Predictions of soil organic carbon using laboratory-based hyperspectral data in the northern Tianshan mountains, China

This paper presents a study dealing with soil organic carbon (SOC) estimation of soil through the combination of soil spectroscopy and multivariate stepwise linear regression. Soil samples were collected in the three sub-regions, dominated by brown calcic soil, in the northern Tianshan Mountains, China. Spectral measurements for all soil samples were performed in a controlled laboratory environment by a portable ASD FieldSpec FR spectrometer (350–2,500 nm). Twelve types of transformations were applied to the soil reflectance to remove the noise and to linearize the correlation between reflectance and SOC content. Based on the spectral reflectance and its derivatives, hyperspectral models can be built using correlation analysis and multivariable statistical methods. The results show that the main response range of soil organic carbon is between 400 and 750 nm. Correlation analysis indicated that SOC has stronger correlation with the second derivative than with the original reflectance and other transformations data. The two models developed with laboratory spectra gave good predictions of SOC, with root mean square error (RMSE) <5.0. The use of the full visible near-infrared spectral range gave better SOC predictions than using visible separately. The multivariate stepwise linear regression of second derivate model (model A) is optimal for estimating SOC content, with a determination coefficient of 0.894 and RMSE of 0.322. The results of this research study indicated that, for the grassland regions, combining soil spectroscopy and mathematical statistical methods does favor accurate prediction of SOC.     

Canopy assessment of biochemical features by ground-based hyperspectral data for an invasive species, giant reed (Arundo donax)

This study explored the potential use of hyperspectral data in the non-destructive assessment of chlorophyll, carbon, and nitrogen content of giant reed at the canopy level. We found that pseudoabsorption and derivatives of original hyperspectral data were able to describe the relationship between spectral data and measured biochemical characteristics. Based on correlogram analyses of ground-based hyperspectral data, we found that derivatives of pseudoabsorption were the best predictors of chlorophyll, carbon, and nitrogen content of giant reed canopies. Within the visible region, spectral data significantly correlated with chlorophyll content at both 461 nm and 693 nm wavelengths. Within the near-infrared region, carbon levels correlated with hyperspectral data at five causal wavelengths: 1038 nm, 1945 nm, 1132 nm, 1525 nm, and 1704 nm. The best spectral wavelength for estimating nitrogen content was 1542 nm. Such relationships between nutrient content and spectral data were best represented by exponential functions in most situations.     

基于实测光谱的海河悬浮物浓度反演研究

悬浮物浓度是评价水质优劣的重要指标之一。以天津滨海新区海河为研究区域,进行光谱数据测量和水体样本采集,并在实验室进行水质参数提取,得到归一化光谱反射率和光谱一阶微分数据。然后对光谱测量数据和悬浮物浓度实测值进行相关性分析,发现896 nm归一化光谱反射率和光谱反射率比值(R_(896)/R_(546))与悬浮物浓度相关性较好。最后,分别建立单波段、波段比值和一阶微分函数拟合模型,进行对比分析。结果表明,基于R_(896)/R_(546)的二次多项式模型拟合效果最好,方差齐性检验(F)值也是最高的,可用于该地区水体的悬浮物浓度反演和预测。     

Changes in labile soil organic matter fractions following land use change from monocropping to poplar-based agroforestry systems in a semiarid region of Northeast China

Labile fractions of soil organic matter (SOM) respond rapidly to land management practices and can be used as a sensitive indicator of changes in SOM. However, there is little information about the effect of agroforestry practices on labile SOM fractions in semiarid regions of China. In order to test the effects of land use change from monocropping to agroforestry systems on labile SOM fractions, we investigated soil microbial biomass C (MBC) and N, particulate organic matter C (POMC) and N (POMN), as well as total organic C (TOC) and total N (TN) in the 0- to 15-cm and the 15- to 30-cm layers in 4-year-old poplar-based agroforestry systems and adjoining monocropping systems with two different soil textures (sandy loam and sandy clay loam) in a semiarid region of Northeast China. Our results showed that poplar-based agroforestry practices affected soil MBC, POMC, and POMN, albeit there was no significant difference in TOC and TN. Agroforestry practices increased MBC, POMC, and POMN in sandy clay loam soils. However, in sandy loam soils, agroforestry practices only increased MBC and even decreased POMC and POMN at the 0- to 15-cm layer. Our results suggest that labile SOM fractions respond sensitively to poplar-based agroforestry practices and can provide early information about the changes in SOM in semiarid regions of Northeast China and highlight that the effects of agroforestry practices on labile SOM fractions vary with soil texture.     

