基于特征筛选结合PSO-BPNN和GA-BPNN算法的土壤重金属高光谱定量反演

以连州地区土壤重金属含量为研究对象,分析包括土壤原始光谱在内的经过数学变换后的光谱数据与重金属含量之间的相关性,再采用VISSA-IRIV算法进行光谱特征提取,分别建立偏最小二乘回归(PLSR)、BP神经网络(BPNN)、粒子群优化BP神经网络、遗传算法优化BP神经网络模型,对比获取土壤重金属元素Cr、Cu含量最优反演模型。结果表明：VISSA-IRIV算法实现了对光谱数据的高效降维;BPNN模型预测效果明显优于PLSR模型;经过优化的BP神经网络模型反演精度和稳定性得到了极大地提升,其中Cr、Cu元素的最佳反演模型组合分别为FD-GABPNN(R²=0.87、RMSE=13.82、RPD=2.95)、SNV-FD-PSO-BPNN(R²=0.92、RMSE=4.25、RPD=3.41)。该研究对土壤重金属含量的准确、快速分析提供了一种有效的方法,对实现土壤重金属污染治理具有重要的现实意义。     

一种基于分段偏最小二乘模型的土壤重金属遥感反演方法

土壤中重金属由于其毒性而成为最有害的环境污染物之一,利用遥感进行土壤重金属检测和分布制图是目前最为高效的手段。采用哨兵二号（Sentinel-2）多光谱影像与实测样品光谱数据,对山西省铜矿峪铜矿尾矿库及其周边农田土壤的铜（Cu）含量进行估算,利用68个土壤样品的反射光谱,优选出适合土壤铜含量预测的波段,结合分段偏最小二乘法（Piecewise Partial Least Squares Regression,P-PLSR）,对土壤铜含量进行估算,将模型用于Sentinel-2影像获得了Cu含量的空间分布。通过P-PLSR对实测样品光谱建模反演Cu含量的决定系数（R²）为0.89,预测偏差比（RPD）为2.82;利用Sentinel-2多光谱影像获得了该区域Cu元素含量空间分布,其Cu含量的估算精度R²为0.74,RPD为1.73,Cu含量高值区空间分布与尾矿库关系密切。Sentinel-2多光谱数据具有高空间分辨率（10、20和60 m）、高时间分辨率和幅宽大（290 km）等优势,通过敏感波段选择并建立反演模型,可实现大范围土壤环境制图。     

滨海盐土重金属含量高光谱遥感研究

高光谱遥感凭借其极高的光谱分辨率在获取有机质、矿物质等土壤组分定量信息的研究中表现出非凡的潜力。以如东县洋口镇为研究区,通过对土壤反射光谱的测量和同步的土壤化学分析,研究了土壤重金属Cr、Cu、Ni与土壤粘土矿物、铁锰氧化物以及碳酸盐之间的赋存关系。利用光谱一阶微分、倒数对数和连续统去除法对土壤光谱的处理,获得了土壤成分的特征波段,通过土壤重金属与土壤光谱变量的相关分析,并利用逐步回归分析方法,确立了3种重金属元素的最佳遥感模型。结果表明,研究区3种重金属与波长429 nm、470 nm、490 nm、1 430 nm、2 398 nm、2 455 nm处光谱变量具有很好的相关性,在所建立的逐步回归模型中,以一阶微分处理的模型精度最高。研究结果可以为高光谱遥感技术反演土壤重金属含量,进一步应用空间或航空遥感进行大尺度环境污染遥感、遥测信息提取和反演提供技术支撑。     

