基于近红外光谱分析的土壤分层氮素含量预测

准确、快速地估测土壤中的氮素含量是推动配方施肥顺利开展的保障。该研究在不同区域随机选取了30个点位,每个点位分别取其表土层(0~30 cm)、心土层(30~48 cm)以及底土层(48~60 cm)3个部位进行取样,利用傅里叶型光谱分析仪MATRIX_I测量了土壤样本在近红外区域的吸收光谱,并使用实验室手段测量了土壤样本的水分及氮素含量。分析了不同层次土壤样本的吸收光谱特性,以及土壤水分、氮素不同层次的变化规律。同时对原始光谱吸收率进行一阶微分处理,而后利用微分光谱与土壤全氮含量进行相关性分析,选取反应土壤全氮含量的敏感波段1 387、1 496、1 738、1 876、2 120以及2 316 nm。利用所得敏感波段与土壤氮素含量分别建立多元线性回归模型,BP神经网络预测模型以及基于遗传算法优化的BP神经网络建模。结果显示,基于遗传算法优化的BP神经网络建模,其决定系数为0.883,均方根误差为0.0278 mg/kg。表土层土壤的预测验证结果决定系数为0.716,均方根误差为0.031 mg/kg;心土层土壤的预测验证结果决定系数为0.801,均方根误差为0.030 mg/kg;底土层土壤的预测验证结果决定系数为0.667,均方根误差为0.033 mg/kg。无论是建模精度还是模型在土壤各个层次的预测精度相比于多元线性回归模型和BP神经网络模型相比都有了显著的提高,说明该方法在土壤全氮含量预测过程中具有明显的优势,可应用于实际生产。     

基于近红外光谱和支持向量机的土壤参数预测

应用支持向量机算法对实时土壤光谱数据进行处理,获得了土壤全氮和有机质的回归模型并研究了模型随参数变化的规律。从中国农业大学试验田采集了150个土样,用光谱仪获取了原始土壤样本的近红外光谱,用实验室分析法获取了各样本的全氮和有机质含量。以近红外光谱数据为自变量对2个土壤参数进行了回归建模并评价了算法各参数对模型的影响。研究表明土壤参数适合于全谱支持向量回归。对于土壤全氮,基于小波降噪NIR光谱的SVM回归模型的标定R2为0.9224,验证R2为0.3667;对于土壤有机质,基于原始NIR光谱的SVM回归模型     

基于近红外光谱分析的土壤全氮含量估测研究

应用近红外光谱分析技术对比研究基于土壤风干样本和鲜样来预测全氮含量的可行性。选取水稻土为研究对象,首先分析了不同水分土壤的光谱特征,显示随水分含量增加,吸光度升高,且鲜样的吸光度高于干样。通过比较不同预处理方法,对土壤干鲜样分别采用逐步多元回归(SMLR)和偏最小二乘法(PLSR)建立了相应的近红外模型。结果表明,利用近红外光谱均可预测干鲜土壤样本的全氮含量,特别是利用偏最小二乘法建立的标定模型,预测精度高,反演性较好,鲜样和干样外部验证决定系数分别达到0.89和0.91,相对误差仅为6.92%和5.92%,研究结果可以为田间土壤全氮含量的估测提供技术依据和参考。     

黑土养分含量的航空高光谱遥感预测

为监测黑龙江省黑土典型区土壤的养分元素含量,综合利用统计理论与光谱分析方法,研究建三江农场黑土土壤的3类养分含量与土壤光谱之间的关系,建立土壤全氮、有效磷、速效钾含量高光谱反演模型,实现土壤养分元素含量定量预测。对黑土土壤航空高光谱数据进行处理,应用偏最小二乘回归（PLSR）和BP神经网络方法分别建立土壤养分元素含量的高光谱定量反演模型,结果表明：全氮PLSR和BP神经网络预测模型的RPIQ值（样本观测值第三和第一四分位数之差与均方根误差的比值）分别为2.42和2.80;有效磷PLSR和BP神经网络模预测型的RPIQ值分别为0.83和1.67;速效钾PLSR和BP神经网络模型的RPIQ值分别为2.00和2.33。试验证明土壤全氮和速效钾的光谱定量预测模型具备较好的精度和预测能力。但有效磷的预测效果不是特别理想,仅可达到近似定量预测的要求;全氮、有效磷和速效钾的预测精度,BP神经网络建模相比偏最小二乘建模有更好的精度和预测能力,预测精度分别提高6.5%、10.1%和6.6%。     