Estimation of chlorophyll a content in inland turbidity waters using WorldView-2 imagery: a case study of the Guanting Reservoir,Beijing, China

Complex optical properties, such as non-pigment suspension and colored dissolved organic matter (CDOM), make it difficult to achieve accurate estimations of remotely sensed chlorophyll a (Chla) content of inland turbidity. Recent attempts have been made to estimate Chla based on red and near-infrared regions where non-pigment suspension and CDOM have little effect on water reflectance. The objective of this study is to validate the applicability of WV-2 imagery with existing effective estimation methods from MERIS when estimating Chla content in inland turbidity waters. The correlation analysis of measured Chla content and WV-2 imagery bands shows that the Chla sensitive bands of WV-2 are red edge, NIR 1, and NIR 2. The coastal band is designed for seawater Chla detection. However, the high correlation with turbidity data and low correlation with Chla made coastal band unsuitable for estimating Chla in inland waters. The high-resolution water body images were extracted by combining the spectral products (NDWI) with the spatial morphological products (sobel edge detection). The estimation results show that the accuracy of the single band and NDCI is not as good as the two-band method, three-band method, stepwise regression algorithm (SRA) and support vector machines (SVM). The SVM estimation accuracy was the highest with an R², RMSE, and URMSE of 0.8387, 0.4714, and 19.11%, respectively. This study demonstrates that the two-band and three-band methods are effective for estimating Chla in inland water for WV-2 imagery. As a high-precision estimation method, SVM has great potential for inland turbidity water Chla estimation.     

基于高光谱分形分析的水稻镉污染动态监测

水稻中过量镉会影响叶绿素含量和细胞结构,进而改变水稻在光谱维上的特征表现.利用长春郊区实测大田水稻高光谱数据和同步测量的土壤、水稻生化参数,系统分析水稻冠层光谱曲线的分形特征及不同镉污染胁迫程度下光谱分维与两类光谱指数的变化情况,构建光谱分维、光谱指数与水稻叶片镉浓度的空间分布图,探讨光谱分维与光谱指数在提取镉污染胁迫等级信息中的相互关系及物理机制.研究表明,光谱分维相对于光谱指数能够更好地综合反映镉污染胁迫下水稻生理特征参数的变化,因而为有效地大范围动态监测水稻镉污染提供诊断依据.     

Characterization of soil organic matter content of two sample size populations along a climatic transect

Soil organic matter (SOM) content was determined in two populations of soil samples that were taken from 0–2 soil depth. One population represented soil samples that were takenfrom a square of 25 cm² in size (small-S population) and the other population represented soil samples that were taken from a square of 2500 cm² in size (large-L population). Thesamples were collected on hillslopes in different climatic regions: Mediterranean (GIV), semi-arid (MAL), mildly-arid (MIS) and arid (KAL). The results of both S and L populationsshowed decreasing SOM mean and variance from the Mediterranean site to the arid site. Statistical and spatial characteristics of each population were compared between the climatic regions. In addition, comparison between the two populations was made foreach site. The difference in sample size did not significantly affect the mean values of SOM of the two populations in sitesGIV, MAL and KAL, but did affect the mean at site MIS. At all study sites, except for site MAL, the variance increased with decreasing sample size. At sites GIV and KAL the coefficient ofvariation of S population was higher (more than 1.5 times) thanthat of L population, whereas at sites MAL and MIS, the differences were negligible. The relationships between the valuesof S and L samples at the individual sampling points defined thebackground of the study sites, which reflects the effect of vegetation (type), grazing, biological crust and soil properties.It was found that at the extreme sites GIV and KAL the backgroundwas characterized by relatively low SOM content with small areas of high organic matter content. At site MIS the background wascharacterized by relatively high SOM with small areas of low organic matter content. At site MAL the background was not dominated by high values of SOM nor by low ones. The spatial pattern of L population became more simple with increasing aridity. At the relatively wet sites the spatial pattern did notdepend on the sample size while in the more arid sites it was sample size dependent. It was indicated that the spatial structure of SOM at the semi-arid and mildly arid sites is anisotropic whereas at the Mediterranean and arid sites it is isotropic.     

