基于高分一号WFV影像的随机森林算法反演水稻LAI

叶面积指数(Leaf Area Index,LAI)是植被生长状况的重要指标,反映了农田生态系统的生产力水平。以江苏省东台市水稻田为研究区,基于多时相高分一号WFV影像提取的水稻植被指数数据,结合样区同步测量的不同生长期水稻叶面积指数数据,利用随机森林算法构建研究区水稻LAI反演模型。研究结果表明:随机森林算法反演的研究区水稻LAI与实测验证值相关性较好,R2达到0.88,RMSE仅为1.03,能准确反映研究区水稻LAI生长季的变化趋势,不同时段LAI测量值与反演值相对误差均值为15%,且GF-1WFV影像对研究区水、路网的分辨能力较高,总体上适用于农田LAI的反演。     

基于BP神经网络的夏玉米多生育期叶面积指数反演研究

叶面积指数（Leaf Area Index,LAI）是生物地球化学循环中重要的植被结构参数。针对目前基于我国GF-1 WFV卫星影像的夏玉米多生育期LAI反演研究较少的问题,基于不同隐含层构建BP神经网络模型（BP1模型和BP2模型）,对比分析BP1模型、BP2模型和6种统计模型（NDVI、RVI、DVI、EVI、SAVI、ARVI）反演之间的精度差异,并根据实测数据绘制BP1模型和BP2模型的夏玉米多生育期LAI动态变化图。结果表明:LAI与6种常用的统计模型均有良好相关性,其中NDVI指数方程式回归模型拟合度最优;BP神经网络模型整体R ²略小于统计模型,而RMSE则小于统计模型,取得了与实测值差异更小的结果,统计模型与BP神经网络模型各有优劣之处;BP2模型在R ²和RMSE均优于BP1模型,能获得更为精确的反演值,BP2整体预测精度更高;基于BP神经网络模拟夏玉米生育期反演,LAI值呈现缓慢升高—快速增长—逐渐减小的S型变化过程,基本符合作物生长规律。该研究结合不同隐含层建立的BP神经网络模型,为GF-1卫星在作物叶面积指数多生育期反演的应用推广提供了方法支撑。     

基于宽波段和窄波段植被指数的草地LAI反演对比研究

叶面积指数是一个重要的植被生理生态参数,为探讨不同植被指数反演叶面积指数的可行性,基于同空间分辨率不同光谱分辨率的HJ\|1B CCD1和Hyperion遥感影像数据,以内蒙古自治区赤峰市克斯克腾旗贡格尔草原为研究对象,选取几种常见宽波段植被指数和高光谱窄波段植被指数并结合4种常用回归模型,比较分析了不同植被指数反演叶面积指数的精度。结果表明:对于全部植被指数而言,PVI、MSAVI等综合考虑了土壤、环境等因素的植被指数较传统植被指数NDVI、RVI反演草地LAI精度更高。通过对比发现,在反演草地LAI方面,窄波段植被指数比宽波段植被指数表现出明显的优势。其中,窄波段垂直植被指数PVI验证模型的确定性系数R²为0.65,均方根误差RMSE为0.15,说明实测LAI和模拟LAI值之间具有较好的变化一致性。最后基于Hyperion影像和窄波段垂直植被指数PVI的估算模型生成研究区叶面积指数空间分布图。     

