基于无人机高光谱的马铃薯冠层叶片全氮含量反演

为实现大田马铃薯冠层叶片全氮含量(LNC)的快速反演,利用低空无人机平台搭载成像光谱仪获取马铃薯冠层光谱数据,在综合比较原始反射率(R)、倒数变换反射率(1/R)、一阶微分变换反射率[D(R)]、二阶微分变换反射率[D(2R)]、倒数之对数变换反射率[lg(1/R)]的基础上,选择[D(2R)]用于后续试验。分别使用相关性分析(CA)、竞争性自适应重加权(CARS)、无信息变量消除(UVE)3种算法筛选特征光谱波段,使用偏最小二乘回归(PLSR)、支持向量机(SVM)构建马铃薯冠层LNC估测模型。结果表明：CA、CARS、UVE算法分别筛选出26、12、19个特征波段。在构建的PLSR模型中,用UVE筛选的特征波段建立的预测模型[UVE-D(2R)-PLSR]效果最好,在验证集上的决定系数(R²)和均方根误差(RMSE)分别为0.806 8和0.193 2;在构建的SVM模型中,用CARS筛选的特征波段建立的预测模型[CARS-D(2R)-SVM]效果最好,在验证集上的R²和RMSE分别为0.831 6和0.183 0。两模型对比,CARS-...     

基于机器学习和多颜色空间的冬小麦叶片氮含量估算方法研究

[目的]基于冠层图像可见光颜色分量进行冬小麦叶片氮含量(leaf nitrogen concentration,LNC)估算的算法研究,旨在为构建LNC估算模型提供方法借鉴。[方法]用Olympus E-620单反相机采集不同种植条件(2个品种、2个种植密度、3个氮处理)下2年生育期(2013年和2014年)的小麦冠层图像。基于H颜色分量的K-means聚类分割冠层图像,分别提取3个颜色空间HSV、L~*a~*b~*和RGB的3种基础颜色分量值,作为输入参数;分别使用多元线性回归(multivariate linear regression,MLR)、支持向量回归(support vector regression,SVR)、随机森林(random forest,RF)构建LNC估算模型,以决定系数R~2和均方根误差(root mean square error,RMSE)为评价指标;10×10嵌套交叉验证法分析3个颜色空间下各算法模型的拟合能力与泛化性能。[结果]单颜色空间下,3个算法模型的拟合与泛化性能以HSV空间下最优,L~*a~*b~*空间下其次,RGB空间下表现最弱;其中,RF拟合能力最强,但方差主导了泛化误差,模型过拟合;SVR拟合能力弱于RF,优于MLR,但该模型泛化性能最优;MLR拟合能力最弱,且偏差主导了泛化误差,模型欠拟合且受噪声干扰。融合3个颜色空间9种基础颜色分量的多颜色空间,3个模型拟合与泛化性能相对单颜色空间均更优,其中RF模型最优。与最优单色空间HSV相比,RF在训练集上R~2提高2.67%,RMSE降低11.59%;测试集上R~2提高7.57%,RMSE降低11.49%。多颜色空间下RF较SVR拟合更优,且泛化性能有效提升,提升比例高于SVR。[结论]基于3种算法构建的LNC估算模型,在融合3个颜色空间的9个基础颜色分量多颜色空间下,RF拟合与泛化性能最优,可为估算冬小麦LNC提供方法参考。     

数字图像技术在草莓氮素营养诊断中的应用研究

为了探讨利用数字图像处理技术进行草莓氮素营养诊断的可行性,通过6个水平的氮肥田间试验,采用数码相机获取草莓冠层图像,分析了不同供氮水平下草莓冠层图像色彩参数与施氮量、土壤无机氮和植株氮素营养指标之间的关系.研究表明:数字图像技术应用于监测草莓的氮素供应状况是可行的.其中,不同氮素处理下G/B值与G/(R+G+B)值的决定系数较高,G/(R+G+B)值与土壤无机氮、叶片硝酸盐及植株全氮之间的决定系数最高,利用G/(R+G+B)值的范围得出草莓开花期与结果期的氮肥推荐用量,从而反馈草莓氮素营养状况,进行氮素营养诊断.     