Monitoring and assessing of landscape heterogeneity at different scales

In numerous studies, spatial and spectral aggregations of pixel information using average values from imaging spectrometer data are suggested to derive spectral indices and the subsequent vegetation parameters that are derived from these. Currently, there are very few empirical studies that use hyperspectral data, to support the hypothesis for deriving land surface variables from different spectral and spatial scales. In the study at hand, for the first time ever, investigations were carried out on fundamental scaling issues using specific experimental test flights with a hyperspectral sensor to investigate how vegetation patterns change as an effect of (1) different spatial resolutions, (2) different spectral resolutions, (3) different spatial and spectral resolutions as well as (4) different spatial and spectral resolutions of originally recorded hyperspectral image data compared to spatial and spectral up- and downscaled image data. For these experiments, the hyperspectral sensor AISA-EAGLE/HAWK (DUAL) was mounted on an aircraft to collect spectral signatures over a very short time sequence of a particular day. In the first experiment, reflectance measurements were collected at three different spatial resolutions ranging from 1 to 3 m over a 2-h period in 1 day. In the second experiment, different spectral image data and different additional spatial data were collected over a 1-h period on a particular day from the same test area. The differently recorded hyperspectral data were then spatially and spectrally rescaled to synthesize different up- and down-rescaled images. The normalised difference vegetation index (NDVI) was determined from all image data. The NDVI heterogeneity of all images was compared based on methods of variography. The results showed that (a) the spatial NDVI patterns of up- and downscaled data do not correspond with the un-scaled image data, (b) only small differences were found between NDVI patterns determined from data recorded and resampled at different spectral resolutions and (c) the overall conclusion from the tests carried out is that the spatial resolution is more important in determining heterogeneity by means of NDVI than the depth of the spectral data. The implications behind these findings are that we need to exercise caution when interpreting and combining spatial structures and spectral indices derived from satellite images with differently recorded geometric resolutions.     

Hyperspectral Characteristics of Canopy Components and Structure for Phenological Assessment of an Invasive Weed

Spectral reflectance values of four canopy components (stems, buds, opening flowers, and postflowers of yellow starthistle (Centaurea solstitialis)) were measured to describe their spectral characteristics. We then physically combined these canopy components to simulate the flowering stage indicated by accumulated flower ratios (AFR) 10%, 40%, 70%, and 90%, respectively. Spectral dissimilarity and spectral angles were calculated to quantitatively identify spectral differences among canopy components and characteristic patterns of these flowering stages. This study demonstrated the ability of hyperspectral data to characterize canopy components, and identify different flowering stages. Stems had a typical spectral profile of green vegetation, which produced a spectral dissimilarity with three reproduction organs (buds, opening flowers, and postflowers). Quantitative differences between simulated flower stages depended on spectral regions and phenological stages examined. Using full-range canopy spectra, the initial flowering stage could be separated from the early peak, peak, and late flowering stages by three spectral regions, i.e. the blue absorption (around 480 nm) and red absorption (around 650 nm) regions and NIR plateau from 730 nm to 950 nm. For airborne CASI data, only the red absorption region and NIR plateau could be used to identify the flowering stages in the field. This study also revealed that the peak flowering stage was more easily recognized than any of the other three stages.     

A new multiscale approach for monitoring vegetation using remote sensing-based indicators in laboratory, field, and landscape

Remote sensing is an important tool for studying patterns in surface processes on different spatiotemporal scales. However, differences in the spatiospectral and temporal resolution of remote sensing data as well as sensor-specific surveying characteristics very often hinder comparative analyses and effective up- and downscaling analyses. This paper presents a new methodical framework for combining hyperspectral remote sensing data on different spatial and temporal scales. We demonstrate the potential of using the “One Sensor at Different Scales” (OSADIS) approach for the laboratory (plot), field (local), and landscape (regional) scales. By implementing the OSADIS approach, we are able (1) to develop suitable stress-controlled vegetation indices for selected variables such as the Leaf Area Index (LAI), chlorophyll, photosynthesis, water content, nutrient content, etc. over a whole vegetation period. Focused laboratory monitoring can help to document additive and counteractive factors and processes of the vegetation and to correctly interpret their spectral response; (2) to transfer the models obtained to the landscape level; (3) to record imaging hyperspectral information on different spatial scales, achieving a true comparison of the structure and process results; (4) to minimize existing errors from geometrical, spectral, and temporal effects due to sensor- and time-specific differences; and (5) to carry out a realistic top- and downscaling by determining scale-dependent correction factors and transfer functions. The first results of OSADIS experiments are provided by controlled whole vegetation experiments on barley under water stress on the plot scale to model LAI using the vegetation indices Normalized Difference Vegetation Index (NDVI) and green NDVI (GNDVI). The regression model ascertained from imaging hyperspectral AISA-EAGLE/HAWK (DUAL) data was used to model LAI. This was done by using the vegetation index GNDVI with an R ² of 0.83, which was transferred to airborne hyperspectral data on the local and regional scales. For this purpose, hyperspectral imagery was collected at three altitudes over a land cover gradient of 25 km within a timeframe of a few minutes, yielding a spatial resolution from 1 to 3 m. For all recorded spatial scales, both the LAI and the NDVI were determined. The spatial properties of LAI and NDVI of all recorded hyperspectral images were compared using semivariance metrics derived from the variogram. The first results show spatial differences in the heterogeneity of LAI and NDVI from 1 to 3 m with the recorded hyperspectral data. That means that differently recorded data on different scales might not sufficiently maintain the spatial properties of high spatial resolution hyperspectral images.     

