Combining discriminant analysis and neural networks for corn variety identification

Variety identification is an indispensable tool to assure grain purity and quality. Based on machine vision and pattern recognition, five China corn varieties were identified according to their external features. Images of non-touching corn kernels were acquired using a flat scanner. A total of 17 geometric features, 13 shape and 28 color features were extracted from color images of corn kernels. Two optimal feature sets were generated by stepwise discriminant analysis, and used as inputs to classifiers. A two-stage classifier combining distance discriminant and a back propagation neural network (BPNN) was built for identification. On the first stage, corn kernels were divided into three types: white, yellow and mixed corn by distance discriminant analysis. And then different varieties in the same type were identified by an improved BPNN classifier. The classification accuracies of BAINUO 6, NONGDA 86, NONGDA 108, GAOYOU 115, and NONGDA 4967 were 100, 94, 92, 88 and 100%, respectively.     

Detection of Fusarium damaged kernels in Canada Western Red Spring wheat using visible/near-infrared hyperspectral imaging and principal component analysis

Fusarium damage in wheat reduces the quality and safety of food and feed products. In this study, the use of hyperspectral imaging was investigated to detect fusarium damaged kernels (FDK) in Canadian wheat samples. Eight hundred kernels of Canada Western Red Spring wheat were segregated into three classes of kernels: sound, mildly damaged and severely damaged. Singulated kernels were scanned with a hyperspectral imaging system in the visible-NIR (400-1000 nm) wavelength range. Principal component analysis (PCA) was performed on the images and the distribution of PCA scores within individual kernels measured to develop linear discriminant analysis (LDA) models for predicting the extent of fusarium damage. An LDA model classified the wheat kernels into sound and FDK categories with an overall accuracy of 92% or better. Classification based on six selected wavelengths was comparable to that based on the full-spectrum data.     

Identification of pummelo cultivars by using Vis/NIR spectra and pattern recognition methods

Vis/NIR spectroscopy was used in combination with pattern recognition methods to identify cultivars of pummelo (Citrus grandis (L.) Osbeck). A total of 240 leaf samples, 60 for each of the four cultivars were analyzed by Vis/NIR spectroscopy. Soft independent modeling of class analogy (SIMCA), partial least square discriminant analysis (PLS-DA), back propagation neural network (BPNN) and least squares support vector machine (LS-SVM) were applied to the spectral data. The first 8 principal components extracted by principal component analysis were used as inputs in building the BPNN and the LS-SVM models. The results showed that a 97.92 % of discrimination accuracy was achieved for both the BPNN and the LS-SVM models when used to identify samples of the validation set, indicating that the performance of the two models was acceptable. Comparatively, the results of the PLS-DA and the SIMCA models were unacceptable because they had lower discrimination accuracy. The overall results demonstrated that use of Vis/NIR spectroscopy coupled with the use of BPNN and LS-SVM could achieve an accurate identification of pummelo cultivars.     

基于高光谱的小麦旗叶净光合速率的遥感反演模型的比较研究

植物净光合速率是植物生产的基础,是体现植物生长状况的重要生理指标。本文将小麦旗叶高光谱波段反射率进行一阶导数变换后与净光合速率（Pn）进行相关性分析确定敏感波段,分别采用二次多项式逐步回归（QPSR）、偏最小二乘法（PLSR）、BP神经网络法（BPNN）3种方法构建小麦旗叶的净光合速率反演模型,并对3种模型的预测精度进行比较分析。结果表明：（1）将小麦旗叶的原始光谱进行一阶导数变换后与Pn进行相关性分析确定的敏感谱区集中在750~925 nm之间,确定的6个敏感波段分别是：760、761、767、814、815、889 nm;（2）基于QPSR、PLSR、BPNN3种方法以及敏感波段的反射率一阶导数构建的Pn估测模型预测精度都较高,说明用这3种方法以及敏感波段对Pn的估测是可行的,其中模型估算能力顺序为QPSR > BPNN > PLSR,说明小麦旗叶Pn的最佳高光谱分析模型为小麦叶片750~925 nm反射率一阶导数变化后的QPSR模型。     

Rapid determination of leaf water content for monitoring waterlogging in winter wheat based on hyperspectral parameters

