基于RGB颜色空间的早稻氮素营养监测研究

针对双季稻区水稻过量施肥带来环境污染和成本提高问题,设计不同品种氮肥梯度大田试验,应用数码相机获取早稻冠层数字图像,研究不同色彩参数及早稻氮素营养指标的时空变化特征,以期确立双季早稻氮素营养预测模型。结果表明:不同品种同一氮肥处理下图像色彩参数差异不大;拔节期数字图像参数对氮素营养指标敏感;模型构建结果显示,图像参数INT与水稻氮素营养指标构建的模型决定系数(R2)最大,模型预测效果最佳,R2分别为0.895 7和0.924 7;进一步采用多元回归分析和BP神经网络分析法进行预测,预测效果均较好。对预测结果进行检验,发现品种对于模型的构建影响不大,以BP神经网络分析法构建的叶片氮浓度(LNC)模型和以INT为敏感色彩参数构建的叶片氮积累量(LNA)回归模型效果最优,而多元回归分析方法则效果不佳。早稻冠层RGB颜色空间敏感参数与氮素营养指标间相关性较好,可以实现氮素营养的无损监测诊断。     

基于高光谱技术的覆膜旱作水稻植株氮含量及籽粒产量估算

为建立利用光谱技术快速诊断覆膜旱作水稻植株氮营养和产量的估算模型,应用高光谱技术分析了长江中下游覆膜旱作区水稻拔节期和抽穗期内5种不同施氮水平(0、60、120、180和240kg/hm~2(N))下植株冠层光谱特征及其与植株氮素含量和产量的关系,并分别构建了植株含氮量和产量的估算模型。结果表明:拔节期和抽穗期内不同供氮水平下冠层光谱的变化规律基本一致,均随着供氮水平的增加,反射率在可见光区降低、在近红外区增大。覆膜旱作水稻植株氮含量与552和890nm 2个敏感波段构成的比值(RVI)和绿色归一化植被指数(GNDVI)的关系最佳。构建的水稻关键生育期植株全氮含量及水稻产量的估算模型预测效果均较好,其中植株全氮含量拟合方程的决定系数为0.730~0.808,采用拔节期的RVI对覆膜旱作水稻进行估产的决定系数达到0.724。本研究构建的模型可以用来估计该地区覆膜旱作水稻的氮素营养状况和作物产量。     

基于LBPHSV+ResNet50融合的水稻冠层氮素营养监测方法

为采用数码相机拍摄的水稻冠层图像来估测作物的氮素含量。以自然环境下获得的水稻冠层图像为研究对象,提出一种基于图像纹理色彩特征(LBPHSV)和ResNet50网络融合算法的氮素含量预测方法。LBPHSV+ResNet50融合算法是通过运用LBP算子和HSV颜色空间矩阵提取图像特征参数,将提取到的融合特征集作为ResNet50模型输入以加强对作物氮素营养的表征,并将预测结果与常用的多元线性回归、随机森林(RF)、支持向量回归模型、多层感知机、卷积神经网络、长短记忆网络(LSTM)及组合模型预测结果进行对比分析。结果显示:相比于浅层机器学习模型,深度学习算法能显著提高预测模型的准确率;LBPHSV+ResNet50融合模型的预测能力和泛化能力达到最优,R²和 RMSE分别为 0.97、0.02。相比于RF、LBP+LSTM、ResNet50,新模型的R²分别提升了16.36%、9.72%、16.55%和1.13%,RMSE 分别下降了 0.35、0.46、0.05和 0.002。因此,LBPHSV+ResNet50融合模型在预测水稻氮素含量时可提供令人满意的性能,能够满足对水稻氮素营养无损精准监测的农业需求。     