贮备饲料近红外光谱模型快速预测青绿饲料营养成分含量

为快速检测饲料的营养成分,该研究利用贮备饲料的近红处技术(near-infrared,NIR)快速分析模型预测青绿饲料的营养成分含量。基于贮备饲料的NIR定标模型,将建模优化模式转移应用到青绿饲料的营养成分定量检测,以判断模型转移能力。在实验室环境下扫描并记录新鲜的青绿饲料样本和储存的贮备饲料样本的近红外反射光谱,利用230个贮备饲料样本进行光谱定标训练,以修正偏最小二乘(modified-partial least squares,M-PLS)建模方法,结合随机局部样本、局部选参、局部非连续性可调、交叉检验等技术相结合的方式建立局部优化模型,分别测试120个贮备饲料样本和120个青绿饲料样本中的氮(nitrogen,N)、中性洗涤纤维(neutral detergent fiber,NDF)、酸性洗涤纤维(acid detergent fiber,ADF)含量。将贮备饲料的定标校正模型应用于贮备饲料验证样本的营养成分测定,其标准误差(square error of prediction,SEP):N为1.02、NDF为16.56和ADF为13.47,相关系数均在0.9以上,相对预测偏差(relative prediction derivation,RPD)均大于3;该模型具有对青绿饲料样本的营养成分预测能力,其预测SEP:N为0.90、NDF为14.11和ADF为9.98,预测相关系数均在0.9以上,预测RPD均大于3,达到快速检测误差标准。由于局部建模过程中考虑了数据的潜在非线性结构和具有近似光谱响应的样本之间的不均匀性,相对全局建模方式而言具有更好的数据驱动性质,其建模效果优于全局建模方法。结果表明,基于贮备饲料样本建立的NIR定标校正模型可以用于青绿饲料营养成分的预测,特别是局部分析模型的应用能够提高NIR快速分析的预测精度。     

基于近红外光谱的土壤全氮含量估算模型

土壤全氮是诊断土壤肥力水平和指导作物精确施肥所需的重要信息,建立土壤全氮的近红外光谱估测模型并对建模波段进行优化选择对于土壤养分信息快速获取和精确农业发展具有重要意义。该研究以中国中、东部地区5种主要类型土壤为研究对象,利用近红外光谱仪采集土壤样品的光谱信息,结合近红外区域分子振动特点选取全谱、合频、一倍频、二倍频和N-H基团及其组合的8个波段,采用多元散射校正等多种预处理方法组合进行处理,结合偏最小二乘法(PLS)对每个波谱区域进行定标建模。结果表明,利用4000～5500cm-1波谱区域结合附加散射校正处理过的原始光谱建立的模型精度表现最好,其内部互验证决定系数达到0.90,均方根误差为0.16。经不同类型土壤的观测资料检验,模型验证决定系数为0.91,均方根误差为0.15,相对分析误差RPD为3.40,表明模型具有极好的预测能力。因此,利用近红外光谱可以实现土壤全氮的快速估测,且以合频波段(4000～5500cm-1)为建模区域可以得到更好的预测效果。     

大尺度采样下不同模型方法预测土壤全氮空间分布研究——以海南岛为例

在大区域尺度、有限土壤样点情况下,为探索准确预测土壤属性的方法,以海南岛为研究区,采用近似网格采样方法,采集130个样点,用多元线性回归(MLR)、普通克里格(OK)和回归克里格(RK)3种模型方法进行土壤全氮预测,并以29个验证点比较了预测精度。结果显示:1)对较大区域进行土壤全氮的空间分布的预测精度为OKRKMLR;2)3种模型对土壤全氮含量空间预测分布趋势基本一致,总趋势为岛内自东向西方向逐渐降低;3)0~5 cm土壤全氮含量与土地利用方式呈极显著相关关系,0~20 cm土壤全氮含量与归一化植被指数呈显著相关,20~40、40~60 cm土壤全氮含量与归一化植被指数、坡度呈极显著或显著相关。     

