不同品种猪肉pH值高光谱检测的模型传递修正算法

针对猪肉pH值高光谱检测模型受品种差异影响存在适用性差的问题,比较了不同算法,提出了一种基于光谱值校正的模型传递算法,以用于品种之间的模型传递。以山黑猪为主品种建立PLS模型,用该模型直接预测零号土猪样本时,预测相关系数仅为0.415,预测均方根误差为0.180 4,预测精度较差。分别用斜率/截距(S/B)算法、模型更新算法以及光谱值校正传递算法对山黑猪模型进行修正或传递并进行了比较。采用S/B法时,山黑猪模型对零号土猪的预测相关系数仍为0.415,预测均方根误差由0.180 4降至0.134 3,下降了25.54%。采用模型更新算法时,把14个零号土猪样本添加到山黑猪校正集,修正后的山黑猪模型对零号土猪样本的预测性能较优,Rp由0.415提高至0.797,提高92.05%,预测均方根误差由0.1804降低为0.1121,下降了37.86%。采用光谱-理化值共生距离法结合DS算法的光谱值校正传递算法时,山黑猪模型对零号土猪样本的预测相关系数由0.415提高至0.837,提高了101.69%,预测均方根误差由0.1804降低至0.0856,下降了52.55%。结果表明,光谱值校正的传递算法能够有效消除品种之间光谱差异,提高了山黑猪模型的适用性,且传递修正效果优于S/B算法和模型更新算法。     

农业仪器的研究与应用现状

该文概述了农业仪器的研究和应用现状,分析了高光谱遥感技术和光谱探测仪器在作物生长中对植株生理生态特征监测的作用。光谱、色谱和质谱等实验室通用分析仪器日新月异地发展,给农产品安全检测和品质分析带来了极大的便利条件,应加大国产仪器的研发力度,促进植保、生命科学和森林测量等仪器设备的产业化发展。      

基于高光谱的冬小麦不同生育时期地上部生物量监测

为实现冬小麦不同生育时期地上部生物量的高光谱监测,2017-2019年分别在河南省鹤壁市、原阳县和温县布置冬小麦氮肥梯度田闻试验,分别于分蘖期、拔节期、抽穗期和灌浆期测定冬小麦地上部生物量及其冠层原位高光谱反射率(400～950 nm),并采用Pearson相关分析明确两者间定量回归关系,再分别利用支持向量机(supp...     

成熟期水稻砷污染胁迫光谱诊断空间模型研究

以分别表征不同生理特征的NDVI等12种高光谱指数为基础, 结合成熟期水稻叶片实测砷含量, 利用相关分析法提取PSNDa(R= -0.89)、DSWI(R= -0.79)、SIPI(R=0.91)作为单级光谱诊断参数, 并以此为因子 组成砷污染胁迫光谱参数诊断空间(PSNDa-DSWI、PSNDa-SIPI、DSWI-SIPI), 其中以表征叶绿素和细胞结构的PSNDa-SIPI组合空间的诊断效果最佳。利用主成分分析综合各光谱参数信息, 从中提取出主成分F1、F2, 其累积贡献率达88%, 由此建立砷污染胁迫主成分诊断空间(F1-F2), 不同污染程度对应不同诊断空间, 高度污染: F1<1.95, F2>0.75; 中度污染: 1.95<F1<3.15, 0.75>F2>0.40; 轻度污染: F1>3.15, F2<0.40。由此从不同层面构成一个严密的水稻砷污染胁迫诊断系统, 为系统、全面、有效监测大面积水稻砷污染提供多元化诊断依据。     

Physical characterization,spectral response and remotely sensed mapping of Mediterranean soil surface crusts

Soil surface crusting and sealing are frequent but unfavorable processes in Mediterranean areas. Soil crust and seals form on bare soil subject to high-intensity rainfall, resulting in a hard, impenetrable layer that impedes infiltration and hampers the germination and establishment of plants. The adverse consequences of overland flow and reduced fertility can lead to erosion and ongoing degradation. Therefore, information on the distribution of surface crusts and their physical properties is essential to combat the undesired effects of crust formation in e.g. soil erosion and soil compaction. We studied the occurrence of crusts in a study area in Mediterranean southern France. Our objectives were to compare the physical and hydrological properties of the crusts and underlying soil, to identify the spectral characteristics (400 to 2500 nm) of the crusted and non-crusted soil surfaces using high-resolution field spectra, and to investigate the potential of mapping crust occurrence using airborne, hyperspectral HyMap images. The differences in some physical properties between crusted and non-crusted surfaces are significant while others are only marginal. Crusting markedly reduces the infiltration capacity and crust strength varies between the different soil types. Spectral differences are small, mainly in albedo values (overall reflectance) and in absorption band depth and shape. Albedo differences range from 8 to 40%. Differences in absorption band features in the spectra of crusts and non-crusted surfaces are small. Sixty percent of the crusted soil surfaces showed stronger absorption features in the clay mineral absorption bands at 2200 nm than non-crusted soils. Increased absorption is due to a relative enrichment in fines. Spectral feature fitting and linear spectral unmixing algorithms were applied to airborne HyMap images to evaluate the possibilities of mapping surface crusts. Crusts could be mapped in fallow, agricultural fields, but the spectral response of natural badlands was too fragmented for crust mapping.     