Molecular-level methods for monitoring soil organic matter responses to global climate change

Soil organic matter (SOM) is one of the most complex natural mixtures on earth. It plays a critical role in many ecosystem functions and is directly responsible for sustaining life on our planet. However, due to its chemical heterogeneity, SOM composition at molecular-level is still not completely clear. Consequently, the response of SOM to global climate change is difficult to predict. Here we highlight applications of two complementary molecular-level methods (biomarkers and nuclear magnetic resonance (NMR)) for ascertaining SOM responses to various environmental changes. Biomarker methods that measure highly specific molecules determine the source and decomposition stage of SOM components. However, biomarkers only make up a small fraction of SOM components because much of SOM is non-extractable. By comparison, (13)C solid-state NMR allows measurement of SOM in its entirety with little or no pretreatment but suffers from poor resolution (due to overlapping of SOM components) and insensitivity, and thus subtle changes in SOM composition may not always be detected. Alternatively, (1)H solution-state NMR methods offer an added benefit of improved resolution and sensitivity but can only analyze SOM components that are fully soluble (humic type molecules) in an NMR compatible solvent. We discuss how these complementary methods have been employed to monitor SOM responses to: soil warming in a temperate forest, elevated atmospheric CO(2) and nitrogen fertilization in a temperate forest, and permafrost thawing in the Canadian High Arctic. These molecular-level methods allow some novel and important observations of soil dynamics and ecosystem function in a changing climate.     

Estimation of chlorophyll-a concentration using field spectral data: a case study in inland Case-II waters, North China

In the remote sensing of chlorophyll-a (Chla) in inland Case-II waters, the assumption that the optical parameter of Chla specific absorption coefficient a*ph remains constant usually restrains application of many models. In this paper, we presented a newly developed model [Rrs(-1)(lambda1) - Rrs(-1)(lambda2)] x Rrs(lambda3) x a*ph(-1)(lambda1) which was improved on a previous three-band model to isolate interferences from a*ph. In terms of the importance of water optical properties in the model development, spectral and absorption characteristics were analyzed for Shitoukoumen Reservoir and Songhua Lake in Northeast China, as typical examples of inland Case-II waters. Both waters showed overwhelming absorption sum of tripton and chromophoric dissolved organic matter (CDOM) owing to their relatively low Chla contents (1.53 to 19.35 microgl(-1)). According to the optical characteristics of waters studied, optimal positions for lambda (1), lambda (2) and lambda (3) were spectrally tuned to be at 664, 684 and 705 nm, respectively. The model allowed accurate Chla estimation with a determination coefficient (R (2)) close to 0.98 and a root mean square error (RMSE) of 0.87 microgl(-1). Comparison of different models further showed the stability of the improved model, implying its potential use in water color remote sensing. Although the findings underline the rationale behind the improved model, an extensive database containing data in different water conditions and water types is required to generalize its application.     

Research on the relationship between the fractional coverage of the submerged plant Vallisneria spiralis and observed spectral parameters

The present paper discusses the relationship between the coverage fraction of submerged plants and the observed spectral characteristics. The purpose of this paper is to validate a remote sensing technology to monitor the change in the plant composition of a water body. In the current study, the reflectance spectra of the submerged plant Vallisneria spiralis at different fraction coverages of the wetland in Hangzhou Bay were measured. The relationships between the fraction coverage of V. spiralis and simulated Quickbird normalized difference vegetation index (NDVI), red edge, and other spectral characteristic parameters were established. The results showed that the spectral reflectance characteristics of submerged plant V. spiralis were mainly in the visible light (490–650 nm) and near infrared (700–900 nm). The rate of change of the blue band curve and simulated Quickbird NDVI showed a higher correlation with the V. spiralis coverage, so estimation models of the fraction coverage were constructed using these parameters. The estimated fraction coverage of V. spiralis with different models were validated with ground data, and the accuracy of estimation models was assessed. The most suitable estimated fraction coverage of V. spiralis was obtained using the rate of change of the blue band curve and simulated Quickbird NDVI. The present work demonstrated a method to monitor the distribution and dynamical variation of submerged plants at the large scale.     