灌溉水中悬浮固体对土壤水分入渗性能的影响

为监测黑龙江省黑土典型区土壤的养分元素含量,综合利用统计理论与光谱分析方法,研究建三江农场黑土土壤的3类养分含量与土壤光谱之间的关系,建立土壤全氮、有效磷、速效钾含量高光谱反演模型,实现土壤养分元素含量定量预测。对黑土土壤航空高光谱数据进行处理,应用偏最小二乘回归（PLSR）和BP神经网络方法分别建立土壤养分元素含量的高光谱定量反演模型,结果表明：全氮PLSR和BP神经网络预测模型的RPIQ值（样本观测值第三和第一四分位数之差与均方根误差的比值）分别为2.42和2.80;有效磷PLSR和BP神经网络模预测型的RPIQ值分别为0.83和1.67;速效钾PLSR和BP神经网络模型的RPIQ值分别为2.00和2.33。试验证明土壤全氮和速效钾的光谱定量预测模型具备较好的精度和预测能力。但有效磷的预测效果不是特别理想,仅可达到近似定量预测的要求;BP神经网络建模相比偏最小二乘建模有更好的精度和预测能力,预测精度分别提高6.5%、10.1%和6.6%。     

基于土壤优化光谱参数估测太湖地区土壤全氮含量

为明确太湖地区土壤全氮的高光谱特征,构建定量分析模型,以江苏省无锡市滨湖区为研究区域,选取地理位置跨度大、土壤质地相似的93个样品,进行土壤风干样品全氮含量测定和光谱数据采集,对光谱反射率进行一阶微分,运用相关系数峰谷值法筛选敏感波长,将敏感波长两两结合进行土壤调节光谱指数(MSASI)运算。将两两结合后敏感波段分别采用多元线性回归分析、人工神经网络分析和偏最小二乘法构建土壤全氮含量的定量高光谱分析模型。结果表明,研究区内土壤全氮含量与光谱反射率呈正相关,敏感波段包括420～444 nm和480～537 nm。基于土壤调节光谱指数的多元线性回归分析对敏感波段诊断的效果最佳(R~2=0.98、RMSE=0.04),其精度高、可靠性强,是筛选出的最佳土壤全氮含量估测模型。偏最小二乘法模型(R~2=0.70、RMSE=0.13)次之,而人工神经网络模型(R~2=0.69、RMSE=0.15)精度最低。该研究结果为太湖地区土壤全氮水平的高光谱快速估测提供了方法借鉴,可为土壤养分精准管理提供技术参考。     

北京典型耕作土壤养分的近红外光谱分析

为研究土壤养分含量分布信息,以从北京郊区一块试验田采集的72个土壤样品为试验材料,应用傅里叶变换近红外光谱技术分析了土样的全氮、全钾、有机质养分含量和pH值。采用偏最小二乘法(PLS)对光谱数据与土壤养分实测值进行回归分析,建立预测模型,以模型决定系数(R2)、校正标准差(RMSECV)、预测标准差(RMSEP)和相对分析误差(RPD)作为模型精度的评价指标。结果表明,利用该模型与光谱数据对土壤全氮、全钾、有机质养分含量和pH值进行预测,结果与实测数据具有较好的一致性,最高决定系数R2达到0.9544。偏最小二乘回归方法建立的养分预测模型能准确地对北京地区褐土土质全氮、有机质、全钾和pH值4种养分进行预测。     

The use of visible and near‐infrared spectroscopy for the analysis of soil water repellency

This study investigated the potential of visible/near‐infrared reflectance spectroscopy (Vis‐NIRS) to predict soil water repellency (SWR). The top 40 mm of soils (n = 288) across 48 sites under pastoral land‐use in the North Island of New Zealand, which represented 10 soil orders and covered five classes of drought proneness, were analysed by standard laboratory methods and Vis‐NIRS. Soil WR was measured by using the molarity of ethanol droplet (MED) and the water drop penetration time (WDPT) tests. Soil organic carbon content (%C) was also measured to examine a possible relationship with SWR. A partial least squares regression (PLSR) model was developed by using Vis‐NIRS spectral data and the reference laboratory data. In addition, we explored the power of discrimination based on WDPT classes using partial least squares discriminant analysis (PLS‐DA). The PLSR of the processed spectra produced moderately accurate prediction for MED (R²_val = 0.61, RPD_val = 1.60, RMSE_val = 0.59) and good prediction for %C (R²_val = 0.82, RPD_val = 2.30, RMSE_val = 2.72). When the data from the 10 soil orders were considered separately and based on soil order rather than being grouped, the prediction of MED was further improved except for the Allophanic, Brown, Organic and Ultic soil orders. The PLS‐DA was successful in classifying 60% of soil samples into the correct WDPT classes. Our results indicate clearly that Vis‐NIRS has the potential to predict SWR. Further improvement in the prediction accuracy of SWR is envisaged by increasing the understanding of the relationship between Vis‐NIRS and the SWR of all New Zealand soil orders as a function of their physical properties and chemical constituents such as hydrophobic compounds.     