融合遥感先验信息的叶面积指数反演

借助植被辐射传输模型,利用遥感观测数据估算LAI是一种较为可靠和稳健的反演方法。然而,地表的复杂性、遥感观测的有限性以及自相关性导致遥感数据包含的信息量不足,不能完全支持LAI等地表参数的估算,易造成“病态”反演。在遥感反演过程中引入先验知识能够有效地解决该问题。研究基于遥感数据提取LAI先验信息,并将其用于代价函数的构建,利用PROSAIL辐射传输模型和遗传算法,分别在500 m和250 m尺度反演LAI。将高空间分辨率LAI分别升尺度到500 m和250 m,验证对应尺度LAI结果,评价引入先验信息对于提高LAI反演精度的作用。研究表明,引入先验信息有助于提高不同分辨率下LAI反演精度,且先验信息的质量一定程度上也影响着LAI反演结果。与未加入先验信息的LAI反演结果相比,以MODIS LAI产品作为先验信息反演的500 m尺度LAI结果精度R²由0.55提高至0.65,RMSE由1.29下降至0.38。在250 m尺度,以500 m LAI反演结果作为先验信息反演的叶面积指数,其精度优于以MODIS LAI产品为先验知识的估算结果,验证精度R²增加了0.08,RMSE减少了0.18。研究使用的先验信息主要来自遥感数据本身,没有地面实测数据的参与,在此基础上发展的多分辨率LAI反演方法具有估算大区域尺度LAI的应用潜力。     

基于DART模型和随机森林估算山地叶面积指数

地形效应会使遥感影像中的地表反射率发生畸变,进而影响基于反射率估算的叶面积指数(Leaf Area Index,LAI)精度。为了减弱或消除地形对LAI反演的影响,基于三维辐射传输模型DART(Discrete Anisotropic Radiative Transfer)构建坡地反射率与LAI数据集作为训练数据。以反射率为输入,LAI为输出,利用随机森林算法进行训练,构建山地LAI反演模型。结合实际遥感影像数据实现山地LAI的估算,并利用实测数据对反演结果开展精度评价。同时,基于DART模型和随机森林构建了平地LAI反演模型作为参照以评价本文发展方法的有效性。结果表明：考虑了地形影响的山地LAI反演模型具有较强的估算能力,验证结果的精度(决定系数(R²)=0.57,均方根误差(RMSE)=0.77 m²/m²)优于平地反演模型(R2=0.46,RMSE=0.86 m2/m2);基于DART模型构建的山地反演模型能够捕捉到坡度和坡向对地表反射率的影响,其反演结果较好地还原了研究区LAI的空间分布,与地面真实情况接近。研究...     

基于吉林一号光谱星影像的农作物叶面积指数反演

吉林一号光谱星的发射提高了我国对地观测能力，并且在农业定量反演方面具有较大的潜力，为了准确、有效地反演农作物关键参数，分析吉林一号光谱星影像的反演能力具有重要意义。以内蒙古乌拉特前旗、正蓝旗、科尔沁右翼前旗的农田为研究区，基于吉林一号光谱星影像，使用优化后的PROSAIL模型和曲线匹配算法，对不同物候期内的玉米和水稻叶面积指数(LAI)进行了反演，并结合实测LAI数据进行了精度验证。结果表明：优化后的PROSAIL模型其参数范围和参数步长更适用于农作物LAI反演，在保证精度的前提下精简了查找表的容量；基于特征值的曲线匹配算法在空间分布高度一致、误差绝对值均值为0.41的情况下，计算效率平均提高了41.43%；研究区不同物候期内的玉米和水稻LAI反演精度R²为0.72～0.9,RMSE为0.32～0.49。其中，玉米开花期精度最高(R²=0.9,RMSE=0.4)，玉米成熟期精度最低(R²=0.72,RMSE=0.47)。综上所述，基于吉林一号光谱星影像反演农作物LAI具有精度高、误差小的特点，研究结果可为该数据在农作物L...     