Canopy reflectance response to plant nitrogen accumulation in rice

Tools to quantify the nitrogen (N) status of a rice canopy during inter-nodal elongation (IE) would be valuable for mid-season N management because N accounts for the largest input cost. The objective of this paper was to study canopy reflectance as a potential tool for assessing the mid-season status of N in a rice crop. Three field plot experiments were conducted in 2002 and 2003 on cultivars Wells and Cocodrie to study the canopy reflectance response of rice to plant N accumulation (PNA) during IE and to identify the wavelengths and vegetation indices that are good indicators of PNA. Each experiment included six pre-flood N treatments of 0, 33.6, 67.2, 100.8, 133.4 and 168 kg N ha⁻¹. Rice canopy reflectance, biomass, tissue N concentration and PNA were measured weekly during IE. The wavelengths most strongly correlated to PNA at the beginning of IE were 937 and 718 nm. Several vegetation indices were examined to determine which were strongly correlated (>0.7) with PNA at the beginning of IE. Multiple linear regression models of PNA on selected vegetation indices explained 53–85% of the variation in PNA during the first week of IE. This study identifies the best combinations of vegetation indices for estimating PNA in rice.     

基于数码图像的甜菜氮素近地遥感监测模型

为定量研究利用数码图像进行甜菜冠层叶片氮含量(Leaf nitrogen content,LNC)时空变化监测的适宜性及准确性,以2014年内蒙古赤峰市松山区太平地镇田间试验为基础,在甜菜各生长阶段采集甜菜冠层数码图像,利用数字图像处理技术对图像进行分割并提取红光值(R)、绿光值(G)和蓝光值(B)。分析R/B、G/B等9个颜色参数与不同生育期冠层LNC的相关性,并研究冠层LNC随施氮量的变化规律,探寻适宜于甜菜氮素营养监测的关键生育时期及最佳颜色参数。分别利用支持向量机(Support vector machine,SVM)和BP人工神经网络(BackPropagation artificial neural network,BP-ANN)建立甜菜冠层LNC预测模型。研究结果表明,BP-ANN预测模型具有较高且较稳定的预测精度,其验证集的决定系数R~2和均方根误差RMSE分别为0.74和2.35,与SVM模型相比,BP-ANN模型的决定系数R~2提高了12.12%,均方根误差RMSE降低了8.09%。     

Monitoring nitrogen,phosphorus and sulphur in hybrid rice (<Emphasis Type="Italic">Oryza sativa</Emphasis> L.) using hyperspectral remote sensing

Till date, the remote sensing research on crop nutrient monitoring has focused mainly on biomass and nitrogen (N) estimation and only a few attempts had been made to characterize and monitor macronutrients other than N. Field experiments were undertaken to study the remote detection of macronutrient status of rice using hyperspectral remote sensing. The variability in soil available N, phosphorus (P) and sulphur (S) and their content in plants were created using artificial fertility gradient design. The leaf and canopy hyperspectral reflectance was captured from variable macronutrient status vegetation. Linear correlation analysis between the spectral reflectance and plant nutrient status revealed significantly (p < 0.05) higher correlation coefficient at 670, 700, 730, 1090, 1260, 1460 nm for the nutrient under study. Published and proposed vegetation indices (VIs) were tested for canopy N, P and S prediction. The results of the investigation revealed that, published VIs (NDVI hyper and NDVI broadbands) could retrieve canopy N with higher accuracy, but not P and S. The predictability of the visible and short wave infrared based VI NRI₁₅₁₀ ((R₁₅₁₀ ? R₆₆₀)/(R₁₅₁₀ + R₆₆₀)) was the highest (r = 0.81, p < 0.01) for predicting N. Based on the outcomes of linear correlation analysis new VIs were proposed for remote detection of P and S. Proposed VI P_670_1260 ((R₁₂₆₀ ? R₆₇₀)/(R₁₂₆₀ + R₆₇₀)) retrieved canopy P status with higher prediction accuracy (r = 0.67, p < 0.01), whereas significantly higher canopy S prediction (r = 0.58, p < 0.01) was obtained using VI S_670_1090 ((R₁₀₉₀ ? R₆₇₀)/(R₁₀₉₀ + R₆₇₀)). The proposed spectral algorithms could be used for real time and site-specific N, P and S management in rice. Nutrient specific wavelengths, identified in the present investigation, could be used for developing relatively low-cost sensors of hand-held instruments to monitor N, P and S status of rice plant.     