不同生育时期冬小麦FPAR高光谱遥感监测模型研究

通过连续5年定位研究不同氮磷耦合水平下,不同生育时期冬小麦群体FPAR与冠层光谱反射率,建立基于不同植被指数的不同生育时期FPAR分段监测模型。结果表明:随着氮磷水平增加FPAR呈递增趋势,不同品种间存在差异;冬小麦群体FPAR与670、850、960 nm具有较高的相关性,在可见光和近红外波段处均有敏感波段;在拔节期、孕穗期、抽穗期、灌浆期和成熟期FPAR与SAVI、NDVI705、EVI、DVI、RVI均达极显著相关,相关系数r范围为0.818~0.942;在不同生育时期,分别基于SAVI、NDVI705、EVI、RVI、RVI能建立较好的FPAR分段监测模型,决定系数R2分别为0.854、0.888、0.811、0.844、0.911;标准误差SE分别为0.054、0.032、0.044、0.047、0.044;以不同年份独立数据对模型进行验证,田间实测值与模型预测值之间相对误差RE分别为14.1%、17.4%、12.8%、18.8%、10.7%;均方根误差RMSE分别为0.139、0.146、0.136、0.158、0.130。该结果较拔节期至成熟期FPAR统一监测模型监测精度及验证效果均有所改善。因此,在拔节期、孕穗期、抽穗期、灌浆期和成熟期可分别用SAVI、NDVI705、EVI、RVI、RVI预测冬小麦群体FPAR,具有较好的年度间重演性和品种间适用性。不同生育时期FPAR分段监测模型较统一监测模型有较好的监测效果。     

以高光谱植被指数研究草坪色泽

色泽是草坪外观质量重要的评价指标之一,叶绿素密度是影响色泽的重要因素。对警犬高羊茅(Festucaarundinacea Schreb.cv.Watchdog)、解放者草地早熟禾(Poa pratensis L.cv.Liberator)、普通狗牙根(Cynodon dac-tylon(L.)Pars.cv.Common)和普特匍匐翦股颖(Agrostis stolonifera L.cv.Putter)等4种草坪进行叶面喷施氮、铁试验,研究反射光谱特征及植被指数与叶绿素密度的相关性。结果显示:喷施氮肥(尿素6 g/m2)或氮肥 铁(尿素6 g/m2 0.6 L/m2EDTA-Fe)能降低草坪在可见光区光谱反射率,提高叶绿素密度,改善草坪色泽;草坪叶绿素密度与可见光波段反射率呈负相关,其中绿波段反射率R(530.94 nm)与叶绿素密度呈极显著相关;由绿波段及蓝绿波段组合的植被指数与草坪叶绿素密度呈显著相关;逐步回归分析结果显示,应用由DVI(530.94 nm,500.12 nm)和NDVI(561.64 nm,520.53 nm)构建的回归方程能够快速定量估测草坪色泽,其结果与目测评分值相关极显著;利用高光谱植被指数可以准确估测叶绿素密度,并对草坪色泽进行定量评价。     

高光谱遥感在土壤研究中的应用

在分析高光谱遥感发展历史、技术特点的基础上,总结了高光谱遥感技术在土壤研究中的主要应用领域,并列举了国内外代表性研究情况。同时,结合现有研究热点,对未来高光谱遥感在土壤研究中的应用进行了分析和展望。     

玉米和大豆LAI高光谱遥感估算模型研究

以ASD Fieldspec光谱仪实测了不同生长季的大田玉米、大豆的冠层高光谱与作物的叶面积指数LAI。采用单变量线性与非线性拟合和逐步回归分析的方式,建立了玉米、大豆LAI高光谱遥感估算模型,并对模型的估算结果进行了初步分析。分析结果表明,绿光波段反射峰区、红光波段以及近红外区的单波段反射率与作物的LAI有较强的相关性,而其他波段的反射率与作物的LAI的相关性相对较弱;以高光谱的窄波段构造的NDVI和RVI与作物的LAI的相关程度高,回归模型的预测水平高;而以多波段逐步回归方式构造的统计模型的预测效果最好。     

Study of How Salt Types Greatly Shape Soil Reflectance Spectra versus Salt Concentrations

In this study, a controlled experiment was designed to determine the potential impacts that salt types impose on soil reflectance spectra. Base soil from a typical inland river basin in China and 99% pure salts of three representive salt types as well as their combinations in different proportions were used to create composite soil samples. Seven levels of salt concentration were assigned for each saline soil treatment. The results revealed that salt concentration did not affect pattern parameters significantly (P > 0.5), where varying the salt concentrations only led to changes in the confined range that corresponded to salt types. Statistical parameter R_mean was sensitive to salt types but was limited to single-type salts. Therefore, salt types largely determined the overall shapes of soil reflectance spectra rather than salt concentrations. It is therefore necessary to identify salt type before retrival of salt concentration in saline soils by use of remote sensing data.