Waterlogging is becoming an obvious constraint on food production due to the frequent occurrence of extremely high-level rainfall events. Leaf water content(LWC) is an important waterlogging indicator, and hyperspectral remote sensing provides a non-destructive, real-time and reliable method to determine LWC. Thus, based on a pot experiment, winter wheat was subjected to different gradients of waterlogging stress at the jointing stage. Leaf hyperspectral data and LWC were collected every 7 days after waterlogging treatment until the winter wheat was mature. Combined with methods such as vegetation index construction, correlation analysis, regression analysis, BP neural network(BPNN), etc., we found that the effect of waterlogging stress on LWC had the characteristics of hysteresis and all waterlogging stress led to the decrease of LWC. LWC decreased faster under severe stress than under slight stress, but the effect of long-term slight stress was greater than that of short-term severe stress. The sensitive spectral bands of LWC were located in the visible(VIS, 400–780 nm) and short-wave infrared(SWIR, 1 400–2 500 nm) regions. The BPNN Model with the original spectrum at 648 nm, the first derivative spectrum at 500 nm, the red edge position(λr), the new vegetation index RVI(437, 466), NDVI(437, 466) and NDVI′(747, 1 956) as independent variables was the best model for inverting the LWC of waterlogging in winter wheat(modeling set: R~2=0.889, RMSE=0.138; validation set: R~2=0.891, RMSE=0.518). These results have important theoretical significance and practical application value for the precise control of waterlogging stress.     

A new approach to aflatoxin detection in chili pepper by machine vision

Aflatoxins are the toxic metabolites of Aspergillus molds, especially by Aspergillus flavus and Aspergillus parasiticus. They have been studied extensively because of being associated with various chronic and acute diseases especially immunosuppression and cancer. Aflatoxin occurrence is influenced by certain environmental conditions such as drought seasons and agronomic practices. Chili pepper may also be contaminated by aflatoxins during harvesting, production and storage. Aflatoxin detection based on chemical methods is fairly accurate. However, they are time consuming, expensive and destructive. We use hyperspectral imaging as an alternative for detection of such contaminants in a rapid and nondestructive manner. In order to classify aflatoxin contaminated chili peppers from uncontaminated ones, a compact machine vision system based on hyperspectral imaging and machine learning is proposed. In this study, both UV and Halogen excitations are used. Energy values of individual spectral bands and also difference images of consecutive spectral bands were utilized as feature vectors. Another set of features were extracted from those features by applying quantization on the histogram of the images. Significant features were selected based on proposed method of hierarchical bottleneck backward elimination (HBBE), Guyon’s SVM-RFE, classical Fisher discrimination power and Principal Component Analysis (PCA). Multi layer perceptrons (MLPs) and linear discriminant analysis (LDA) were used as the classifiers. It was observed that with the proposed features and selection methods, robust and higher classification performance was achieved with fewer numbers of spectral bands enabling the design of simpler machine vision systems.     

基于近红外光谱的纽荷尔脐橙产地识别研究

【目的】中国柑橘产区分布广、生态类型复杂,不同产地纽荷尔脐橙果实品质和市场效应具有较大差异。研究基于近红外光谱技术的柑橘产地识别技术,利于不同柑橘产地果品的识别和鉴伪。【方法】从中国南方17个纽荷尔脐橙主要产地选择代表性成年果园,分别采摘成熟鲜果样品100个。利用SupNIR-1500近红外分析仪采集脐橙果实赤道部、肩部表面以及果汁滤液的近红外反射光谱,光谱波长范围为1 000-2 499 nm。采用主成分分析法对原始光谱数据进行预处理,提取近红外光谱的特征信息以降低数据集维度以及噪声。研究人工神经网络理论,构建由一个输入层、一个具有非线性激励函数的隐藏层和一个输出层组成的典型的3层人工神经网络识别模型。研究由径向基函数作为核函数、以光谱主成分作为输入的支持向量机模型,构建由126个分类器组成的一对一扩展支持向量机模型。研究遗传算法优异的自然选择特性,利用遗传算法从光谱主成分中选择出最优的特征基因子集作为支持向量机的输入,构建遗传算法-支持向量机模型。利用3种模型分别对果汁滤液的近红外反射光谱数据进行分类,从而实现产地识别测试,并根据产地识别精度筛选出最优的产地识别模型。进一步对比该最优识别模型对果实赤道部、肩部反射光谱数据的识别精度,从而确定识别精度最高的光谱数据采集源。【结果】利用所建立的3层人工神经网络模型对纽荷尔脐橙果汁滤液的近红外光谱进行产地识别测试,确定当输入神经元数量为11、隐藏神经元数量为13时,模型对果实产地识别的最佳精度达81.45%。采用一对一扩展方式建立支持向量机产地识别模型,研究确定采用径向基函数作为核函数,当主成分数量为20时,脐橙产地识别精度最高可达86.98%。测试利用遗传算法-支持向量机混合模型进行脐橙产地分类识别,确定当种群数量为200、遗传代数为100、交叉概率0.7、突变概率0.01时,遗传算法选择出最优的基因子集进行产地识别,遗传算法-支持向量机模型的产地识别精度最高可达89.72%,优于人工神经网络分类模型和支持向量机分类模型的产地识别精度。进一步利用遗传算法-支持向量机产地识别模型对果实赤道部及肩部的果面反射光谱进行产地识别测试,得到对应的最高识别精度分别为80.00%和69.00%。【结论】遗传算法-支持向量机模型对果汁反射近红外光谱进行产地识别精度最高,优于人工神经网络模型和支持向量机模型。该模型对果实赤道部反射光谱进行分类的精度次于果汁滤液反射光谱但优于果实肩部反射光谱,因此,可利用赤道部的反射光谱实现非破坏性果实产地分类识别。     

