有效叶面积指数(LAI_(e))之间的关系;进而,利用该关系对基于PROSAIL模型反演得到的LAI进行修正.结果表明,修正后的LAI精度有明显提高,R^(2)从0.27提高到了0.55.该方法有望提高中高分辨率遥感数据在农作物LAI反演精度.
文摘以大豆叶面积指数(Leaf area index,LAI)反演为研究目标,利用PROSAIL模型和遗传算法优化后的BP神经网络模型,分别对重组自交系(Recombinant Inbred Lines,RIL)和自然野生大豆种群的LAI进行反演。结果表明,在对人工定向培育的RIL大豆种群冠层叶片LAI反演中,PROSAIL模型表现出了更优异的反演能力,而对品种繁多的自然野生大豆种群LAI反演中,遗传算法优化后的BP神经网络模型表现出了更好的适用性,并且上述2种模型在始粒期(R5)时性能最佳,PROSAIL模型和遗传算法优化后的BP神经网络模型R2分别为0.89和0.85,RMSE分别为0.11和0.13,EA均为97%,典型生育期内的反演性能均优于全生育期综合反演性能。因此,针对同一农作物不同种群的表型特征反演,需要根据研究对象的特征来选择合适的模型,以便于精确的估测大豆长势情况,为农作物的规模化育种监测提供数据支持。
基金funded by the National Natural Science Foundation of China(42071300)the Fujian Province Natural Science(2020J01504)+4 种基金the China Postdoctoral Science Foundation(2018M630728)the Open Fund of Fujian Provincial Key Laboratory of Resources and Environment Monitoring&Sustainable Management and Utilization(ZD202102)the Program for Innovative Research Team in Science and Technology in Fujian Province University(KC190002)the Open Fund of University Key Lab of Geomatics Technology and Optimize Resources Utilization in Fujian Province(fafugeo201901)supported by the Research Project of Jinjiang Fuda Science and Education Park Development Center(2019-JJFDKY-17)。
文摘Biochemical components of Moso bamboo(Phyllostachys pubescens)are critical to physiological and ecological processes and play an important role in the material and energy cycles of the ecosystem.The coupled PROSPECT with SAIL(PROSAIL)radiative transfer model is widely used for vegetation biochemical component content inversion.However,the presence of leaf-eating pests,such as Pantana phyllostachysae Chao(PPC),weakens the performance of the model for estimating biochemical components of Moso bamboo and thus must be considered.Therefore,this study considered pest-induced stress signals associated with Sentinel-2A/B images and field data and established multiple sets of biochemical canopy reflectance look-up tables(LUTs)based on the PROSAIL framework by setting different parameter ranges according to infestation levels.Quantitative inversions of leaf area index(LAI),leaf chlorophyll content(LCC),and leaf equivalent water thickness(LEWT)were derived.The scale conversions from LCC to canopy chlorophyll content(CCC)and LEWT to canopy equivalent water thickness(CEWT)were calculated.The results showed that LAI,CCC,and CEWT were inversely related with PPC-induced stress.When applying multiple LUTs,the p-values were<0.01;the R2 values for LAI,CCC,and CEWT were 0.71,0.68,and 0.65 with root mean square error(RMSE)(normalized RMSE,NRMSE)values of 0.38(0.16),17.56μg cm-2(0.20),and 0.02 cm(0.51),respectively.Compared to the values obtained for the traditional PROSAIL model,for October,R2 values increased by 0.05 and 0.10 and NRMSE decreased by 0.09 and 0.02 for CCC and CEWT,respectively and RMSE decreased by 0.35μg cm-2 for CCC.The feasibility of the inverse strategy for integrating pest-induced stress factors into the PROSAIL model,while establishing multiple LUTs under different pest-induced damage levels,was successfully demonstrated and can potentially enhance future vegetation parameter inversion and monitoring of bamboo forest health and ecosystems.