关中地区小麦冠层光谱与氮素的定量关系

【目的】分析不同生育期及整个生育期小麦叶片氮含量(LNC)与冠层光谱反射特征的关系,以实现对田间小麦活体氮素营养状况的监测,为小麦叶片氮素状况的精确诊断提供依据。【方法】以位于陕西关中地区杨凌揉谷镇、扶风马席村和巨良农场的3个小麦试验田为研究对象,测定不同长势及生育期小麦LNC及冠层光谱反射率,分析不同长势下小麦LNC和反射率的变化,并研究氮含量与冠层光谱反射率的相关性,以及小麦LNC与比值植被指数(RVI)、归一化植被指数(NDVI)的相关性,建立小麦LNC的敏感波段及光谱监测模型。【结果】在同一生育期,长势差的小麦叶片氮含量较低,长势较好的叶片氮含量高。与单波段相比,组合波段构成的植被指数RVI、NDVI与LNC的相关性明显提高,近红外波段(730~1 075nm)和红波段630,660,690nm组成组合波段的RVI、NDVI与LNC呈极显著正相关,其中LNC与RVI的相关性较高。利用独立的小麦田间试验数据,采用通用的均方根差(RMSE)、决定系数(R2)、准确度(斜率)3个指标对所建立的模型进行检验,最终选取RVI(970,690)为监测小麦LNC的最佳光谱参数,构建的最佳模型为LNC=0.176 3×RVI(970,690)0.775 6,R2为0.863,RMSE为0.137,准确度为0.979,接近于1。【结论】利用小麦冠层光谱反射率构建了预测小麦LNC的最佳模型,该模型具有较好的准确度和普适性,适用于整个生育期小麦叶片氮含量的监测。     

基于双波段光谱仪CGMD-302的小麦叶面积指数和叶干重监测

为探究双波段光谱仪CGMD-302在监测小麦长势上的可靠性与精准性,同时使用高光谱仪UniSpec SC与双波段光谱仪CGMD-302测试各生育时期小麦冠层信息,并定量分析了植被指数NDVI、RVI、DVI与叶面积指数和叶片干重之间的线性关系。结果表明,基于相同波段反射率计算出的高光谱仪植被指数和双波段光谱仪植被指数均能较好监测小麦群体长势。在CGMD-302监测的叶面积指数模型中,拟合方程的决定系数(R~2)均高于0.89,用以检验模型的均方根误差(RMSE)和相对误差(RE)分别小于0.792和0.225;叶片干重模型中,决定系数(R2)均高于0.85,用以检验模型的均方根误差(RMSE)和相对误差(RE)分别小于440kg/hm~2和0.239。通过分析发现,施氮270kg/hm~2既能保证产量又能兼顾品质,可作为适宜施氮量。适宜施氮量下,拔节期和孕穗期小麦适宜叶面积指数分别为:3.65±0.09和5.95±0.32;适宜叶干重分别为:(1 554±168)和(2 231±130)kg/hm~2。结合CGMD-302监测模型可推算出拔节期和孕穗期适宜冠层群体的植被指数区间并应用于冠层群体诊断。     

基于高光谱的玉米叶片氮含量监测模型

通过实测获取不同氮素营养水平下玉米冠层叶片光谱特性数据,利用可见光-近红外区域标准反射曲线的斜率和夹角等新型特征参数,结合预测学中最优权重组合原理,构建玉米叶片氮含量(LNC)监测模型。结果表明:光谱曲线夹角参数和斜率参数均与LNC存在较好的相关性,其中夹角参数Aγ和Aδ以及斜率参数Kr、Kb和Kn1的相关系数均在0.7左右;在单项监测模型中M(Kr/Kb)和M(Kb)以及M(Aδ/Aβ)和M(Aδ/Aα)的模型效果最好;最优模型由M(Kn1)和M(Kr/Kb)2个单项模型组合而成,其权重分别为0.245 5和0.754 5,最优模型的监测结果的决定系数(R2)为0.752 7,均方根误差(RMSE)为0.534,监测精度较单项模型明显提高。     