Exploring the relationship between vegetation spectra and eco-geo-environmental conditions in karst region, Southwest China

Remote sensing of local environmental conditions is not accessible if substrates are covered with vegetation. This study explored the relationship between vegetation spectra and karst eco-geo-environmental conditions. Hyperspectral remote sensing techniques showed that there were significant differences between spectral features of vegetation mainly distributed in karst and non-karst regions, and combination of 1,300- to 2,500-nm reflectance and 400- to 680-nm first-derivative spectra could delineate karst and non-karst vegetation groups. Canonical correspondence analysis (CCA) successfully assessed to what extent the variation of vegetation spectral features can be explained by associated eco-geo-environmental variables, and it was found that soil moisture and calcium carbonate contents had the most significant effects on vegetation spectral features in karst region. Our study indicates that vegetation spectra is tightly linked to eco-geo-environmental conditions and CCA is an effective means of studying the relationship between vegetation spectral features and eco-geo-environmental variables. Employing a combination of spectral and spatial analysis, it is anticipated that hyperspectral imagery can be used in interpreting or mapping eco-geo-environmental conditions covered with vegetation in karst region.     

基于无人机高光谱遥感在太湖蓝藻水华监测中的一次应用

采用机载高光谱视频相机,在4个季节对太湖蓝藻进行7次、18个架次的有效拍摄。对拍摄到的高光谱影像进行辐射定标、几何拼接等预处理后,提取不同浓度蓝藻和水草等其他物体的高光谱数据,发现不同浓度的蓝藻光谱在680 nm后表现出较大差异。采用主成分分析（PCA）对高光谱数据降维后,结合k-近邻（kNN）分类算法,可实现对蓝藻的精准定位。定性识别结果经光谱预处理后,采用连续投影算法（SPA）进行特征波段提取,发现蓝藻光谱的季节差异主要表现在450 nm~570 nm和760 nm~910 nm波段。     

Predicting field capacity,wilting point,and the other physical properties of soils using hyperspectral reflectance spectroscopy: two different statistical approaches

In this study, we examined the ability of reflectance spectroscopy to predict some of the most important soil parameters for irrigation such as field capacity (FC), wilting point (WP), clay, sand, and silt content. FC and WP were determined for 305 soil samples. In addition to these soil analyses, clay, silt, and sand contents of 145 soil samples were detected. Raw spectral reflectance (raw) of these soil samples, between 350 and 2,500-nm wavelengths, was measured. In addition, first order derivatives of the reflectance (first) were calculated. Two different statistical approaches were used in detecting soil properties from hyperspectral data. Models were evaluated using the correlation of coefficient (r), coefficient of determination (R ²), root mean square error (RMSE), and residual prediction deviation (RPD). In the first method, two appropriate wavelengths were selected for raw reflectance and first derivative separately for each soil property. Selection of wavelengths was carried out based on the highest positive and negative correlations between soil property and raw reflectance or first order derivatives. By means of detected wavelengths, new combinations for each soil property were calculated using rationing, differencing, normalized differencing, and multiple regression techniques. Of these techniques, multiple regression provided the best correlation (P?<?0.01) for selected wavelengths and all soil properties. To estimate FC, WP, clay, sand, and silt, multiple regression equations based on first(2,310)-first(2,360), first(2,310)-first(2,360), first(2,240)-first(1,320), first(2,240)-first(1,330), and raw(2,260)-raw(360) were used. Partial least square regression (PLSR) was performed as the second method. Raw reflectance was a better predictor of WP and FC, whereas first order derivative was a better predictor of clay, sand, and silt content. According to RPD values, statistically excellent predictions were obtained for FC (2.18), and estimations for WP (2.0), clay (1.8), and silt (1.63) were acceptable. However, sand values were poorly predicted (RDP?=?0.63). In conclusion, both of the methods examined here offer quick and inexpensive means of predicting soil properties using spectral reflectance data.     