水稻冠层与土壤高光谱反演土壤重金属对比研究

针对土壤高光谱与水稻冠层高光谱反演多种土壤重金属含量精度差异较大的问题,探讨2种光谱反演土壤重金属适应性。通过对水稻冠层光谱与其土壤光谱的光谱指标变换,进行对比分析,在此基础上研究多元逐步回归、偏最小二乘回归在不同光谱指标下反演土壤重金属(Fe、Zn、Cu、Pb、Cd)的含量。结果表明,红光、近红外波段为土壤重金属含量敏感波段;水稻冠层波谱反演Fe、Zn、Pb、Cd含量的精度高于土壤波谱的反演精度;土壤波谱反演Cu含量的精度高于水稻冠层波谱反演精度。     

珠海一号高光谱遥感的表层土壤有机质含量反演方法

表层土壤有机质含量影响土壤光谱特性且在空间分布上呈异质性。采用光谱分辨率高、波段连续性强的高光谱遥感影像反演区域表层土壤有机质空间分布状况,可为精准农业提供科学管理依据。针对以往方法很少基于高光谱影像大尺度反演土壤表层有机质含量,以安徽省淮南市舜耕山以南的三和镇、曹庵镇为研究区,探索珠海一号高光谱遥感反演表层土壤有机质含量的方法。研究结果表明,研究区表层土壤有机质含量与珠海一号高光谱影像原始光谱反射率最大相关波段为656nm(r=-0.680);采用小波包分解原始光谱后,低频分量和高频分量与表层土壤有机质的最大相关性均有所提高,低频分量最大相关波段为656nm(r=-0.797),高频分量最大相关波段为700nm(r=-0.804)。采用多元线性回归对原始光谱、小波包分解低频分量、小波包分解高频分量建立土壤有机质预测模型取得了良好的效果,R2分别为0.747、0.770、0.789。依据小波包分解的低频分量、小波包分解的高频分量建立的基于高斯核变换的支持向量回归模型预测效果优于多元线性回归模型,预测值与实测值更接近。研究结果为开展大尺度遥感反演表层土壤有机质工作提供了新方法、新思路。     

基于高光谱的土壤重金属铜的反演研究

为探讨高光谱遥感反演红壤重金属铜含量的可行性,研究采集了34个红壤性土壤样品,通过对350~2 500 nm波段范围光谱曲线进行测试和分析,建立了不同的土壤光谱变量与重金属铜含量多元回归关系模型,分析了土壤重金属铜与土壤化学组分以及土壤特征光谱的关系。结果表明,土壤重金属铜含量与土壤全铁和镁含量显著相关,而与土壤有机质的相关性不显著,表明红壤性土壤粘土矿物对土壤铜含量影响较大;与重金属铜含量相关性较好的波段在830 nm、1 000 nm和2 250 nm附近,且一阶微分模型精度(79%)高于反射率模型(66.26%)和倒数对数模型(67%)的精度。因此,一阶微分高光谱反演模型具有较好的快速估算土壤中重金属铜含量的潜力。     

基于无人机高光谱数据的玉米叶面积指数估算

无人机高光谱遥感是低成本、高精度获取精细尺度农作物生物物理参数和生物化学参数的新型手段,以此快速反演叶面积指数(Leaf Area Index,LAI)对作物长势评价、产量预测具有重要意义。以山东禹城市玉米为研究对象,利用PROSAIL辐射传输模型模拟玉米冠层反射率获取LAI特征响应波段结合相关性定量分析获取对LAI变化最为敏感的波段,并以此计算6种植被指数(Vegetation Index,VI),利用6种回归模型分别对单一特征波段和VI进行反演建模,以实测LAI评定模型精度。研究表明,光谱反射率中516、636、702、760和867 nm等波段对LAI变化最为敏感,以此建立的单一特征波段反演模型预测LAI精度R~2为0.44～0.58;RMSE为0.16～0.18,其中636 nm建立的模型(LAI=21.86exp(-29.47R636))相比其他反演模型预测精度较高(R~2=0.58,RMSE=0.16);6种植被指数与LAI高度相关,相关性系数R2为0.85～0.86,以此建立的反演模型相比单一特征波段反演模型精度有所提高,R~2为0.66～0.72,RMSE为0.12～0.14;其中mNDVI构建的LAI估算模型(LAI=exp(2.76～1.77/mNDVI))精度最高(R~2=0.72,RMSE=0.13)。无人机高光谱遥感是快速、无损监测农作物生长信息的有效手段,为指导精细化尺度作物管理提供依据。     