Comparison of passive and active canopy sensors for the estimation of vine biomass production

Recent advances in optical designs and electronic circuits have allowed the transition from passive to active proximal sensors. Instead of relying on the reflectance of natural sunlight, the active sensors measure the reflectance of modulated light from the crop and so they can operate under all lighting conditions. This study compared the potential of active and passive canopy sensors for predicting biomass production in 25–32 randomly selected positions of a Merlot vineyard. Both sensors provided estimates of the normalized difference vegetation index (NDVI) from a nadir view of the canopy at veraison that were good predictors of pruning weight. Although the red NDVI of the passive sensors explained more of the variation in biomass (R ² = 0.82), its relationship to pruning weight was nonlinear and was best described by a quadratic regression (NDVI = 0.55 + 0.50 wt−0.21 wt²). The theoretically greater linearity of the amber NDVI-biomass relationship could not be verified under conditions of high biomass. The linear correlation to stable isotope content in leaves (¹³C and ¹⁵N) provided evidence that canopy reflectance detected plant stresses as a result of water shortage and limited fertilizer N uptake. Thus, the canopy reflectance data provided by these mobile sensors can be used to improve site-specific management practices of vineyards.     

A comparison of three methods for estimating leaf area index of paddy rice from optimal hyperspectral bands

This paper follows previous research that identified 15 hyperspectral wavebands that were suitable to estimate paddy rice leaf area index (LAI). The objectives of the study were to: (1) test the efficiency of the wavebands selected in the previous study, (2) to evaluate the potential of least squares support vector machines (LS-SVM) to estimate paddy rice LAI from canopy hyperspectral reflectance and (3) to compare multiple linear regression-MLR, partial least squares-PLS regression and LS-SVM to determine paddy rice LAI using the selected wavebands. In the study, measurements of hyperspectral reflectance (350–2500 nm) and corresponding LAI were made for a paddy rice canopy throughout the growing seasons. On the basis of the wavebands selected previously, models based on MLR, PLS and LS-SVM to estimate rice LAI were compared using the data from 123 observations, which were split randomly for model calibration (2/3) and validation (1/3). Root mean square errors (RMSEs) and the correlation coefficients (r) between measured and predicted LAI values from model calibration and validation were calculated to evaluate the quality of the models. The results showed that the LS-SVM model using the 15 selected wavebands produced more accurate estimates of paddy rice LAI than the PLS and MLR models. We concluded that the LS-SVM approach may provide a useful exploratory and predictive tool for estimating paddy rice LAI when applied to reflectance data using the 15 selected wavebands.     

东北水稻叶片SPAD遥感光谱估算模型

为通过构建高精度SPAD遥感估算模型,实现对水稻叶片叶绿素含量进行实时无损的监测,以东北地区多时期不同施氮水平下水稻叶片光谱反射率为研究对象,采用回归模型与BP神经网络算法构建不同输入量的SPAD高光谱估算模型,通过模型精度评价指标决定系数R~2、均方根误差RMSE,确定最优输入量和最优模型。结果表明:1)不同品种水稻成熟时期不同导致在孕穗期和抽穗期之间光谱反射率出现差异;2)回归模型中以DVI(D755,D930)为变量建立多项式模型估算精度最高;3)与回归模型相比,不同波长处单波段反射率作为输入量的BP神经网络模型估算精度显著提高,R~2为0.98。BP神经网络模型在隐藏节点数为7时估算精度达到稳定,在可见光和近红外处经过不同波段反射率作为输入量的尝试说明神经网络模型较为稳定,可以用来反演叶绿素相对含量。     

Combining leaf fluorescence and active canopy reflectance sensing technologies to diagnose maize nitrogen status across growth stages