Yield-related agronomic traits evaluation for hybrid wheat and relations of ethylene and polyamines biosynthesis to filling at the mid-grain filling stage

Because of the yield increase of 3.5–15% compared to conventional wheat, hybrid wheat is considered to be one of the main ways to greatly improve the wheat yield in the future. In this study, we performed a principal component analysis(PCA) on two-line hybrids wheat and their parents using the grain weight(GW), the length of spike(LS), the kernel number of spike(KSN), and spike number(SPN) as variables. The results showed that the variables could be classified into three main factors, the weight factor(factor 1), the quantity factor 1(factor 2) and the quantity factor 2(factor 3), which accounted for 37.1, 22.6 and 18.5%, respectively of the total variance in different agronomic variables, suggesting that the GW is an important indicator for evaluating hybrid combinations, and the grain weight of restorer line(RGW) could be used as a reference for parents selection. The hybrid combination with a higher score in factor 1 direction and larger mid-parent heterosis(MPH) of the GW and its parents were used to carry out the analysis of grain filling, 1-aminocylopropane-1-carboxylicacid(ACC) and polyamine synthesis related genes. The results suggested that the GW of superior grain was significantly higher than that of inferior grains in BS1453×JS1(H) and its parents. Both grain types showed a weight of H between BS1453(M) and JS1(R), and a larger MPH, which may be caused by their differences in the active filling stage and the grain filling rate. The ADP-glucose pyrophosphorylase(AGPase), granule bound starch synthase Ⅰ(GBSSI), starch synthase Ⅲ(SSS), and starch branching enzyme-I(SBE-I) expression levels of H were intermediated between M and R, which might be closely related to MPH formation of the GW. Compared with R and H, the GW of M at maturity was the lowest. The expression levels of spermidine synthase 2(Spd2), ornithine decarboxylase(ODC) and S-adenosylmethionine decarboxylase(SAMDC) had significantly positive correlations with the grain filling rate(r=0.77*, 0.51*, 0.59*), suggesting their major roles in the grain filling and heterosis formation. These provide a theoretical basis for improving the GW of photo-thermo-sensitive male sterile lines(PTSMSL) by changing the endogenous polyamine synthesis in commercial applications.     

The vegetation classification in coal mine overburden dump using canopy spectral reflectance