The potential of hyperspectral images and partial least square regression for predicting total carbon,total nitrogen and their isotope composition in forest litterfall samples

Purpose

The main objective of this study was to examine the potential of using hyperspectral image analysis for prediction of total carbon (TC), total nitrogen (TN) and their isotope composition (δ¹³C and δ¹⁵N) in forest leaf litterfall samples.

Materials and methods

Hyperspectral images were captured from ground litterfall samples of a natural forest in the spectral range of 400–1700 nm. A partial least-square regression model (PLSR) was used to correlate the relative reflectance spectra with TC, TN, δ¹³C and δ¹⁵N in the litterfall samples. The most important wavelengths were selected using β coefficient, and the final models were developed using the most important wavelengths. The models were, then, tested using an external validation set.

Results and discussion

The results showed that the data of TC and δ¹³C could not be fitted to the PLSR model, possibly due to small variations observed in the TC and δ¹³C data. The model, however, was fitted well to TN and δ¹⁵N. The cross-validation R² _cv of the models for TN and δ¹⁵N were 0.74 and 0.67 with the RMSE_cv of 0.53% and 1.07‰, respectively. The external validation R² _ex of the prediction was 0.64 and 0.67, and the RMSE_ex was 0.53% and 1.19 ‰, for TN and δ¹⁵N, respectively. The ratio of performance to deviation (RPD) of the predictions was 1.48 and 1.53, respectively, for TN and δ¹⁵N, showing that the models were reliable for the prediction of TN and δ¹⁵N in new forest leaf litterfall samples.

Conclusions

The PLSR model was not successful in predicting TC and δ¹³C in forest leaf litterfall samples using hyperspectral data. The predictions of TN and δ¹⁵N values in the external litterfall samples were reliable, and PLSR can be used for future prediction.

     

淮北平原土壤高光谱特征及有机质含量预测

以安徽省淮北平原的蒙城县为研究区,采集131个表层土壤(0～20 cm)样品。采用Cary 5000分光光度计测定土壤光谱反射率,分析该地区典型土壤类型的光谱特征,利用偏最小二乘回归方法建立土壤有机质光谱预测模型。首先比较不同光谱变换对土壤有机质含量光谱预测建模的影响;其次根据光谱相似性对土壤样品进行分类,比较不同土壤类型和不同光谱分类的有机质光谱预测精度。结果表明:①不同土壤有机质含量和不同土壤类型光谱曲线在整体波段范围内趋势基本一致;有机质含量与光谱反射率呈显著负相关;有机质含量越低,曲线特征差异明显,可能是受其他因素的影响;②土壤光谱反射率经倒数的对数处理后,有机质光谱建模的决定系数和相对分析误差均有所提高,均方根误差降低,模型预测效果较优;③按照光谱相似性分类后建立的有机质光谱预测模型,比按土壤类型建立的光谱预测模型精度明显提高。     

同时估测土壤全氮、有机质和速效氮含量的光谱指数研究

通过系统分析我国中、东部地区5种不同类型土壤风干样本的有机质、全氮及速效氮含量与近红外(1 000～2 500 nm)光谱反射率之间的关系,进而构建了适合同时估测这3种养分含量的光谱参数及定量估算模型。结果表明,同时与3种养分指标相关较高的波段范围为1 879～1 890与2 050～2 100 nm,其中1 881和2 070 nm两个波段的反射率经多元散射校正及Savitzky-Golay平滑处理并构建而成的差值指数DI(CR1 881,CR2 070)与土壤有机质、全氮及速效氮含量具有良好的线性相关性。独立的观测资料检验显示,基于DI(CR1 881,CR2 070)的估测模型对全氮、有机质和速效氮的预测决定系数R2分别为0.83、0.79和0.72,均方根误差(RMSE)分别为0.20 g kg-1、4.71 g kg-1和23.96 mg kg-1,相对分析误差(RPD)分别为2.56、2.30和2.93。表明DI(CR1 881,CR2 070)是一种可同时估测土壤中3种养分含量的良好光谱指数。     

Determination of total soil organic C and hot water‐extractable C from VIS‐NIR soil reflectance with partial least squares regression and spectral feature selection techniques