融合遥感先验信息的叶面积指数反演

借助植被辐射传输模型，利用遥感观测数据估算LAI是一种较为可靠和稳健的反演方法。然而，地表的复杂性、遥感观测的有限性以及自相关性导致遥感数据包含的信息量不足，不能完全支持LAI等地表参数的估算，易造成“病态”反演。在遥感反演过程中引入先验知识能够有效地解决该问题。研究基于遥感数据提取LAI先验信息，并将其用于代价函数的构建，利用PROSAIL辐射传输模型和遗传算法，分别在500 m和250 m尺度反演LAI。将高空间分辨率LAI分别升尺度到500 m和250 m，验证对应尺度LAI结果，评价引入先验信息对于提高LAI反演精度的作用。研究表明，引入先验信息有助于提高不同分辨率下LAI反演精度，且先验信息的质量一定程度上也影响着LAI反演结果。与未加入先验信息的LAI反演结果相比，以MODIS LAI产品作为先验信息反演的500 m尺度LAI结果精度R²由0.55提高至0.65，RMSE由1.29下降至0.38。在250 m尺度，以500 m LAI反演结果作为先验信息反演的叶面积指数，其精度优于以MODIS LAI产品为先验知识的估算结果，验证精度R²增加了0.08，RMSE减少了0.18。研究使用的先验信息主要来自遥感数据本身，没有地面实测数据的参与，在此基础上发展的多分辨率LAI反演方法具有估算大区域尺度LAI的应用潜力。     

基于无人机遥感的叶面积指数反演

叶面积指数定量遥感产品的真实性检验需要地面数据进行支撑。目前常用的叶面积指数测量仪器,如LAI2000、AccuPAR、Sunscan、Demon和TRAC等,需要工作人员进入样地进行手持测量,效率较低,人工测量引入的不确定性大。近年来基于无线传感器网络技术进行叶面积指数长时间自动观测取得了很多进展,但是投入成本大、移动不便等因素制约了其大范围应用。随着无人机的快速发展,利用无人机采集遥感数据具有极大的灵活性。本文利用轻型无人机获取了玉米地不同生长期的高分辨率光学影像,采用图像处理的算法进行植被与非植被的区分,最后利用辐射传输模型与聚集指数理论进行了叶面积指数反演。通过对比表明,在玉米成熟前期,反演得到的叶面积指数与LAI2200采集得到的数据,以及LI-3000C得到的真实叶面积指数有较高的一致性。基于无人机影像的LAI测量方法可作为一种快速准确的手段得以推广应用。     

联合无人机激光雷达和高光谱数据反演玉米叶面积指数

叶面积指数（Leaf Area Index, LAI）是作物长势监测及产量估算的重要指标,准确高效的LAI反演对农田经济的宏观管理具有重要作用。研究探索了联合无人机激光雷达（Light Detec-tion and Ranging, LiDAR） 和高光谱数据反演玉米叶面积指数的潜力,并分析了LiDAR数据不同采样尺寸、高度阈值、点密度对LAI反演精度的影响同时确定三者的最优值。该研究分别从重采样的LiDAR数据和高光谱影像中提取了LiDAR变量和植被指数,然后基于偏最小二乘回归（Partial Least Square Regression,PLSR）和随机森林（Random Forest, RF） 回归两种算法分别利用LiDAR变量、植被指数、联合LiDAR变量和植被指数构建预测模型,并确定反演玉米LAI的最优预测模型。结果表明：反演玉米LAI的最优采样尺寸、高度阈值、点密度分别为5.5 m、0.55 m、18 points/m²,研究发现最高的点密度（420 points/m²）并没有产生最优的玉米LAI反演精度,因此单独依靠增加点密度的方法提高LAI的反演精度并不可靠。基于LiDAR变量获得的LAI反演精度（PLSR：R²=0.874,RMSE=0.317;RF：R²=0.942,RMSE=0.222）高于基于植被指数获得的LAI反演精度（PLSR： R²=0.741,RMSE=0.454;RF：R²=0.861,RMSE=0.338）,而使用组合变量构建预测模型的反演精度（PLSR：R²=0.885, RMSE=0.304;RF：R²=0.950,RMSE=0.203）优于使用单一变量建立的LAI预测模型,其中利用联合LiDAR变量和植被指数建立的随机森林回归模型为最优预测模型。因此,将两种数据源融合在提高植被LAI反演精度方面具有一定的潜力。     