The canopy spectral characteristics of typical plants in the overburden of the Fuxin coal mine dump were measured and analyzed. The reflectance of Leymus chinensis was affected by the soil, with a slight shift from green (550 nm) to the near infrared (NIR) region. Changes in chlorophyll and water absorption were not significant in the red (670 nm) and NIR bands, respectively. The reflectance curve trend for Artemisia lavandulaefolia was similar to those of Sophora japonica and Ulmus pumila, while the reflectance of S. japonica and U. pumila fluctuated in the NIR region (760-1200 nm), especially with greater water absorption around 930 and 1120 nm. In contrast, the reflectance of A. lavandulaefolia fluctuated slightly around 930 nm and a significant peak appeared at 1127 nm. In addition, the spectral reflectance of S. japonica was lower than for the other species in the visible band (400-700 nm). However, it was higher than for L. chinensis in the NIR region (780-1200 nm). Three classifiers, the self-organizing map (SOM), learning-vector quantization (LVQ), and a probabilistic neural network (PNN), were used to classify the vegetation and the results of all classifiers were compared based on total spectral reflectance data from 400 to 1200 nm. The PNN was the best classifier in terms of training and testing accuracy. The first difference reflectance was calculated, and the red edge parameter was able to classify the herbs (L. chinensis and A. lavandulaefolia) and the arbores (S. japonica and U. pumila) with an accuracy of 77 and 84%, respectively, although it did not perform as well for detail species. A mixing parameter matrix was built based on the sensitive wavelengths (550, 674, 810, 935, and 1125 nm), the vegetation indices (SAVI and NDGI), and the water absorption slope. High classification accuracy was obtained by applying the mixing parameter matrix. This method could be used for revegetation monitoring and in decision making.     

A new algorithm for automatic Rumex obtusifolius detection in digital images using colour and texture features and the influence of image resolution

In Gebhardt et al. (2006) an object-oriented image classification algorithm was introduced for detecting Rumex obtusifolius (RUMOB) and other weeds in mixed grassland swards, based on shape, colour and texture features. This paper describes a new algorithm that improves classification accuracy. The leaves of the typical grassland weeds (RUMOB, Taraxacum officinale, Plantago major) and other homogeneous regions were segmented automatically in digital colour images using local homogeneity and morphological operations. Additional texture and colour features were identified that contribute to the differentiation between grassland weeds using a stepwise discriminant analysis. Maximum-likelihood classification was performed on the variables retained after discriminant analysis. Classification accuracy was improved by up to 83% and Rumex detection rates of 93% were achieved. The effect of image resolution on classification results was investigated. The eight million pixel images were upscaled in six stages to create images with decreasing pixel resolution. Rumex detection rates of over 90% were obtained at almost all resolutions, and there was only moderate misclassification of other objects to RUMOB. Image processing time ranged from 45 s for the full resolution images to 2.5 s for the lowest resolution ones.     

A double-layer model for improving the estimation of wheat canopy nitrogen content from unmanned aerial vehicle multispectral imagery

The accurate and rapid estimation of canopy nitrogen content (CNC) in crops is the key to optimizing in-season nitrogen fertilizer application in precision agriculture. However, the determination of CNC from field sampling data for leaf area index (LAI), canopy photosynthetic pigments (CPP; including chlorophyll a, chlorophyll b and carotenoids) and leaf nitrogen concentration (LNC) can be time-consuming and costly. Here we evaluated the use of high-precision unmanned aerial vehicle (UAV) multispectral imagery for estimating the LAI, CPP and CNC of winter wheat over the whole growth period. A total of 23 spectral features (SFs; five original spectrum bands, 17 vegetation indices and the gray scale of the RGB image) and eight texture features (TFs; contrast, entropy, variance, mean, homogeneity, dissimilarity, second moment, and correlation) were selected as inputs for the models. Six machine learning methods, i.e., multiple stepwise regression (MSR), support vector regression (SVR), gradient boosting decision tree (GBDT), Gaussian process regression (GPR), back propagation neural network (BPNN) and radial basis function neural network (RBFNN), were compared for the retrieval of winter wheat LAI, CPP and CNC values, and a double-layer model was proposed for estimating CNC based on LAI and CPP. The results showed that the inversion of winter wheat LAI, CPP and CNC by the combination of SFs+TFs greatly improved the estimation accuracy compared with that by using only the SFs. The RBFNN and BPNN models outperformed the other machine learning models in estimating winter wheat LAI, CPP and CNC. The proposed double-layer models (R²=0.67–0.89, RMSE=13.63–23.71 mg g^–1, MAE=10.75–17.59 mg g^–1) performed better than the direct inversion models (R²=0.61–0.80, RMSE=18.01–25.12 mg g^–1, MAE=12.96–18.88 mg g^–1) in estimating winter wheat CNC. The best winter wheat CNC accuracy was obtained by the double-layer RBFNN model with SFs+TFs as inputs (R²=0.89, RMSE=13.63 mg g^–1, MAE=10.75 mg g^–1). The results of this study can provide guidance for the accurate and rapid determination of winter wheat canopy nitrogen content in the field.     