Rapid methods allowing for non-destructive crop monitoring are imperative for accurate in-season nitrogen (N) status assessment and precision N management. The objectives of this paper were to (1) compare the performance of a leaf fluorescence sensor Dualex 4 and an active canopy reflectance sensor Crop Circle ACS-430 for estimating maize (Zea mays L.) N status indicators across growth stages; (2) evaluate the potential of N status prediction across growth stages using the reflectance parameters acquired from the canopy sensor at an early growth stage; and, (3) investigate the prospect of combining the active canopy sensor and leaf fluorescence sensor data to estimate N nutrition index (NNI) indirectly using a general model across growth stages. The results indicated that data from both sensors were closely related to NNI across stages. However, using the direct NNI estimation method, among the tested indices, only the N balance index (NBI) could diagnose N status satisfactorily, based on the Kappa statistics. The effect of growth stages on proximal sensing was reduced by incorporating the information of days after sowing. It was found that the leaf fluorescence sensor performed relatively better in estimating plant N concentration whereas the canopy reflectance sensor performed better in aboveground biomass estimation. Their combination significantly improved the reliability of N diagnosis, including NNI prediction. In addition, the study confirmed that N status can be assessed by predicting aboveground biomass at the later stages using the canopy reflectance measurements at an early stage. Furthermore, the integrated NBI was verified to be a more robust and sensitive N status indicator than the chlorophyll concentration index. It is concluded that combining active canopy sensor data, of an early growth stage (e.g. V8), with leaf fluorescence sensor data, modified using days after sowing, can improve the accuracy of corn N status diagnosis across growth stages.

     

Leaf area index based nitrogen diagnosis in irrigated lowland rice

Leaf area index(LAI)is used for crop growth monitoring in agronomic research,and is promising to diagnose the nitrogen(N)status of crops.This study was conducted to develop appropriate LAI-based N diagnostic models in irrigated lowland rice.Four field experiments were carried out in Jiangsu Province of East China from 2009 to 2014.Different N application rates and plant densities were used to generate contrasting conditions of N availability or population densities in rice.LAI was determined by LI-3000,and estimated indirectly by LAI-2000 during vegetative growth period.Group and individual plant characters(e.g.,tiller number(TN)and plant height(H))were investigated simultaneously.Two N indicators of plant N accumulation(NA)and N nutrition index(NNI)were measured as well.A calibration equation(LAI=1.7787LAI_(2000)–0.8816,R~2=0.870~(**))was developed for LAI-2000.The linear regression analysis showed a significant relationship between NA and actual LAI(R~2=0.863~(**)).For the NNI,the relative LAI(R~2=0.808~(**))was a relatively unbiased variable in the regression than the LAI(R~2=0.33~(**)).The results were used to formulate two LAI-based N diagnostic models for irrigated lowland rice(NA=29.778LAI–5.9397;NNI=0.7705RLAI+0.2764).Finally,a simple LAI deterministic model was developed to estimate the actual LAI using the characters of TN and H(LAI=–0.3375(TH×H×0.01)~2+3.665(TH×H×0.01)–1.8249,R~2=0.875~(**)).With these models,the N status of rice can be diagnosed conveniently in the field.     

