基于灰度关联-岭回归的荒漠土壤有机质含量高光谱估算

为改善高光谱技术对荒漠土壤有机质的估测效果,该文采集了以色列Seder Boker地区的荒漠土壤,经预处理、理化分析后将土样分为砂质土和黏壤土2类,再通过光谱采集、处理得到6种光谱指标:反射率(reflectivity,REF)、倒数之对数变换(inverse-log reflectance,LR)、去包络线处理(continuum removal,CR)、标准正态变量变换(standard normal variable reflectance,SNV)、一阶微分变换(first order differential reflectance,FDR)和二阶微分变换(second order differential reflectance,SDR)。通过灰度关联(gray correlation,GC)法确定SNV、FDR、SDR为敏感光谱指标,采用偏最小二乘回归(partial least squares regression,PLSR)法和岭回归(ridge regression,RR)法,构建基于敏感光谱指标的土壤有机质高光谱反演模型,并对模型精度进行比较。结果表明:砂质土有机质含量的反演效果要优于黏壤土;基于SNV指标建立的模型决定系数R~2和相对分析误差RPD均为最高、均方根误差RMSE最低,所以SNV是土壤有机质的最佳光谱反演指标;对SNV-PLSR模型和SNV-RR模型综合比较得出,SNV-RR模型仅用全谱4%左右的波段建模,实现了更为理想的反演效果:其中,对砂质土有机质的预测能力极强(R_p~2为0.866,RMSE为0.610 g/kg、RPD为2.72),对黏壤土有机质的预测能力很好(Rp2为0.863,RMSE为0.898 g/kg、RPD为2.37)。荒漠土壤有机质GC-SNV-RR反演模型的建立为高光谱模型的优化、土壤有机质的快速测定提供了一种新的途径。     

Ground-level hyperspectral imagery for detecting weeds in wheat fields

Site-specific weed management can allow more efficient weed control from both an environmental and an economic perspective. Spectral differences between plant species may lead to the ability to separate wheat from weeds. The study used ground-level image spectroscopy data, with high spectral and spatial resolutions, for detecting annual grasses and broadleaf weeds in wheat fields. The image pixels were used to cross-validate partial least squares discriminant analysis classification models. The best model was chosen by comparing the cross-validation confusion matrices in terms of their variances and Cohen’s Kappa values. This best model used four classes: broadleaf, grass weeds, soil and wheat and resulted in Kappa of 0.79 and total accuracy of 85 %. Each of the classes contains both sunlit and shaded data. The variable importance in projection method was applied in order to locate the most important spectral regions for each of the classes. It was found that the red-edge is the most important region for the vegetation classes. Ground truth pixels were randomly selected and their confusion matrix resulted in a Kappa of 0.63 and total accuracy of 72 %. The results obtained were reasonable although the model used wheat and weeds from different growth stages, acquisition dates and fields. It was concluded that high spectral and spatial resolutions can provide separation between wheat and weeds based on their spectral data. The results show feasibility for up-scaling the spectral methods to air or spaceborne sensors as well as developing ground-level application.     

基于FOD和SVMDA-RF的土壤有机质含量高光谱预测

为探讨分数阶微分(FOD)联合支持向量机分类-随机森林模型改善高光谱监测荒漠土壤有机质含量(SOM)的效果,对以色列Sde Boker荒漠地区采集的砂质土(SS)和黏壤土(CLS)样品进行理化分析和室内光谱测定,依据光谱的平均反射率建立支持向量机分类模型(SVMAD),并对不同土质高光谱原始反射率分别经0～2阶(间隔0. 2)的分数阶微分处理,构建归一化光谱指数(NDI),分析NDI和SOM之间的二维相关性,并筛选敏感的NDI指数,以此建立不同FOD的随机森林(RF)模型,并以不同土质中的最佳模型进行组合,构建新的SVMDA-RF模型。结果表明:基于光谱平均反射率的SVMDA对土壤质地的分类正确率可达100%;分数阶微分耦合光谱指数具有放大波长间与SOM有关隐含信息的能力,经FOD提升敏感指数的数量在0. 6阶时达到峰值,但黏壤土的敏感指数数量远大于沙质土;由不同FOD敏感指数建立的RF模型中,砂质土在1. 2阶的模型最佳(R_C~2=0. 962,R_P~2=0. 920,RMSEP为0. 435 g/kg,RPD为3. 658),黏壤土在0. 6阶的模型最佳(R_C~2=0. 942,R_P~2=0. 944,RMSEP为0. 554 g/kg,RPD为4. 316);经最佳模型组合后的SVMDA-RF模型,砂质土和黏壤土的模型精度都有所提高,其中R_C~2=0. 980,R_P~2=0. 979,RMSEP为0. 481 g/kg,RPD为7. 004。研究成果可为快速评估荒漠土壤有机质含量提供依据。     

Leaf nitrogen estimation in potato based on spectral data and on simulated bands of the VENμS satellite