Analysis of spectral difference between the foreside and backside of leaves in yellow rust disease detection for winter wheat

Disease detection by means of hyperspectral reflectance is inevitably influenced by the spectral difference between foreside (adaxial surface) and backside (abaxial surface) of a leaf. Taking yellow rust disease in winter wheat as an example, the spectral differences between the foreside and backside of healthy and diseased wheat leaves at both jointing stage and grain filling stage were investigated based on spectral measurements with a large sample size. The spectral difference between leaf orientations was found to be confused with disease signals to some extent. Firstly, the original bands and spectral features (SFs) that were sensitive to the disease were identified through a correlation analysis. Then, to eliminate the influence of leaf orientation, a pairwise t test was used to screen for the orientation insensitive bands and SFs. By conducting an overlapping procedure, the bands/SFs that were sensitive to the disease yet insensitive to the leaf orientations were selected and tested for disease detection. The results suggested that the Ref_{525–745 nm}, Ref_{1060–1068 nm}, DEP920–1120, DEP1070–1320, AREA1070–1320, SR and NDVI at the jointing stage, and the Ref_{606–697 nm}, Ref_{740–752 nm}, WID550–770, SR, NDVI, GNDVI, RDVI, GI and MCARI at the grain filling stage were capable of eliminating the influence of leaf orientation, and were retained for disease detection. Given these features, models based on the partial least square regression analysis showed a better performance at the grain filling stage, with the R ² of 0.854 and RMSE of 0.104. This result indicated that reliable estimation of disease severity can be made until the grain filling stage. In the future, more attention should be given to leaf orientation when detecting disease at the canopy level.     

Development of an agricultural crops spectral library and classification of crops at cultivar level using hyperspectral data

In the context of a growing interest in remote sensing for precision agriculture applications, the utility of space-borne hyperspectral imaging for the development of a crop-specific spectral library and automatic identification and classification of three cultivars for each of rice (Oryza sativa L.), chilli (Capsicum annuum L.), sugarcane (Saccharum officinarum L.) and cotton (Gossipium hirsutum L.) crops have been investigated in this study. The classification of crops at cultivar level using two spectral libraries developed using hyperspectral reflectance data at canopy scale (in-situ hyperspectral measurements) and at pixel scale (Hyperion data) has shown promising results with 86.5 and 88.8% overall classification accuracy, respectively. This observation highlights the possible integration of in-situ hyperspectral measurements with space-borne hyperspectral remote sensing data for automatic identification and discrimination of various crop cultivars. However, considerable spectral similarity is observed between cultivars of rice and sugarcane crops which may pose problems in the accurate identification of various crop cultivars.     

AdaBoost classifiers for pecan defect classification

One of the constraints in the adoption of machine vision inspection systems for food products is low classification accuracy. This study attempts to improve pecan defect classification accuracy by using machine learning classifiers: AdaBoost and support vector machine (SVM). X-ray images of good and defective pecans, 100 each, were segmented and features were extracted. Twenty classification runs were made to adjust parameters and 300 classification runs to compare classifiers. The Real AdaBoost classifier gave average classification accuracy of 92.2% for the Reverse water flow segmentation method and 92.3% for the Twice Otsu segmentation method. The Linear SVM classifier gave average classification accuracy of 90.1% for the Reverse water flow method and 92.7% for the Twice Otsu method. Computational time for the classifiers varied by two orders of magnitude: Bayesian (10⁻⁴ s), SVM (10⁻⁵ s), and AdaBoost (10⁻⁶ s). AdaBoost classifiers improved classification accuracy by 7% when Bayesian accuracy was poor (less than 89%). The AdaBoost classifiers also adapted well to data variability and segmentation methods. A minimalist AdaBoost classifier, more suitable for real time applications, using fewer features can be built. Overall, the selected AdaBoost classifiers improved classification accuracy, reduced classification time, and performed consistently better for pecan defect classification.     