基于冠层RGB图像的冬小麦氮素营养指标监测

[目的]本文旨在探索基于冬小麦冠层RGB图像的氮素营养指标实时监测方法,为实现简便、准确的冬小麦氮素营养诊断和推荐施肥奠定基础。[方法]基于3年次的冬小麦大田氮肥梯度试验,采用数码相机在返青期和拔节期垂直拍摄冠层RGB图像。分析图像特征参数绿红通道比值(G/R)、绿红通道差值(GMR)、红光标准化值(NRI)、绿光标准化值(NGI)、色相(H)和冠层覆盖度(CC)与植株氮素生理指标间的关系,筛选氮素营养监测指标的最优图像特征参数,构建氮素营养指标估算模型。[结果]CC与冬小麦地上部生物量、氮积累量和叶面积指数(LAI)三者间的相关系数最高,分别为0.87、0.85和0.84(P0.01);其他特征参数与三者间的相关系数相对较低,其中H为0.81、0.77和0.79,NRI为-0.80、-0.77和-0.77,G/R为0.73、0.63和0.76,GMR为0.66、0.67和0.63。采用CC作为冬小麦氮素营养指标估算模型的输入参数,并分别使用异速生长函数和指数函数建立地上部生物量、氮积累量和LAI估算模型,异速生长函数这3个指标的估算模型R~2分别为0.82、0.76和0.82(P0.01),指数函数的R~2分别为0.80、0.74和0.85(P0.01)。利用独立试验数据对模型进行验证,异速生长函数模型预测值和观测值间的R~2平均为0.89(P0.01),地上部生物量、氮积累量和LAI预测值的均方根误差(RMSE)分别为31.09 g·m~(-2)、1.37 g·m~(-2)和0.16;指数函数模型预测值和观测值间的R~2平均也为0.89(P0.01),地上部生物量、氮积累量和LAI预测值的RMSE分别为28.95 g·m~(-2)、1.34 g·m~(-2)和0.17。[结论]异速生长函数和指数函数模型在利用CC对冬小麦氮素营养指标进行估算时均具有较好的预测性。基于RGB图像的监测方法操作简单、准确度高,可实时获取监测结果,具有较高的推广应用价值。     

基于冬小麦冠层数码图像的叶面积指数和叶片SPAD值的估算

为探究冠层图像分析技术在冬小麦长势监测中应用,6个施氮水平的田间试验条件下,在冬小麦拔节期采集冠层图像,并同步测定冬小麦叶面积指数和叶片SPAD值.通过图像分析软件计算了冬小麦冠层覆盖度及红、绿、蓝亮度值等10种色彩指数,分析了叶面积指数及叶片SPAD值与色彩指数和冠层覆盖度的相关性,利用逐步回归方法构建了叶面积指数及叶片SPAD值的估算模型.结果表明:冬小麦拔节期叶面积指数与冠层覆盖度及几个色彩指数呈极显著相关;叶片SPAD值与红光标准化值等几个色彩指数呈极显著相关;利用叶面积指数估算模型计算的预测值与实测值的线性回归方程的决定系数为0.771,相对均方根误差为25.181％;利用叶片SPAD值估算模型计算的预测值与实测值的线性回归方程的决定系数为0.644,相对均方根误差为6.734％.相关分析和估算模型验证结果表明,基于冠层图像分析的冬小麦拔节期叶面积指数和叶片SPAD值的监测是可行的.     

数字图像技术在黄瓜氮素营养诊断上的应用研究

试验设计了6种氮素水平处理温室黄瓜,利用数码相机获取黄瓜冠层图像,分析了不同供氮水平下黄瓜冠层图像参数与施氮量、土壤无机氮(Nmin)和植株氮素营养指标之间的关系.结果表明:黄瓜冠层图像G/(R G B)为诊断黄瓜结果期氮素营养状况的适宜数字化参数.建立了评价黄瓜结果期氮素营养丰缺的冠层图像数字化指标G/(R G B)的量值标准和氮肥推荐标准.     