基于无人机高光谱数据的玉米叶面积指数估算

无人机高光谱遥感是低成本、高精度获取精细尺度农作物生物物理参数和生物化学参数的新型手段,以此快速反演叶面积指数（Leaf Area Index, LAI）对作物长势评价、产量预测具有重要意义。以山东禹城市玉米为研究对象,利用PROSAIL辐射传输模型模拟玉米冠层反射率获取LAI特征响应波段结合相关性定量分析获取对LAI变化最为敏感的波段,并以此计算6种植被指数（Vegetation Index,VI）,利用6种回归模型分别对单一特征波段和VI进行反演建模,以实测LAI评定模型精度。研究表明,光谱反射率中516、636、702、760和867 nm等波段对LAI变化最为敏感,以此建立的单一特征波段反演模型预测LAI精度R2为0.44~0.58;RMSE为0.16~0.18,其中636 nm建立的模型(LAI=21.86exp(-29.47R₆₃₆))相比其他反演模型预测精度较高（R2=0.58,RMSE=0.16）;6种植被指数与LAI高度相关,相关性系数R ²为0.85~0.86,以此建立的反演模型相比单一特征波段反演模型精度有所提高,R2为0.66~0.72,RMSE为0.12~0.14;其中mNDVI构建的LAI估算模型（LAI=exp(2.76~1.77/mNDVI))精度最高（R2=0.72,RMSE=0.13）。无人机高光谱遥感是快速、无损监测农作物生长信息的有效手段,为指导精细化尺度作物管理提供依据。     

云南个旧矿区土壤锌污染遥感反演研究

土壤重金属锌污染作为现代工矿业发展的产物,已逐渐入侵到人类日常的生产和生活中,危害人们的身心健康。传统的重金属监测方法在面对大规模土壤环境监测时费时费力。遥感技术由于具有宏观、快速、高效的特点已成为新时代环境监测的重要工具。以云南个旧矿区为典型区,通过野外土壤样品采集、光谱与Zn元素测量,提出了乘积变换的波段变换方法以增强Zn元素与光谱敏感波段之间的相关性,应用其建立了Zn含量最优预测模型并基于ASTER影像开展了污染制图。研究表明:(1)Zn元素的最大相关波段是B515波段,该波段处于闪锌矿、红锌矿、菱锌矿等含锌矿物的吸收峰附近,是反演土壤锌元素的重要波段;(2)光谱乘积变换在突出Zn元素敏感波段的同时,最大程度地保留了土壤原有的敏感波段信息;(3)研究区土壤锌含量的高光谱反演模型中,偏最小二乘法建立的模型精度最高(建模精度R=0.90,验证精度R=0.70);(4)基于ASTER影像的反演结果表明了土壤Zn元素污染与矿业活动的显著相关性(制图验证精度R=0.694)。研究结果可以为遥感定量反演重金属含量,以及大规模的环境污染监测提供研究基础与技术支持。     