The calibration of soil organic C (SOC) and hot water‐extractable C (HWE‐C) from visible and near‐infrared soil reflectance spectra is hindered by the complex spectral interaction of soil chromophores that usually varies from one soil or soil type to another. The exploitation of spectral variables from spectroradiometer data is further affected by multicollinearity and noise. In this study, a set of soil samples (Fluvisols, Podzols, Cambisols and Chernozems; n = 48) representing a wide range of properties was analysed. Spectral readings with a fibre‐optics visible to near‐infrared instrument were used to estimate SOC and HWE‐C contents by partial least squares regression (PLS). In addition to full‐spectrum PLS, spectral feature selection techniques were applied with PLS (uninformative variable elimination, UVE‐PLS, and a genetic algorithm, GA‐PLS). On the basis of normalized spectra (mean centring + vector normalization), the order of prediction accuracy was GA‐PLS ? UVE‐PLS > PLS for SOC; for HWE‐C, it was GA‐PLS > UVE‐PLS, PLS. With GA‐PLS, acceptable cross‐validated (cv) prediction accuracies were obtained for the complete dataset (SOC, , RPD_cv = 2.42; HWE‐C_cv, , RPD_cv = 2.13). Splitting the soil data into two groups with different basic properties (Podzols compared with Fluvisols/Cambisols; n = 21 and n = 23, respectively) improved SOC predictions with GA‐PLS distinctly (Podzols, , RPD_cv = 3.14; Fluvisols/Cambisols, , RPD_cv = 3.64). This demonstrates the importance of using stratified models for successful quantitative approaches after an initial rough screening. GA selection frequencies suggest that the spectral region over 1900 nm, and in particular the hydroxyl band at 2200 nm are of great importance for the spectral prediction of both SOC and HWE‐C.     

煤矿区土壤有机碳含量的高光谱预测模型

可见—近红外光谱已被证明是一种快速、及时、有效的土壤有机碳含量预测工具。利用Field Spec4对济宁鲍店矿区的104个土壤样品进行光谱测量,采用Savitzky-Golay卷积平滑(SG)、多元散射校正(MSC)及数学变换等多种方式组合对光谱预处理,并运用偏最小二乘回归分析建立土壤有机碳含量预测模型,进而探讨煤矿区土壤有机碳含量的高精度预测方法。结果表明:(1)不同的光谱预处理方法对建模结果影响差异较大,建模结果以SG加MSC预处理再结合光谱反射率的一阶微分变换最优,建模R~2=0.86,RMSE=2.0g/kg,验证R~2=0.78,RMSE=1.81g/kg,RPD=2.69。(2)倒数和倒数的对数与土壤有机碳含量的相关性曲线接近重合,与反射率曲线成反比,但是建模效果远低于反射率;光谱反射率的一阶微分能明显提高500~600nm波段相关性。(3)光谱反射率随土壤有机碳的含量减少而增大,当有机碳含量较低时,其波谱的近红外波段反射率响应能力也随之降低,反射率直接建模难度加大。     

基于光谱吸收特征的土壤含水量预测模型研究

为了定量分析土壤含水量与反射光谱特征之间关系,并为土壤含水量速测提供理论依据。以黑土作为研究对象,测定实验室光谱反射率,利用去包络线方法提取反射光谱特征指标,建立土壤水分含量高光谱预测模型。结果表明:黑土含水量与1 420 nm、1 920 nm附近吸收谷的主要光谱特征(吸收谷深度、宽度、面积)呈显著正相关;1 920 nm附近吸收谷可作为黑土土壤水分的特征吸收谷,由其光谱特征参数预测黑土含水量;以1 920 nm附近吸收谷面积为自变量建立的一元线性回归模型预测精度高,输入量少,可以作为土壤含水量速测仪器研制的理论依据。     

Near infrared reflectance spectroscopy: A tool to characterize the composition of different types of exogenous organic matter and their behaviour in soil