融合遥感先验信息的叶面积指数反演

借助植被辐射传输模型,利用遥感观测数据估算LAI是一种较为可靠和稳健的反演方法。然而,地表的复杂性、遥感观测的有限性以及自相关性导致遥感数据包含的信息量不足,不能完全支持LAI等地表参数的估算,易造成"病态"反演。在遥感反演过程中引入先验知识能够有效地解决该问题。研究基于遥感数据提取LAI先验信息,并将其用于代价函数的构建,利用PROSAIL辐射传输模型和遗传算法,分别在500 m和250 m尺度反演LAI。将高空间分辨率LAI分别升尺度到500 m和250 m,验证对应尺度LAI结果,评价引入先验信息对于提高LAI反演精度的作用。研究表明,引入先验信息有助于提高不同分辨率下LAI反演精度,且先验信息的质量一定程度上也影响着LAI反演结果。与未加入先验信息的LAI反演结果相比,以MODIS LAI产品作为先验信息反演的500 m尺度LAI结果精度R2由0.55提高至0.65,RMSE由1.29下降至0.38。在250 m尺度,以500 m LAI反演结果作为先验信息反演的叶面积指数,其精度优于以MODIS LAI产品为先验知识的估算结果,验证精度R2增加了0.08,RMSE减少了0.18。研究使用的先验信息主要来自遥感数据本身,没有地面实测数据的参与,在此基础上发展的多分辨率LAI反演方法具有估算大区域尺度LAI的应用潜力。     

The relationship between the leaf area index (LAI) of rice and the C‐band SAR vertical/horizontal (VV/HH) polarization ratio

We investigated the relationship between the leaf area index (LAI) of rice and the ENVISAT Advanced Synthetic Aperture Radar (ASAR) vertical/horizontal (VV/HH) polarization ratio. Four alternating polarization ASAR images of swaths IS4 and IS5 over rice fields were used in the study. The VV/HH polarization ratio correlates well with the field‐measured LAI and an empirical relationship was established to estimate the LAI of rice using the VV/HH polarization ratio. A theoretical radiative transfer model was adopted to analyse the relationship. The error of the estimated LAI was 0.17 for the test site and a better correlation was found when LAI was less than 3.5. The results suggest that ASAR alternating polarization data can be used to estimate the LAI of rice for wide‐area monitoring of rice growth.     

Analysis of ENVISAT ASAR data for forest parameter retrieval and forest type classification—a case study over deciduous forests of central India

The present paper gives an account of potential of Environment Satellite‐Advanced Synthetic Aperture Radar (ENVISAT‐ASAR) C‐band data in forest parameter retrieval and forest type classification over deciduous forests of Tadoba Andhari Tiger Reserve (TATR), central India. Ground data on phyto‐sociology and Leaf Area Index (LAI) over the study area was collected in 23 sampling points (20m×20m) over the study area. Phyto‐sociological data collected over the study area was used to compute plot‐wise biometric parameters like basal area, volume, stem density and dominant height. ENVISAT ASAR data covering the study area, pertaining to 24 November 2005, has been geo‐referenced and digital number (DN) values were converted to radar backscatter values. Regression analysis between backscatter and the retrieved biometric variables has been done to explain the relationships between SAR backscatter and forest parameters. Analysis showed a significant correlation between backscatter and biometric parameters and backscatter values typically increased with increase in basal area, volume, stem density and dominant height. The scatter observed between ASAR backscatter and stem density, basal area and dominant height suggested limitation of C‐band data in estimating biometric variables in heterogeneous forest systems. Further, ASAR data was used in conjunction with Indian Remote sensing Satellite (IRS‐P6)—Linear Imaging Self Scanner (LISS) III data of 16 October 2004 to classify the study area into different land use/land cover (LU/LC) classes. Various texture and adaptive filters were applied on ASAR image to reduce speckle noise and enhance image features. An attempt is made to merge ASAR image with LISS‐III to enhance feature discrimination. Training sets corresponding to the ground data have been used to derive confusion matrices for the ASAR and LISS‐III images. Results suggested better discrimination of vegetation types in the merged data suggesting the possible use of ASAR data in forest type discrimination.     