基于无人机高光谱数据的玉米叶面积指数估算

无人机高光谱遥感是低成本、高精度获取精细尺度农作物生物物理参数和生物化学参数的新型手段，以此快速反演叶面积指数（Leaf Area Index, LAI）对作物长势评价、产量预测具有重要意义。以山东禹城市玉米为研究对象，利用PROSAIL辐射传输模型模拟玉米冠层反射率获取LAI特征响应波段结合相关性定量分析获取对LAI变化最为敏感的波段，并以此计算6种植被指数（Vegetation Index,VI），利用6种回归模型分别对单一特征波段和VI进行反演建模，以实测LAI评定模型精度。研究表明，光谱反射率中516、636、702、760和867 nm等波段对LAI变化最为敏感，以此建立的单一特征波段反演模型预测LAI精度R2为0.44~0.58；RMSE为0.16~0.18，其中636 nm建立的模型(LAI=21.86exp(-29.47R₆₃₆))相比其他反演模型预测精度较高（R2=0.58，RMSE=0.16）；6种植被指数与LAI高度相关，相关性系数R ²为0.85~0.86，以此建立的反演模型相比单一特征波段反演模型精度有所提高，R2为0.66~0.72，RMSE为0.12~0.14；其中mNDVI构建的LAI估算模型（LAI=exp(2.76~1.77/mNDVI))精度最高（R2=0.72，RMSE=0.13）。无人机高光谱遥感是快速、无损监测农作物生长信息的有效手段，为指导精细化尺度作物管理提供依据。     

Integrating spectral indices with environmental parameters for estimating heavy metal concentrations in rice using a dynamic fuzzy neural-network model

A generalized dynamic fuzzy neural network (GDFNN) was created to estimate heavy metal concentrations in rice by integrating spectral indices and environmental parameters. Hyperspectral data, environmental parameters, and heavy metal content were collected from field experiments with different levels of heavy metal pollution (Cu and Cd). Input variables used in the GDFNN model were derived from 10 variables acquired by gray relational analysis. The assessment models for Cd and Cu concentration employed five and six input variables, respectively. The results showed that the GDFNN for estimating Cu and Cd concentrations in rice performed well at prediction with a compact network structure using the training, validation, and testing sets (for Cu, fuzzy rules=9, R² greater than 0.75, and RMSE less than 2.5; for Cd, fuzzy rules=9, R² greater than 0.75, and RMSE less than 1.0). The final GDFNN model was then compared with a back-propagation (BP) neural network model, adaptive-network-based fuzzy interference systems (ANFIS), and a regression model. The accuracies of GDFNN model prediction were usually slightly better than those of the other three models. This demonstrates that the GDFNN model is more suitable for predicting heavy metal concentrations in rice.     

基于偏最小二乘的土壤重金属铜含量高光谱估算

为探究高光谱数据估算土壤重金属铜含量的可行性,以石家庄市水源保护区褐土为研究对象,对不同光谱变换数据与重金属铜含量做了相关分析,建立了土壤重金属铜的单光谱变换指标偏最小二乘模型和多光谱变换指标偏最小二乘模型。结果表明：光谱反射率（R）经倒数一阶微分（RTFD）变换后与铜含量的相关性有所提高;光谱敏感波段为418、427、435、446、490、673、1 909、1 920和2 221 nm,基本位于土壤氧化铁、粘土矿物的特征吸收区域;对土壤重金属铜含量估算效果最好的单光谱变换指标偏最小二乘模型为RTFD模型,其模型决定系数（R²）为0.649,均方根误差（RMSE）为1.477;多光谱变换指标偏最小二乘模型R²和RMSE分别为0.751和1.162,建模效果优于单光谱变换指标模型。研究结果可为北方地区褐土类型土壤重金属铜的高光谱估算提供借鉴。     

基于偏最小二乘的土壤重金属铜含量高光谱估算

为探究高光谱数据估算土壤重金属铜含量的可行性，以石家庄市水源保护区褐土为研究对象，对不同光谱变换数据与重金属铜含量做了相关分析，建立了土壤重金属铜的单光谱变换指标偏最小二乘模型和多光谱变换指标偏最小二乘模型。结果表明：光谱反射率（R）经倒数一阶微分（RTFD）变换后与铜含量的相关性有所提高；光谱敏感波段为418、427、435、446、490、673、1 909、1 920和2 221 nm，基本位于土壤氧化铁、粘土矿物的特征吸收区域；对土壤重金属铜含量估算效果最好的单光谱变换指标偏最小二乘模型为RTFD模型，其模型决定系数（R²）为0.649，均方根误差（RMSE）为1.477；多光谱变换指标偏最小二乘模型R²和RMSE分别为0.751和1.162，建模效果优于单光谱变换指标模型。研究结果可为北方地区褐土类型土壤重金属铜的高光谱估算提供借鉴。     