Uncertainty assessment of mapping mercury contaminated soils of a rapidly industrializing city in the Yangtze River Delta of China using sequential indicator co-simulation

Accurate characterization of heavy-metal contaminated areas and quantification of the uncertainties inherent in spatial prediction are crucial for risk assessment, soil remediation, and effective management recommendations. Topsoil samples (0–15 cm) (n = 547) were collected from the Zhangjiagang suburbs of China. The sequential indicator co-simulation (SIcS) method was applied for incorporating the soft data derived from soil organic matter (SOM) to simulate Hg concentrations, map Hg contaminated areas, and evaluate the associated uncertainties. High variability of Hg concentrations was observed in the study area. Total Hg concentrations varied from 0.004 to 1.510 mg kg⁻¹ and the coefficient of variation (CV) accounts for 70%. Distribution patterns of Hg were identified as higher Hg concentrations occurred mainly at the southern part of the study area and relatively lower concentrations were found in north. The Hg contaminated areas, identified using the Chinese Environmental Quality Standard for Soils critical values through SIcS, were limited and distributed in the south where the SOM concentration is high, soil pH is low, and paddy soils are the dominant soil types. The spatial correlations between Hg and SOM can be preserved by co-simulation and the realizations generated by SIcS represent the possible spatial patterns of Hg concentrations without a smoothing effect. Once the Hg concentration critical limit is given, SIcS can be used to map Hg contaminated areas and quantitatively assess the uncertainties inherent in the spatial prediction by setting a given critical probability and calculating the joint probability of the obtained areas.     

植物大气污染响应高光谱监测实例研究

选择某钢铁企业绿化树种桂花当年生叶片作为供试样本,测试叶片光谱反射率之后分别测试叶液pH值、叶片含硫量、叶绿素含量、叶片含水量,研究生长在S02 污染环境下桂花叶片光谱的变化以及相应的部分生理生化指标的变化.研究表明,污染较严重生产区采集的叶片光谱反射率和红边斜率均较生活区低.叶片含硫量随大气SO2 浓度的增减而相应地变化,叶片叶绿素含量、叶液pH值、叶片含水量的变化规律与叶片含硫量和大气SO2浓度的变化规律相反.     

Study of the effect of pH, salinity and DOC on fluorescence of synthetic mixtures of freshwater and marine salts

In order to provide support for the discussion of the fate of organic matter in estuaries, a laboratory simulation was performed by changing freshwater ionic strength, pH and organic matter content. The change in spectroscopic characteristics caused by variations in salinity, pH and organic matter concentration in the filtered samples was observed by UV-Vis and fluorescence spectroscopy. The increase in emission fluorescence intensity of dissolved organic matter (DOM) due to increasing salinity (in the range 0 to 5 g l-1) is affected by the pH of the samples. The emission fluorescence intensity at the three maxima observed in the fluorescence spectra, is linearly correlated with dissolved organic carbon (DOC) concentration at several salinity values in the same sample. The increase in organic matter concentration caused a shift in the emission peak wavelength at 410 nm for several salinity values. We concluded that it is necessary to take into account the influence of salinity and pH on emission fluorescence of dissolved organic matter if it is to be used as a tracer in estuarine or near shore areas.     

Pollution of intensively managed greenhouse soils by nutrients and heavy metals in the Yellow River Irrigation Region,Northwest China

The present study aimed to assess the potential ecological risk of heavy metals and nutrient accumulation in polytunnel greenhouse soils in the Yellow River irrigation region (YRIR), Northwest China, and to identify the potential sources of these heavy metals using principal component analysis. Contents of available nitrogen (AN), phosphorus (AP), and potassium (AK) in the surface polytunnel greenhouse soils (0–20 cm) varied from 13.42 to 486.78, from 39.10 to 566.97, and from 21.64 to 1,156.40 mg kg^?1, respectively, as well as AP, soil organic matter (SOM) and AK contents tended to increase significantly at the 0–20- and 20–40-cm soil layers. Heavy metal accumulations occurred in the polytunnel greenhouse soils as compared to arable soils, especially at a depth of 20 cm where Cd, Zn and Cu contents were significantly higher than arable soil. Cd and As were found to be the two main polluting elements in the greenhouse soils because their contents exceeded the thresholds established for greenhouse vegetable production HJ333-2006 in China and the background of Gansu province. It has been shown that Cd, Cu, Pb and Zn at the 0–20-cm soil layer were derived mainly from agricultural production activities, whereas contents of Cr and Ni at the same soil layer were determined by ‘natural’ factors and As originated from natural sources, deposition and irrigation water.