Relationships between leaf spectral reflectance at 400–900 nm and nitrogen levels in potato petioles and leaves were studied. Five nitrogen (N) fertilizer treatments were applied to build up levels of nitrogen variation in potato fields in Israel in spring 2006 and 2007. Reflectance of leaves was measured in the field over a spectral range of 400–900 nm. The leaves were sampled and analyzed for petiole NO₃–N and leaf percentage N (leaf-%N). Prediction models of leaf nitrogen content were developed based on an optical index named transformed chlorophyll absorption reflectance index (TCARI) and on partial least squares regression (PLSR). Prediction models were also developed based on simulated bands of the future VENμS satellite (Vegetation and Environment monitoring on a New Micro-Satellite). Leaf spectral reflectance correlated better with leaf-%N than with petiole NO₃–N. The TCARI provided strong correlations with leaf-%N, but only at the tuber-bulking stage. The PLSR analysis resulted in a stronger correlation than TCARI with leaf-%N. An R ² of 0.95 (p < 0.01) and overall accuracy of 80.5% (Kappa = 74%) were determined for both vegetative and tuber-bulking periods. The simulated VENμS bands gave a similar correlation with leaf-%N to that of the spectrometer spectra. The satellite has significant potential for spatial analysis of nitrogen levels with inexpensive images that cover large areas every 2 days.     

A spectral index to monitor the head-emergence of wheat in semi-arid conditions

Harvesting wheat (Triticium aestivum L.) for forage or leaving it for grain is the main decision uncertainty growers face in semi-arid regions during mid-season. To facilitate decision-making, a decision support system (DSS) has recently been proposed that requires information about crop water and nutritional status during spike emergence. Though remote sensing has been used to provide site-specific crop status information, a spectral vegetation index is needed to ensure that the information has been acquired during spike emergence. The objective of this study was to propose a spectral index sensitive to spike emergence and validate its suitability across different commercial farm fields by using ground spectral measurements and multispectral satellite imagery. To develop the index, controlled experiments with commonly grown wheat varieties were conducted during the 2004/2005 and 2005/2006 growing season in the agricultural area of the northern Negev desert of Israel. The experiments showed that spike emergence correlated most strongly (r = 0.7, p < 0.05) with spectral changes near the 1.2 μm water absorption feature in contrast to the band at 1.1 μm which appeared to be only weakly correlated. Thus, the spike emergence sensitive band at 1.2 μm has been combined with the insensitive band at 1.1 μm as reference to form the ratio-based normalized heading index (NHI). Experimental data were then used to establish an index threshold that helps separate data acquired before and after spike emergence. The proposed NHI was able to identify spike emergence with a classification accuracy varying between 53 and 83%. Accuracy was influenced by season, and whether narrow or broad spectral bands were used. Validation of the index in commercial farm fields in Israel and the United States showed that the classification accuracy was similar for ground spectral measurements and the advanced land imager (ALI) satellite imagery. These results suggest that the NHI is suited for identifying the onset of heading throughout wheat-growing areas without the need for characterizing seasonal trends.     

Assessment of maize yield and phenology by drone-mounted superspectral camera

Precision Agriculture - The capability of unmanned aerial vehicle (UAV) spectral imagery to assess maize yield under full and deficit irrigation is demonstrated by a Tetracam MiniMCA12 11 bands...     

Tracking Regional Anthropogenic Air Pollution: A Case Study in Israel

Typical mid-winter anthropogenic air pollution episodes are caused when pollutants are trapped in the lower atmospheric boundary layer due to the generation of surface inversion favored by synoptic conditions. We analyzed the optical properties of atmospheric aerosol particles obtained during one such episode using a sun/sky radiometer at two measurement sites: one located in the densely populated and industrialized central part of Israel, and the other in a reference site, about 150 km away. Aerosol optical thickness and volume size distributions showed an increased burden of fine aerosol particles in the central part of Israel. In order to verify the local origin and anthropogenic nature of the effect, the analysis was accompanied by examinations of the synoptic conditions, air mass backward trajectories, and conventional in situ air pollution measurements made by a ground-based sampling station. This case study shows the ability of optical measurements to track urban and industrial atmospheric air pollution expressed by high concentration of fine aerosol particles. In addition, it emphasizes the role of local Israeli air pollution sources and may explain the difference in the properties of long-term aerosol optical observations between the two sites. The advantages of the optical method presented are speed (almost instantaneous), automated measurement, and sensitivity to aerosol particle concentration as well as aerosol size fraction. The drawback is that the optical measurements discussed deal only with aerosol particles and cannot distinguish between different types of pollutant gases.     

Marked subcutaneous mast cell and eosinophilic infiltration associated with the presence of multiple Dirofilaria repens microfilariae in 4 dogs

Dirofilaria repens is a parasitic nematode in the subcutaneous tissue of carnivores, including dogs and cats, transmitted by mosquitoes. Human beings may be accidental hosts. Infection of a dog with D repens was first reported in Palestine in 1934, and 2 additional cases were reported in dogs in Israel to date. This report describes D repens infection in 4 dogs in Israel that presented with subcutaneous masses, which were cytologically characterized by marked mast cell and eosinophil infiltration. In 3 cases, multiple microfilariae were present in the lesions; rare microfilariae were present in the 4^th case. In all 4 dogs, PCR of fine‐needle aspirates from the lesions were positive for D repens. The mast cells observed in all lesions were uniform and highly granulated, and with the presence of the microfilariae, a mast cell tumor was considered less likely. This report suggests that D repens infection‐associated subcutaneous lesions, characterized cytologically by massive mast cell and eosinophil infiltration, should be considered a differential diagnosis for mast cell tumor, especially in geographic locations endemic for this nematode. Notably, all 4 dogs were infected with D repens despite a routine prophylactic doramectin therapy administered every 3 months, probably due to the relatively long time interval between treatments.