小麦籽粒灌浆参数的基因型差异及其稳定性分析

四川盆地小麦的粒重优势明显,但其稳定性受生育后期不利环境的影响较大,研究籽粒灌浆参数的稳定性对于提高粒重和产量的稳定性十分重要。以10个粒重、熟期和株叶型差异明显的小麦品种为材料,设置5个试验环境(年份×地点),研究小麦籽粒灌浆参数的基因型差异及其稳定性。结果表明,参试品种的千粒重差异很大,环境均值变动在42.9~53.0g之间。多数灌浆参数均存在显著的基因型差异和环境效应,且年份效应大于地点效应。相同年份,粒重(TKW)、最大灌浆速率(Rmax)、平均灌浆速率(Rmean)、渐增期(T1)、缓增期(T3)及快增期灌浆速率(R2)主要受基因型的影响。CM104、CM42、CY23和NM836等品种的所有灌浆参数的稳定性都较好。相关分析表明,TKW与灌浆速率之间存在显著的正相关,而TKW的稳定性则主要与籽粒生长起始势(C0)、渐增期灌浆速率(R1)的稳定性关系密切,C0和R1的稳定性是决定粒重稳定的主要因素。     

Hyperspectral species mapping for automatic weed control in tomato under thermal environmental stress

This work studied the impacts of variations in environmental temperature on hyperspectral imaging features in the visible and near infrared regions for robust species identification for weed mapping in tomato production. Six major Californian processing tomato cultivars, black nightshade (Solanum nigrum L.) and redroot pigweed (Amaranthus retroflexus L.) were grown under a variety of diurnal temperature ranges simulating conditions common in the Californian springtime planting period and one additional treatment simulating greenhouse growing conditions. The principal change in canopy reflectance with varying temperature occurred in the 480-670 and 720-810 nm regions. The overall classification rate ranged from 62.5% to 91.6% when classifiers trained under single temperatures were applied to plants grown at different temperatures. Eliminating the 480-670 nm region from the classifier’s feature set mitigated the temperature effect by stabilizing the total crop vs. weed classification rate at 86.4% over the temperature ranges. A site-specific recalibration method was also successful in alleviating the bias created by calibrating the models on the extreme temperatures and increased the classification accuracy to 90.3%. A global calibration method, incorporating all four temperature conditions in the classifier feature space, provided the best average total classification accuracy of 92.2% out of the methods studied, and was fairly robust to the varying diurnal temperature conditions.     

Classification of oil palm fresh fruit bunches based on their maturity using portable four-band sensor system

Classification of oil palm fresh fruit bunch (FFB) maturity is a critical factor that dictates the quality of produced palm oil. This study evaluates a multi-band portable, active optical sensor system; comprising of four spectral bands, 570, 670, 750, and 870 nm, to detect oil palm FFB maturity. The in-field spectral reflectance data were collected using the sensor system from a total of 120 fresh fruit bunches. These fruit bunches were categories into unripe, ripe, and overripe classes. Different classifiers were applied to assess the applicability of using the sensor system. Based on the classification accuracies, data analysis on the spectral features (reflectance data and other features extracted from vegetation indices) indicated that the spectral reflectance data could be valuable in predicting the maturity of the fruit bunches. The quadratic discriminant analysis and discriminant analysis with Mahalanobis distance classifiers yielded highest average overall accuracies of greater than 85% in classifying oil palm FFB maturity. Additionally, the average individual class (unripe, ripe, and overripe) classification accuracies were also higher than 80%. Thus, optical sensing using four-band sensor system could be useful for oil palm FFB maturity classification under field condition.     

Identification of yellow rust in wheat using in-situ spectral reflectance measurements and airborne hyperspectral imaging

The aim of this study was to evaluate the accuracy of the spectro-optical, photochemical reflectance index (PRI) for quantifying the disease index (DI) of yellow rust (Biotroph Puccinia striiformis) in wheat (Triticum aestivum L.), and its applicability in the detection of the disease using hyperspectral imagery. Over two successive seasons, canopy reflectance spectra and disease index (DI) were measured five times during the growth of wheat plants (3 varieties) infected with varying amounts of yellow rust. Airborne hyperspectral images of the field site were also acquired in the second season. The PRI exhibited a significant, negative, linear, relationship with DI in the first season (r ² = 0.91, n = 64), which was insensitive to both variety and stage of crop development from Zadoks stage 3–9. Application of the PRI regression equation to measured spectral data in the second season yielded a coefficient of determination of r ² = 0.97 (n = 80). Application of the same PRI regression equation to airborne hyperspectral imagery in the second season also yielded a coefficient of determination of DI of r ² = 0.91 (n = 120). The results show clearly the potential of PRI for quantifying yellow rust levels in winter wheat, and as the basis for developing a proximal, or airborne/spaceborne imaging sensor of yellow rust in fields of winter wheat.     