基于冠层高光谱信息的苹果树花量估测

【目的】研究基于盛花期冠层高光谱数据的苹果花量估测技术,为植株花果管理和生产力预测技术的建立奠定基础。【方法】以5年生M₉无性系砧木‘米奇嘎啦’苹果（Malus pumila‘Mitch Gala’）、树形为高纺锤形的植株为试材,在盛花期采集植株冠层可见-近红外高光谱图像,人工统计供试植株花量,比对分析基于原始光谱反射率（original reflectance spectra,OS）与Savitzky-Golay平滑法（savitzky-golay smoothing,SG）、正态变量标准化（standardization of normal variables,SNV）、标准化（Normalize）、一阶求导（first derivation,l^st Der）、二阶求导（second derivation,2^nd Der）共5种预处理的高光谱数据的偏最小二乘法（partial least squares method,PLS）模型,以及基于载荷系数法（x-loading weight,x-LW）提取的特征波长的PLS模型、人工神经网络（the back-propagation neural network,BPNN）、最小二乘支持向量机（the least squares support vector machines,LS-SVM）等模型对单株单位面积花量实时估测精度的影响。【结果】苹果树单株花量与单株单位面积花量具有较高的相关系数,表明采用冠层单位面积花量替代单株总花量进行树体花量估测可行。单株单位面积花量与植株冠层光谱反射率在紫外-可见光波长（308—700 nm）呈极显著正相关,在近红外波长（750—1 000 nm）相关性不显著。基于全波长,以Normalize预处理光谱建立的PLS模型对单株单位面积花量的预测效果最好,校正集决定系数（R_c²）和预测集决定系数（R_p²）分别为0.794和0.804,校正集均方根误差（RMSEC）和预测集均方根误差（RMSEP）分别为0.084、0.062,预测相对误差（RE%）为3.940。基于特征波长的BPNN模型稳定性差,而LS-SVM模型的建模效果较好,R_c²和R_p²分别为0.826和0.804,RMSEC和RMSEP分别为0.077、0.064,RE%为12.160。【结论】基于Normalize预处理的PLS模型对高纺锤形苹果树冠层单位面积花量的预测效果最优,同时,本研究利用高光谱成像仪获取的数据,经过分析处理对提取特征信息进行简化,可为多光谱遥感数据的应用提供依据。     

基于近红外光谱技术的大米掺伪定量判别

针对现在市场上常见的两种大米掺伪现象,利用近红外光谱技术结合化学计量学方法分别建立了大米中掺入低档米和掺入矿物油的定量分析模型。制配不同掺伪比例的大米样品,采集其近红外光谱,并选用标准正态变量变换、最大最小归一化、平滑和一阶导数4种方法对原始光谱进行预处理,分别结合偏最小二乘法建立PLS定量分析模型。通过对比建模结果选出的最优预处理方法是最大最小归一化,建立的掺低档米模型的校正集和预测集相关系数分别为0.9698和0.9845,均方根误差分别为8.66和6.46;掺矿物油米模型的校正集和预测集相关系数分别为0.9739和0.9888,均方根误差分别为0.106和0.0698。模型的预测精度和稳定性均很好,实现了对两种掺伪大米快速、准确的定量判别,为大米的品质监控提供了一种新的方法思路。     