In addition to total organic carbon and nitrogen, potential organic carbon mineralization under controlled laboratory conditions and indicators such as the indicator of remaining organic carbon in soil (I_ROC), based on Van Soest biochemical fractionation and short-term carbon mineralization in soil, are used to predict the evolution of exogenous organic matter (EOM) after its application to soils. The purpose of this study was to develop near infrared reflectance spectroscopy (NIRS) calibration models that could predict these characteristics in a large dataset including 300 EOMs representative of the broad range of such materials applied to cultivated soils (plant materials, animal manures, composts, sludges, etc.). The NIRS predictions of total organic matter and total organic carbon were satisfactory (R²_P = 0.80 and 0.85, ratio of performance to deviation, RPD_P = 2.2 and 2.6, respectively), and prediction of the Van Soest soluble, cellulose and holocellulose fractions were acceptable (R²_P = 0.82, 0.73 and 0.70, RPD_P = 2.3, 1.9 and 1.8, respectively) with coefficients of variation close to those of the reference methods. The NIRS prediction of carbon mineralization during incubation was satisfactory and indeed better regarding the short-term results of mineralization (R²_P = 0.78 and 0.78, and RPD_P = 2.1 and 2.0 for 3 and 7 days of incubation, respectively). The I_ROC indicator was predicted with fairly good accuracy (R²_P = 0.79, RPD_P = 2.2). Variables related to the long-term C mineralization of EOM in soil were not predicted accurately, except for I_ROC which was based on analytical and well-identified characteristics, probably because of the increasing interactions and complexity of the factors governing EOM mineralization in soil as a function of incubation time. This study demonstrated the possibility of developing NIRS predictive models for EOM characteristics in heterogeneous datasets of EOMs. However, specific NIRS predictive models still remain necessary for sludges, organo-mineral fertilizers and liquid manures.     

Rapid determination of lime requirement by mid‐infrared spectroscopy: A promising approach for precision agriculture

Mid‐infrared spectroscopy (MIRS) has proven to be a cost‐effective, high throughput measurement technique for soil analysis. After multivariate calibration mid‐infrared spectra can be used to predict various soil properties, some of which are related to lime requirement (LR). The objective of this study was to test the performance of MIRS for recommending variable rate liming on typical Central European soils in view of precision agriculture applications. In Germany, LR of arable topsoils is commonly derived from the parameters organic matter content (SOM), clay content, and soil pH (CaCl₂) as recommended by the Association of German Agricultural Analytical and Research Institutes (VDLUFA). We analysed a total of 458 samples from six locations across Germany, which all revealed large within‐field soil heterogeneity. Calcareous topsoils were observed at some positions of three locations (79 samples). To exclude such samples from LR determination, peak height at 2513 cm^?1 of the MIR spectrum was used for identification. Spectra‐based identification was accurate for carbonate contents > 0.5%. Subsequent LR derivation (LR_SPP) from MIRS‐PLSR predictions of SOM, clay, and pH (CaCl₂) for non‐calcareous soil samples using the VDLUFA look‐up tables was successful for all locations (R² = 0.54–0.82; RMSE = 857–1414 kg CaO ha^?1). Alternatively, we tested direct LR prediction (LR_DP) by MIRS‐PLSR and also achieved satisfactory performance (R² = 0.52–0.77; RMSE = 811–1420 kg CaO ha^?1; RPD = 1.44–2.08). Further improvement was achieved by refining the VDLUFA tables towards a stepless algorithm. It can be concluded that MIRS provides a promising approach for precise LR estimation on heterogeneous arable fields. Large sample numbers can be processed with low effort which is an essential prerequisite for variable rate liming in precision agriculture.     

基于地类分层的土壤有机质光谱反演校正样本集的构建

以江汉平原滨湖地区不同土地利用类型的土壤样本为例,比较了基于目标土壤理化性质的浓度梯度法、扩展的基于多种理化性质的综合法(P-KS)、基于光谱信息的KS法、最邻近样本去除法(reduce nearest neighbor samples,RNNS)法和基于浓度分层并结合光谱信息的C-KS、C-RNNS法,基于地类分层再结合上述方法,构建具有不同层次土壤信息代表性的校正集,采用偏最小二乘回归法,建立土壤有机质可见光/近红外光谱反演模型。结果表明,具有单一代表性的浓度梯度法、KS法、RNNS法难以建立适用模型;具有光谱与理化性质二元代表性的C-KS方法模型预测精度得到了明显的提升,相对分析误差(ratio of performance to standard deviation,RPD)为1.66;考虑土地利用类型后,浓度梯度法、RNNS法与C-KS法模型预测精度有明显的提升,RPD分别达到了1.84、1.51、1.75,模型具有良好的适用性。说明具有多层次土壤信息代表性的校正集构建方法对提高土壤有机质可见光/近红外光谱反演模型的适用性具有较好作用。