Remote Sensing Inversion of Soil Zinc Pollution in Gejiu Mining Area of Yunnan

As a product of the development of modern industry and mining industry,heavy metal Zn pollution has gradually invaded the daily production and life of human beings,which is harmful to our physical and mental health.In dealing with large-scale soil environmental monitoring.The traditional heavy metal monitoring method is time-consuming and laborious.Due to its characteristics of high speed,high speed and high efficiency,remote sensing technology has become an important tool for environmental monitoring in the new era.This study takes Yunnan Gejiu mining area as a typical area,collecting sample in field soil and measurement of soil sample spectra and Zn content.Then the band transform method based on the multiplicative transformation was proposed to enhance product conversion relationship between Zn elements and spectral sensitive bands,using the established prediction model and optimal Zn content based on ASTER images to carry out pollution mapping.Research shows that：①the maximum correlation band of Zn elements is the B515 band,close to the absorption peak of sphalerite and smithsonite zinc containing minerals,is an important band of zinc element inversion of soil;②the spectral multiplicative transformation can highlight the sensitive bands of Zn elements,and retain the most sensitive information of the original soil;③in the hypersecretion inversion model of soil zinc content in the study area,the precision of the model established by partial least squares(R=0.90)is the highest(R=0.70);④The inversion results based on ASTER images show that there is a significant correlation between soil Zn pollution and mining activities(Verification accuracy of map R=0.694).The results of this study can provide the basis and technical support for remote sensing quantitative inversion of heavy metal content and large-scale environmental pollution monitoring.     

Soil Heavy Metals Estimation based on Hyperspectral in Urban Residential

To explore the possibility of using soil spectral reflectance to estimate soil heavy metal content in urban residential area，this study chooses 30 soil samples of Cu，Pb and Zn in Minhang Residential area，Shanghai Province.Through the spectral factor transform to highlight its eigenvalues，constructed Multiple Linear Stepwise Regression(MLSR) model and Partial Least Squares Regression(PLSR) model based on spectral reflectance of soil heavy metals.The results show that the reciprocal first-order and the logarithmic first-order differential transformation can effectively enhance the heavy metal soil spectral characteristics.The best characteristic bands of Cu，Pb and Zn are 1 042.7 nm、706.84 nm and 1 404.8 nm.In terms of model stability and accuracy，PLSR model is better than MLSR model.The RMSE of Cu and Zn were only about 10% of the mean value of heavy metals in the study area，and the accuracy of the model was high.Compared with Cu and Zn，the R2 of Pb is between 0.64~0.88 which with higher model stability.By preprocessing the spectral data，the partial least-squares regression can effectively improve the accuracy of estimating the heavy metal content in urban residential areas. 〖WTHZ〗Key words:〖WT〗 Urban residential area;Soil heavy metals;Hyperspectral;Multivariate Linear Stepwise Regression(MLSR) model;Partial Least Squares Regression(PLSR) model 〖HT〗〖ST〗〖HJ〗〖WT〗〖JP〗〖LM〗     

Comparing Sentinel-2 MSI and Landsat 8 OLI in soil salinity detection: a case study of agricultural lands in coastal North Carolina