矮化中间砧短枝富士苹果高纺锤树形冠层结构与光能截获的三维模拟

【目的】建立苹果冠层结构的三维虚拟植物模型,为精确、量化评价果树冠层空间结构及光截获提供方法指导,为果树树形选择及优化提供理论依据。【方法】以14年生矮化中间砧高纺锤形富士（西府海棠/M26/礼泉短富）为试材,以田间树体数字化测定为基础,获取枝叶特定形态参数,利用计算机模拟重建三维虚拟植物。借助虚拟植物模型进行虚拟试验,定量研究冠层结构、光截获和果实的空间分布。【结果】确定了树体枝叶间异速生长关系：枝（梢）长度分别与枝（梢）叶片数量和枝（梢）的总叶面积、叶片长度分别与叶柄长度和叶片宽度、叶片长度的平方与叶面积间均呈显著线性关系。结合三维数字化技术与枝叶间异速生长关系,构建了苹果冠层三维虚拟植物模型,结果表明：叶片数量及叶片面积模拟值与实测值间决定系数分别为0.92和0.95,均方根误差分别为1.18和31.5 cm²,相对误差分别为7.15%和5.86%。模型精度可满足冠层结构与光截获评价要求;模型可量化模拟各类枝梢及整体冠层叶面积、体积、光照射叶片面积、冠层或枝（梢）叶片被光线照射到的照射叶面积与总叶面积比值（STAR）、郁闭度、叶面积密度相对方差（ξ）及STAR值的日动态变化。矮化中间砧高纺锤树形枝（梢）主要分布于冠层高度0.5-2.5 m和距树干水平距离20-80 cm空间范围内,占总枝梢叶面积比例74.88%。整体冠层体积、郁闭度分比为4.47 m³、44.62%;营养短枝（梢）、营养长枝（梢）及果台副梢分别占树体体积的69.73%、43.50%和41.26%,三者郁闭度分别为60.77%、54.12%和83.15%,平均STAR值分别为0.10、0.23和0.13;各类枝（梢）STAR值空间分布规律明显,随冠层高度及距树干水平距离增大而逐渐增加。果实主要分布于冠层高度0.5-2.0 m和距树干水平距离20-60 cm空间区域;单位面积产量4.1×10⁴ kg/667 m²;冠层适宜的营养短枝（梢）、营养长枝（梢）、果台副梢叶面积比例为69.70%-73.13%﹕11.25%-15.18%﹕11.16%-14.27%,适宜STAR值分别为0.08、0.14和0.11。整体冠层、营养短枝（梢）和果台副梢的STAR值日变化曲线近似于双峰曲线,营养长枝（梢）STAR值日变化为单峰曲线;各类枝（梢）STAR值皆与果台副梢数量呈显著负相关关系,与果实可溶性固形物、单果重及横径呈显著正相关关系。【结论】三维虚拟植物模型可用于果树冠层结构及光截获的精准量化评价,STAR值可量化评价冠层光截获效率。     

Improving in-season estimation of rice yield potential and responsiveness to topdressing nitrogen application with Crop Circle active crop canopy sensor

In-season site-specific nitrogen (N) management is a promising strategy to improve crop N use efficiency and reduce risks of environmental contamination. To successfully implement such precision management strategies, it is important to accurately estimate yield potential without additional topdressing N application (YP₀) as well as precisely assess the responsiveness to additional N application (RI) during the growing season. Previous research has mainly used normalized difference vegetation index (NDVI) or ratio vegetation index (RVI) obtained from GreenSeeker active crop canopy sensor with two fixed bands in red and near-infrared (NIR) spectrums to estimate these two parameters. The development of three-band Crop Circle active sensor provides a potential to improve in-season estimation of YP₀ and RI. The objectives of this study were twofold: (1) identify important vegetation indices obtained from Crop Circle ACS-470 sensor for estimating rice YP₀ and RI; and (2) evaluate their potential improvements over GreenSeeker NDVI and RVI. Four site-years of field N rate experiments were conducted in 2012 and 2013 at the Jiansanjiang Experiment Station of China Agricultural University located in Northeast China. The GreenSeeker and Crop Circle ACS-470 active canopy sensor with green, red edge, and NIR bands were used to collect rice canopy reflectance data at different key growth stages. The results indicated that both the GreenSeeker (best R² = 0.66 and 0.70, respectively) and Crop Circle (best R² = 0.71 and 0.77, respectively) sensors worked well for estimating YP₀ and RI at the stem elongation stage. At the booting stage, Crop Circle red edge optimized soil adjusted vegetation index (REOSAVI, R² = 0.82) and green ratio vegetation index (R² = 0.73) explained 26 and 22 % more variability in YP₀ and RI, respectively, than GreenSeeker NDVI or RVI. At the heading stage, the GreenSeeker sensor indices became saturated and consequently could not be used for YP₀ or RI estimation, while Crop Circle REOSAVI and normalized green index could still explain more than 70 % of YP₀ and RI variability. It is concluded that both sensors performed similarly at the stem elongation stage, but significantly better results were obtained by the Crop Circle sensor at the booting and heading stages. Furthermore, the results revealed that Crop Circle green band-based vegetation indices performed well for RI estimation while the red edge-based vegetation indices were the best for estimating YP₀ at later growth stages.     