冬小麦冠层叶绿素质量分数高光谱遥感反演研究

叶绿素质量分数是评估冬小麦生长状况和预测产量的重要参数,估算叶绿素质量分数对于冬小麦的生长监测具有重要意义。利用SPAD-502叶绿素仪和SVCHR 1024i型便携式高光谱仪对冬小麦冠层叶绿素质量分数和光谱特征进行田间测量,分别利用回归分析方法和BP神经网络方法搭建冬小麦叶绿素质量分数的估算模型,并将模型估算的叶绿素质量分数与田间实测的叶绿素质量分数进行对比,分析反演精度,从中筛选出精度最高的模型。结果表明:基于BP神经网络的冬小麦冠层叶绿素质量分数估算模型拟合精度要优于其他7种基于植被指数的估算模型,其相关系数(R)为0.961 4,均方根误差(RMSE)为1.875 4,相对误差(RE)为2.815 2%,以及检验方程的决定系数(R~2)为0.704 8,RMSE为1.744 6,RE为2.845 1%。研究结果为估测冬小麦冠层叶绿素质量分数提供参考,从而为冬小麦叶绿素质量分数的实时、快速、无损监测奠定基础。     

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。     

Nitrogen nutrition diagnosis for cotton under mulched drip irrigation using unmanned aerial vehicle multispectral images

Remote sensing has been increasingly used for precision nitrogen management to assess the plant nitrogen status in a spatial and real-time manner. The nitrogen nutrition index (NNI) can quantitatively describe the nitrogen status of crops. Nevertheless, the NNI diagnosis for cotton with unmanned aerial vehicle (UAV) multispectral images has not been evaluated yet. This study aimed to evaluate the performance of three machine learning models, i.e., support vector machine (SVM), back propagation neural network (BPNN), and extreme gradient boosting (XGB) for predicting canopy nitrogen weight and NNI of cotton over the whole growing season from UAV images. The results indicated that the models performed better when the top 15 vegetation indices were used as input variables based on their correlation ranking with nitrogen weight and NNI. The XGB model performed the best among the three models in predicting nitrogen weight. The prediction accuracy of nitrogen weight at the upper half-leaf level (R²=0.89, RMSE=0.68 g m^–2, RE=14.62% for calibration and R²=0.83, RMSE=1.08 g m^–2, RE=19.71% for validation) was much better than that at the all-leaf level (R²=0.73, RMSE=2.20 g m^–2, RE=26.70% for calibration and R²=0.70, RMSE=2.48 g m^–2, RE=31.49% for validation) and at the plant level (R²=0.66, RMSE=4.46 g m^–2, RE=30.96% for calibration and R²=0.63, RMSE=3.69 g m^–2, RE=24.81% for validation). Similarly, the XGB model (R²=0.65, RMSE=0.09, RE=8.59% for calibration and R²=0.63, RMSE=0.09, RE=8.87% for validation) also outperformed the SVM model (R²=0.62, RMSE=0.10, RE=7.92% for calibration and R²=0.60, RMSE=0.09, RE=8.03% for validation) and BPNN model (R²=0.64, RMSE=0.09, RE=9.24% for calibration and R²=0.62, RMSE=0.09, RE=8.38% for validation) in predicting NNI. The NNI predictive map generated from the optimal XGB model can intuitively diagnose the spatial distribution and dynamics of nitrogen nutrition in cotton fields, which can help farmers implement precise cotton nitrogen management in a timely and accurate manner.     

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.     

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.     

Spring Wheat Response to Fertilizer Nitrogen Following a Sugar Beet Crop Varying in Canopy Color

Experiments were conducted in the Red River Valley (RRV) of Minnesota to determine the responses of hard red spring wheat (Triticum aerstivum L.) to fertilizer N after a sugar beet (Beta vulgaris L.) crop that varied spatially in canopy color and N content. A color aerial photograph was acquired of the sugar beet field just prior to root harvest, and six sites were selected that varied in sugar beet canopy color, three each of green and yellow canopy sites. The three green sugar beet canopies returned 369, 265, and 266 kg N ha^–1 to the soil while the three yellow sugar beet canopies returned 124, 71, and 73 kg N ha^–1 to the soil. Spring wheat response to fall-applied urea-N fertilizer (0, 45, 90, 135, and 180 kg N ha^–1) was determined the following year at each of the above antecedent canopy sites. Soil NO₃-N in the top 0.6 m of soil varied among the locations with a range of 35 to 407 kg NO₃-N ha^–1 at the green canopy sites and 12 to 23 kg NO₃-N ha^–1 at the yellow canopy sites. Application of fertilizer N according to traditional recommendation methods would have resulted in fertilizer applications at all three yellow canopy sites and two of the three green canopy sites. At the antecedent green sugar beet canopy sites, fertilizer N had little or no effect on spring wheat grain yields, grain N concentration, anthesis dry matter, and anthesis N content. In contrast, fertilizer N increased all four parameters at the antecedent yellow sugar beet canopy sites. The data indicate that fertilizer N management can be improved by using remote sensing to delineate management zones according to antecedent sugar beet canopy color.     