This study presents the first comparison of Landsat 8 Operational Land Imager (OLI) and Sentinel-2 Multispectral Instrument (MSI) in identifying soil salinity using soil physiochemical, spectral, statistical, and image analysis techniques. By the end of the century, intermediate sea level rise scenarios project approximately 1.3 meters of sea level rise along the coast of the southeastern United States. One of the most vulnerable areas is Hyde County, North Carolina, where 1140 km² of agricultural lands are being salinized, endangering 4,200 people and $40 million USD of property. To determine the best multispectral sensor to map the extent of salinization, this study compared the feasibility of OLI and MSI to estimate electrical conductivity (EC). The EC of field samples were correlated with handheld spectrometer spectra resampled into multispectral sensor bands. Using an iterative ordinary least squares regression, it was found that EC was sensitive to OLI bands 2 (452 nm – 512 nm) and 4 (636 nm – 673 nm) and MSI bands 2 (457.5 nm – 522.5 nm) and 4 (650 nm – 680 nm). Respectively, the R²_Adj and Root Mean Square Error (RMSE) of 0.04–0.54 and 1.15 for OLI, and 0.05–0.67 and 1.17 for MSI, suggests that the two sensors have similar salinity modelling skill. The extracted saline soils make up approximately 1,703 hectares for OLI and 118 hectares for MSI, indicating overestimation from the OLI image due to its coarser spatial resolution. Additionally, field samples indicate that nearby vegetated land is saline, indicating an underestimation of total impacted land. As sea levels rise, accurately monitoring soil salinization will be critical to protecting coastal agricultural lands. MSI’s spatial and temporal resolution makes it superior to OLI for salinity tracking though they have roughly equivalent spectral resolutions. This study demonstrates that visible spectral bands are sensitive to soil salinity with the Blue and Red spectral ranges producing the highest model accuracy; however, the low accuracies for both sensors indicate the need of narrowband sensors. The HyspIRI to be launched in the early 2020s by NASA may provide ideal data source in soil salinity studies.     

Potential signatures of heavy metal complexes in lichen reflectance spectra

Lichens are sensitive to atmospheric pollutants emitted from anthropogenic activities and are thus effective biomonitors. A variety of heavy metals, such as nickel (Ni), iron (Fe), lead (Pb), copper (Cu), and cadmium (Cd), can be emitted by metal smelters. The purpose of this study was twofold: (1) to measure the spectral reflectance properties (350–2500 nm) of expected heavy metal complexes in lichens (oxalates and sulphides); and (2) to determine whether these complexes contribute features to reflectance spectra of lichens from the vicinity of a heavy metal smelter. Some metal oxalate spectra are characterized by crystal field transition absorption bands in the 500–1300 nm region, which are specific to the particular metal cation they contain and its oxidation state. The 1900–2500 nm region exhibits multiple absorption bands attributable to the oxalate molecule. The metal sulphide reflectance spectra are characterized by generally low reflectance and few if any strong or diagnostic spectral features; those that are found can be related to a specific cation and its oxidation state. These spectra were used to determine whether reflectance spectra of a diverse suite of lichens collected downwind of a smelter showed spectral evidence indicative of heavy metal oxalates or sulphides. The lichen spectra, coupled with the oxalate and sulphide spectra and independently determined heavy metal concentration, failed to reveal spectral features that could be unambiguously related to heavy metal complexes. This was likely due to a number of causes: lichen reflectance spectra have absorption bands that overlap those of oxalates; oxalate and sulphide concentrations may have been too low to allow for their unambiguous identification, and lichen spectra are naturally diverse in the region below 1300 nm. There were no strong or significant linear trends between metal concentrations and distance from the smelter (coefficient of determination (R²) values <0.05), or between absorption band depths in the lichen spectra and distance from the smelter (R² values <0.06). This was likely due to the inclusion of multiple lichen species in the analysis, which may interact with airborne pollutants in different ways, and microenvironmental effects.     