Identifying the limiting factors driving the winter wheat yield gap on smallholder farms by agronomic diagnosis in North China Plain

North China Plain(NCP) is the primary winter wheat production region in China, characterized by smallholder farming systems. Whereas the winter wheat average yield of smallholder farmers is currently low, the yield potential and limiting factors driving the current yield gap remain unclear. Therefore, increasing the wheat yield in NCP is essential for the national food security. This study monitored wheat yield, management practices and soil nutrient data in 132 farmers' fields of Xushui County, Baoding City, Hebei Province during 2014–2016. These data were analyzed using variance and path analysis to determine the yield gap and the contribution of yield components(i.e., spikes per hectare, grain number per spike and 1 000-grain weight) to wheat yield. Then, the limiting factors of yield components and the optimizing strategies were identified by a boundary line approach. The results showed that the attainable potential yield for winter wheat was 10 514 kg ha~(–1). The yield gaps varied strongly between three yield groups(i.e., high, middle and low), which were divided by yield level and contained 44 farmers in each group, and amounted to 2 493, 1 636 and 814 kg ha~(–1), respectively. For the three yield components, only spikes per hectare was significantly different(P0.01) among the three yield groups. For all 132 farmers' fields, correlation between yield and spikes per hectare(r=0.51, P0.01), was significantly positive, while correlations with grain number per spike(r=–0.16) and 1 000-grain weight(r=–0.10) were not significant. The path analysis also showed that the spikes per hectare of winter wheat were the most important component to the wheat yield. Boundary line analysis showed that seeding date was the most limiting factor of spikes per hectare with the highest contribution rate(26.7%), followed by basal N input(22.1%) and seeding rate(14.5%), which indicated that management factors in the seeding step were the most important for affecting spikes per hectare. For desired spikes per hectare(6.598×10~6 ha~(–1)),the seeding rate should range from 210–300 kg ha~(–1), seeding date should range from 3th to 8th October, and basal N input should range from 90~(–1)80 kg ha~(–1). Compared to these reasonable ranges of management measures, most of the farmers' practices were not suitable, and both lower and higher levels of management existed. It is concluded that the strategies for optimizing yield components could be achieved by improving wheat seeding quality and optimizing farmers' nutrient management practices in the NCP.     

Evaluating hyperspectral vegetation indices for estimating nitrogen concentration of winter wheat at different growth stages

Many hyperspectral vegetation indices (VIs) have been developed to estimate crop nitrogen (N) status at leaf and canopy levels. However, most of these indices have not been evaluated for estimating plant N concentration (PNC) of winter wheat (Triticum aestivum L.) at different growth stages using a common on-farm dataset. The objective of this study was to evaluate published VIs for estimating PNC of winter wheat in the North China Plain for different growth stages and years using data from both N experiments and farmers’ fields, and to identify alternative promising hyperspectral VIs through a thorough evaluation of all possible two band combinations in the range of 350–1075 nm. Three field experiments involving different winter wheat cultivars and 4–6 N rates were conducted with cooperative farmers from 2005 to 2007 in Shandong Province, China. Data from 69 farmers’ fields were also collected to evaluate further the published and newly identified hyperspectral VIs. The results indicated that best performing published and newly identified VIs could explain 51% (R₇₀₀/R₆₇₀) and 57% (R₄₁₈/R₄₀₅), respectively, of the variation in PNC at later growth stages (Feekes 8–10), but only 22% (modified chlorophyll absorption ratio index, MCARI) and 43% (R₇₆₃/R₇₆₁), respectively, at the early stages (Feekes 4–7). Red edge and near infrared (NIR) bands were more effective for PNC estimation at Feekes 4–7, but visible bands, especially ultraviolet, violet and blue bands, were more sensitive at Feekes 8–10. Across site-years, cultivars and growth stages, the combination of R₃₇₀ and R₄₀₀ as either simple ratio or a normalized difference index performed most consistently in both experimental (R ² = 0.58) and farmers’ fields (R ² = 0.51). We conclude that growth stage has a significant influence on the performance of different vegetation indices and the selection of sensitive wavelengths for PNC estimation, and new approaches need to be developed for monitoring N status at early growth stages.