氮磷互作对水稻冠层光谱的影响及其PNN识别

【目的】氮、磷均为作物必需的大量营养元素,其丰缺诊断直接关系到合理科学施肥,进而影响产量、效益以及环境。本文旨在研究准确、快捷、无损地区分水稻缺氮和缺磷信息的光谱识别方法,从而指导田间施肥决策,精确作物管理、节约种植成本并控制农田面源污染。【方法】基于水稻6个氮素及两个磷素营养水平交互下的盆栽试验,分别在分蘖、拔节和抽穗期测定水稻冠层的可见近红外反射光谱（350-1 330 nm）及植株全氮（TN）和全磷（TP）含量等数据,分析氮磷互作对水稻植株体内TN和TP含量以及冠层反射光谱的影响,并运用概率神经网络(PNN)分别对不同生育时期的冠层光谱进行氮水平、磷水平、氮磷交互水平和缺素水平4个尺度下的分类识别。为避免光谱测量时仪器误差和光照、风力、温度、水分等环境条件所造成光谱数据批次间的差异,PNN分类识别前对光谱数据进行标准化处理,并将其中2/3作为训练集,另外1/3作为测试集。【结果】植株全氮含量受氮肥、磷肥和氮磷交互作用的影响显著;植株全磷含量则主要受磷肥和氮肥水平的双重影响,但不存在氮磷交互作用。水稻冠层光谱对氮肥的响应规律不受磷肥水平的影响,缺氮使可见光区反射率升高,近红外区反射率下降。缺磷使近红外区反射率下降,但可见光区的响应则受氮肥水平的影响,施氮处理呈上升趋势,氮胁迫处理则呈现分蘖期下降、拔节期上升、抽穗期下降的趋势。利用冠层光谱PNN模型可以对各个生育时期氮水平、磷水平、氮磷交互水平和缺素水平等不同施肥尺度进行识别,拔节期分类精度最高,抽穗期分类精度相对最低。4种分类尺度下PNN模型对磷素水平的分类精度最高,分蘖期和拔节期分别为83%和94%;其次是缺素水平,分别为78%和88%;对氮素水平以及氮磷交互水平等有较多个分类输出的识别精度较低,为61%-75%。值得一提的是,PNN模型对水稻施肥关键生育时期分蘖期和拔节期水稻植株缺氮缺磷、缺氮不缺磷、缺磷不缺氮、不缺氮不缺磷等4种缺素水平的分类中,所有只缺氮处理没有被预测为只缺磷处理,所有只缺磷处理也没有被误判为只缺氮处理,表明冠层光谱PNN模型能有效区分开氮磷胁迫。【结论】水稻的冠层光谱受到氮、磷水平的共同影响,利用水稻冠层光谱建立的PNN模型不仅能分别辨识各氮素、磷素施肥水平,并且能有效地区分开水稻缺磷和缺氮处理,避免混淆,对有目的性的指导施肥具有重要的意义和价值,可避免不恰当的施肥策略造成的环境、产量和经济损失。     

Using optimal combination method and in situ hyperspectral measurements to estimate leaf nitrogen concentration in barley

Leaf nitrogen concentration (LNC), a good indicator of nitrogen (N) status in crops, is of special significance to diagnose nutrient stress and guide N fertilization in fields. Due to non-destructive and quick detectability, hyperspectral remote sensing plays a unique role in detecting LNC in crops. Barley, especially malting barley, is very demanding for N nutrition and requires timely monitoring and accurate estimation of N concentration in barley leaves. Hyperspectral techniques can help make effective diagnosis and facilitate dynamic regulation of plant N status. In this study, canopy reflectance spectra (between 350 and 1 050 nm) from 38 typical barley fields were measured as well as the corresponding LNC in Hailar Nongken, China’s Inner Mongolia Autonomous Region in July, 2010. Existing spectral indices that are considered to be good indicators for assessing N status in crops were selected to estimate LNC in barley. In addition, the optimal combination (OC) method was tested to extract the sensitive indices and first-order spectral derivative wavebands that are responsible for variation of leaf N in barley, and expected to develop some combination models for improving the accuracy of LNC estimates. The results showed that most of the selected indices (such as NPCI, PRI and DCNI) could adequately describe the dynamic changes of LNC in barley. The combined models based on OC performed better in comparison with the individual models using either spectral indices or first-order derivatives and the other methods (such as PCA). A combined model that integrated the first-order derivatives from five wavebands with OC performed well with R ² of 0.82 and RMSE of 0.50 for LNC in barley. This good correlation with ground measurements indicates that hyperspectral reflectance and the OC method have good potential for assessing N status in barley.     