Hyperspectral field estimation and remote-sensing inversion of salt content in coastal saline soils of the Yellow River Delta

This study explored hyperspectral field and satellite-based remote sensing of soil salt content. Using Kenli County in the Yellow River Delta as the study area, in situ soil field spectra and satellite-based remote-sensing images were integrated with laboratory measurements of soil sample salinity to improve remote sensing-based soil salt estimation and inversion procedures. First, the narrow-band hyperspectral reflectance field data were used to model the wide-band reflectance data from Landsat 7. Second, the bands and spectral features sensitive to soil salt content were identified through correlation analysis and band combination. Stepwise multiple linear regression was used to find a best model, which was then inverted to predict soil salt content using remote-sensing images from Landsat 7 and Landsat 8. The applicability of the model was verified by ground-checking the inversion results. The results show that the bands sensitive to soil salinity are mainly in the visible and near-infrared (NIR) regions. Combining information from these bands can eliminate some background effects and significantly improve the correlation with salinity. The best model of soil salinity is y = 1.345 ? 25.898 × g_SWIR1 ? 245.440 × g_Red × (g_Red ? g_NIR) ? 0.252 × (g_Red + g_NIR)/(g_Red ? g_NIR) ? 19.563 × (g_Red ? g_SWIR1). This model has a coefficient of determination (R²) of 0.896, a verification R² of 0.867, a relative prediction deviation (RPD) of 2.135, and a root mean square error (RMSE) of 0.264. The model fits well and is highly stable. The inversion results based on Landsat 7 and Landsat 8 images are consistent with the actual situation of soil salinity in the study area. This study provides an effective and feasible method for the estimation of soil salt content in coastal regions based on field spectral measurements and remote-sensing inversion.     

Using WorldView-2 bands and indices to predict bronze bug (Thaumastocoris peregrinus) damage in plantation forests

Thaumastocoris peregrinus is an insect that causes significant damage to Eucalyptus plantations internationally. This bug inhibits the photosynthetic ability of the tree, resulting in stunted growth and even death of severely infested trees. This study uses high spatial resolution satellite imagery (WorldView-2 sensor data), with unique band settings for the prediction of T. peregrinus damage in plantation forests using partial least squares (PLS) regression. The PLS models developed from the WorldView-2 sensor bands and indices were inverted to map the severity of the damage caused by the pest. The WorldView-2 sensor bands and indices predicted T. peregrinus damage with an R ² value of 0.65 and a root mean square error (RMSE) of 3.62% on an independent test data set. The red-edge and near-infrared bands of the WorldView-2 sensor and pigment-specific indices and red-edge indices were identified as significant bands by variable importance scores for the prediction of T. peregrinus damage. This study demonstrates the potential of WorldView-2 sensor data in successfully predicting T. peregrinus damage using PLS regression and identifies important spectral variables for the prediction of forest damage in plantation forests.     

Inverting a forest canopy reflectance model to retrieve the overstorey and understorey leaf area index for forest stands

In this article, the Kuusk–Nilson forest reflectance and transmittance (FRT) model was inverted to retrieve the overstorey and understorey leaf area index (OU-LAI) of forest stands in the Longmenhe forest nature reserve in China. Data from detailed sample sites were collected in 30 forest stands representing the typical vegetation community in the study area. An uncertainty and sensitivity matrix (USM) was used to analyse the sensitivity of the FRT model parameters based on these data. The results indicated that overstorey LAI strongly influenced stand reflectance, whereas understorey LAI had a much lower impact. To predict OU-LAI in forest stands, FRT model inversion is carried out by minimizing a merit function that provides a measure of the difference between the reflectance simulated by the FRT model and the reflectance originating from optimal band selection of Hyperion data. Various combinations of Hyperion bands were tested to evaluate the most effective wavelengths for the inversion of OU-LAI. The best estimates from 17 Hyperion bands (5 VIS, 8 NIR, 4 SWIR) by the FRT model inversion showed an R ²?=?0.41 and RMSE/mean?=?0.21 for overstorey LAI and R ²?=?0.49 and RMSE/mean?=?0.91 for understorey LAI. Advantages and disadvantages of FRT inversion for retrieval OU-LAI combined with Hyperion data are discussed.