基于数字图像技术的冬油菜氮素营养诊断

【目的】利用田间氮肥梯度试验探讨数字图像技术对冬油菜氮素营养无损评估预测的可行性,明确该技术的最佳数码参数和方程模型,为数字图像技术进行冬油菜氮素无损诊断提供依据。【方法】2013-2014年在湖北省武穴市开展不同施氮处理田间试验,以冬油菜为试验材料,设置不同氮素水平（0、90、180、270和360 kg·hm^-2）,分别于六叶期、十叶期、蕾薹期和开花期,利用数码相机获取冠层数字图像数据,同时采集植株样品分析其生长特征值,研究其相关性并建立氮素营养参数的方程模型。利用2014-2015年独立氮肥水平试验,对上述方程模型拟合精度进行验证并绘制1﹕1线性关系图。【结果】数字图像红光值（R）、红光标准化值（NRI）和绿光与蓝光比值（G/B）与冬油菜氮营养状况常规诊断指标地上部生物量、叶片氮浓度和叶绿素浓度等呈负相关关系,而绿光值（G）、蓝光值（B）、绿光与红光比值（G/R）、蓝光与红光比值（B/R）、绿光标准化值（NGI）和蓝光标准化值（NBI）则与上述指标呈正相关关系,红光标准化值（NRI）与其他数码参数相比能更好地表征冬油菜的氮素营养状况,蕾薹期红光标准化值NRI与氮肥用量、地上部生物量、叶片氮浓度、叶绿素浓度、氮素吸收量和氮营养指数之间的关系可分别用线性方程y(t·hm^-2)=-8.003x+2.706、y(t·hm^-2)=-106.072x+38.200、y(g·kg^-1)=-692.99x+ 261.84、y(mg·g^-1)=-12.750x+5.665、y(kg·hm^-2)=-4087.416x+1414.274和y=-27.198x+9.812来表达,其相关性达到极显著水平。2014-2015年独立试验模型检验结果表明,叶片氮浓度、叶绿素浓度和氮营养指数实测值与预测值的决定系数R²分别为0.917^**、0.746^**和0.953^**;均方根误差RMSE分别为0.821、0.330和0.228;相对误差RE %分别为26.32%、28.57%和28.39%,模型预测精度较好。【结论】数字图像技术可以用于冬油菜氮素营养的评估预测,评估时期为蕾薹期（包括）之前均可,最佳预测参数为红光标准化值NRI,参数的最佳方程模型为直线方程函数。     

Estimating nitrogen concentration in rape from hyperspectral data at canopy level using support vector machines

The estimation of nitrogen concentration from remotely sensed data has been the subject of some work. However, few studies have addressed the effective model for monitoring nitrogen status at canopy level using Support Vector Machines (SVM). The present study is focused on the assessment of an estimation model for nitrogen concentration of rape canopy with hyperspectral data. Two types of estimation model, the traditional statistical method based on stepwise linear regression (SLR) and the emerging computationally powerful techniques based on support vector machines were applied The Root Mean Square Error (RMSE) and T values were used to assess their predictability. The results show that a better agreement between the observed and the predicted nitrogen concentration were obtained by using the SVM model. Compared to the SLR model, the SVM model improved the results by lowering RMSE by 11.86–21.13 %, and by increasing T by 20.00–29.41 % for different spectral transformations. The study demonstrated the potential of SVM to estimate nitrogen concentration using canopy level hyperspectral data and it was concluded that SVM may provide a useful exploratory and predictive tool when applied to canopy-level hyperspectral reflectance data for monitoring nitrogen status of rape.