Integration of multitemporal/polarization C‐band SAR data sets for land‐cover classification

This paper investigates the potential of multitemporal/polarization C‐band SAR data for land‐cover classification. Multitemporal Radarsat‐1 data with HH polarization and ENVISAT ASAR data with VV polarization acquired in the Yedang plain, Korea are used for the classification of typical five land‐cover classes in an agricultural area. The presented methodologies consist of two analytical stages: one for feature extraction and the other for classification based on the combination of features. Both a traditional SAR signal property analysis‐based approach and principal‐component analysis (PCA) are applied in the feature extraction stage. Special concerns are in the interpretation of each principal component by using principal‐component loading. The tau model applied as a decision‐level fusion methodology can provide a formal framework in which the posteriori probabilities derived from different sensor data can be combined. From the case study results, the combination of PCA‐based features showed improved classification accuracy for both Radarsat‐1 and ENVISAT ASAR data, as compared with the traditional SAR signal property analysis‐based approach. The integration of PCA‐based features based on multiple polarization (i.e. HH from Radarsat‐1, and both VV and VH from ENVISAT ASAR) and different incidence angles contributed to a significant improvement of discrimination capability for dry fields which could not be properly classified by using only Radarsat‐1 or ENVISAT ASAR data, and thus showed the best classification accuracy. The results of this case study indicate that the use of multiple polarization SAR data with a proper feature extraction stage would improve classification accuracy in multitemporal SAR data classification, although further consideration should be given to the polarization and incidence angle dependency of complex land‐cover classes through more experiments.     

A semi‐empirical backscattering model for estimation of leaf area index (LAI) of rice in southern China

Most paddy rice in southern China grows in warm, humid and rainy areas where it is hard to acquire optical remote sensing data. In this study, a semi‐empirical backscattering model was proposed to estimate leaf area index (LAI) of rice in the area using ENVISAT Advanced Synthetic Aperture Radar (ASAR) alternating polarization data. Ground measurements of LAI, water content and height of rice in the test site were collected and the model fitted at the same time as the acquisition of ASAR data. LAI estimated from the model was compared with ground measurements to evaluate the accuracy of the model. The results showed that the model provides a promising alternative to optical remote sensing data for predicting LAI of rice in southern China.     

On the LAI of mixed soils–forests regions

In homogeneous forest textures, it has been recently confirmed experimentally that, for sufficiently large ground surfaces, the Leaf Area Index (LAI) has weak variations with respect to ground surface variations. This allows computing the LAI of mixed pixels on regions composed of forests and soils, with the use of the Perpendicular Vegetation Index (PVI). In the present paper, we study the accuracy of the method with experimental data.     

Retrieval Rice Leaf Area Index Using Random Forest Algorithm based on GF-1 WFV Remote Sensing Data

Leaf Area Index(LAI) is an important indicator of vegetation growth and reflects the productivity of farmland ecosystems.In this study,rice LAI was mapping using LAI retrieved model based on rice vegetation indexes from multi\|temporal GF\|1 WFV and situ LAI measurements data obtained in different rice growing periods over rice fields taking Dongtai county,Jiangsu province as a case study.The LAI retrieval model was constructed using random forest algorithm(RF).Results showed that the RF model achieved high accuracy that the RMSE was 1.03 and the coefficients of determination(R²) between retrieved LAI and measured LAI reached 0.88.The mean relative error between retrieved LAI and measured LAI in different growing periods was 15%.The trend of rice LAI could be reflected by the retrieved value and GF\|1 WFV data has high ability to distinguish the waters and road network in study area.     