不同水氮耦合条件下棉花冠层NDVI分析与产量估算

为研究冠层归一化差值植被指数(Normalized difference vegetation index,NDVI)在棉花重要生育时期估算棉花产量的可行性,使用GreenSeeker分别对不同灌水施肥条件下棉花光谱反射率NDVI值进行测定优化,建立NDVI值与产量关系数学模型,并对模型精度进行验证。结果显示:不同水氮组合随着生育期的推移棉花冠层NDVI值变化趋势基本一致,都呈"低-高-低"的变化规律;选取在棉花出苗后80、105和140d冠层NDVI值分别与产量进行相关性分析,得出冠层NDVI值与产量具有明显的正相关关系,相关系数分别为R2=0.376 0,0.093 4,0.363 9。利用独立的试验数据对相关性最高的水氮组合棉花出苗后80d的产量模型进行模型验证,其相关系数R2=0.712 6,均方根误差(Root mean square error,RMSE)561.04kg/hm2。因此,棉花出苗后80d的冠层NDVI值可以估测棉花产量。     

基于反射光谱全波段与双波段甘蔗叶片叶绿素预测模型比较

以甘蔗品种新台糖22号(ROC22)叶片为研究对象,针对全波段和双敏感波段处的反射率分别建立甘蔗叶片叶绿素含量的预测模型,对比各模型的精度。全波段方面,以可见-近红外光谱反射率为输入量,提取出前5个主成分后,分别采用多元线性回归(MLR)与BP神经网络(BPNN)方法建立全波段模型M1与M2;敏感波段方面,选择731和785nm这2个敏感波段及由二者计算出的植被指数为输入量,建立一元线性回归(SLR)模型M3、MLR模型M4以及BPNN模型M5。研究结果表明:M1与M2的预测值与实测值间的决定系数R~2分别为0.792 4和0.892 9;M3、M4、M5的R~2分别为0.821 2、0.840 1和0.848 2;BPNN模型精度高于线性回归模型;虽M5的精度稍低于M2的精度,但M5只包含2个敏感波段信息,具有更高的工程应用价值。     

Tectona grandis canopy cover predicted by remote sensing

The phytosanitary status of Tectona grandis plantations are monitored conventionally with periodic data collection in the field, which is often costly and has low efficiency. The objective of this research was to develop a methodology to predict the canopy cover of T. grandis plantations using multispectral images of the Sentinel-2 (S2) satellite and photographic imagery. The study was carried out in a T. grandis plantation of seminal origin, in Cáceres, Mato Grosso state, Brazil. Hemispherical photographic (HP) images of the plant canopy were obtained with a digital camera coupled to a “fisheye” lens fixed at 1.3 m high at two dates in the rainy and the dry season. Cloudless and no shadow images of the S2 satellite bands were concurrently obtained with the field images. Multivariate permutative analysis of variance (PERMANOVA) and partial least squares regression (PLSR) were used to predict canopy cover percentage. The accuracy of the predicted T. grandis canopy cover (%) by the PLSR model approach was 77.8?±?0.09%. The results indicate that a PLS model calibrated with 28 HP sample images can accurately estimate the percentage canopy cover for a continuous area of T. grandis plantations and facilitate mapping of canopy heterogeneity to monitor threats of diseases, mortality, fires, pests and other disturbances.