Assimilation of leaf area index derived from ASAR and MERIS data into CERES-Wheat model to map wheat yield

This study presents a method to assimilate leaf area index retrieved from ENVISAT ASAR and MERIS data into CERES-Wheat crop growth model with the objective to improve the accuracy of the wheat yield predictions at catchment scale. The assimilation method consists in re-initialising the model with optimal input parameters allowing a better temporal agreement between the LAI simulated by the model and the LAI estimated by remote sensing data. A variational assimilation algorithm has been applied to minimise the difference between simulated and remotely-sensed LAI and to determine the optimal set of input parameters. After the re-initialisation, the wheat yield maps have been obtained and their accuracy evaluated.The method has been applied over Matera site located in Southern Italy and validated by using the dataset of an experimental campaign carried out during the 2004 wheat growing season.Results indicate that, LAI maps retrieved from MERIS and ASAR data can be effectively assimilated into CERES-Wheat model thus leading to accuracies of the yield maps ranging from 360 kg/ha to 420 kg/ha.     

Soil moisture mapping based on ASAR/ENVISAT radar data over a Sahelian region

The analysis of feedbacks between continental surfaces and the atmosphere is one of the key factors to understanding African Monsoon dynamics. For this reason, the monitoring of surface parameters, in particular soil moisture, is very important. Satellite remote sensing appears to be the most suitable means of obtaining data relevant to such parameters. The present paper presents a methodology applied to the mapping and monitoring of surface soil moisture over the Kori Dantiandou region in Niger, using data provided by the ASAR/ENVISAT radar instrument. The study is based on 15 sets of ASAR/ENVISAT C‐band radar data, acquired during the 2004 and 2005 rainy seasons. Simultaneously with radar acquisitions, ground soil moisture measurements were carried out in a large number of test fields. Soil moisture was estimated only for fields with bare soil or low‐density vegetation, using low‐incidence‐angle radar data (IS1 configuration). A mask was developed, using SPOT/HRV data and DTM, for use over areas characterized by high‐density vegetation cover, pools, and areas with high slopes. Soil moisture estimations are based on horizontal‐ and vertical‐polarization radar data. In order to double the temporal frequency of soil moisture estimations, IS2 data were used with IS1 data, with all data normalized to a single incidence angle. A high correlation is observed between in situ measurements and processed radar data. An empirical inversion technique is proposed, to estimate surface soil moisture from dual‐polarization data with a spatial resolution of approximately 1 km. Surface soil moisture maps are presented for all the studied sites, at various dates in 2004 and 2005. Of particular interest, these maps reveal convective precipitation scales associated with strong spatial variations in surface soil moisture.     

基于L演化系统3D树木构建及其光学功能初探

以L系统演化理论为主要算法,将枝、叶和芽作为单位转化子,通过海龟几何变换,最终构建形成三维树木模型,进而可以方便快捷地计算出不同的树高层面上的叶面积指数和光斑,对于植物光和作用的研究有重要意义。结果表明,此模型具有一定的形态逼真性和运用价值,经扩展后可用于植物理论生物学和实验观测生物学的仿真研究。     

Multi-polarization Envisat-ASAR images as a function of leaf area index (LAI) of White Poplar and Desert Date plantations

The relationship between leaf area index (LAI) of plantations and multi-polarization Synthetic Aperture Radar (SAR) data (Envisat-ASAR) was investigated for White Poplar (Populus tomentosa Carr) and Desert Date (Elaeagnus angustifolius) in Heihe district, northwest china. The study showed that, for homogeneous White Poplar plantations, HH and HV polarization data (where H and V represent horizontal and vertical polarizations, respectively, and the first of the two letters refers to the transmission polarization and the second to the received polarization) were sensitive to LAI and the r² (logistic relationship fits) values between HH polarization and LAI, 0.56 and 0.58 on 25 and 28 June images respectively, was much higher than that for the other polarizations of VV, VH and HV. For Desert Date plantations, the heterogeneity of the forests results in a more complex backscattering than that for White Poplar. Incidence angle also plays an important role in SAR backscattering, so a suitable SAR mode should be chosen to avoid scattering saturation when the LAI and incidence angle exceed certain values. The logistic polarization ratios of HH/HV and VV/VH showed varying correlation with LAI over White Poplar plantations, probably due to incidence angle.