lable at ScienceDirect

Environmental Pollution 159 (2011) 3057e3070

Contents lists avai

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

Assessment of heavy metal pollution from a Fe-smelting plant in urban riversediments using environmental magnetic and geochemical methods

Chunxia Zhang a,*, Qingqing Qiao a, John D.A. Piper b, Baochun Huang a

a State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences, No. 19 Bei Tucheng Xilu, Chaoyang Dist., Beijing 100029, ChinabGeomagnetism Laboratory, Department of Earth and Ocean Science, University of Liverpool, Liverpool L69 7ZE, UK

a r t i c l e i n f o

Article history:Received 10 November 2010Received in revised form1 April 2011Accepted 7 April 2011

Keywords:Environmental magnetismHeavy metalsRiver sedimentsIndustrial plantHunanChina

* Corresponding author.E-mail address: cxzhang@mail.iggcas.ac.cn (C. Zha

0269-7491/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.envpol.2011.04.006

a b s t r a c t

Environmental magnetic proxies provide a rapid means of assessing the degree of industrial heavy metalpollution in soils and sediments. To test the efficiency of magnetic methods for detecting contaminatesfrom a Fe-smelting plant in Loudi City, Hunan Province (China) we investigated river sediments fromLianshui River. Both magnetic and non-magnetic (microscopic, chemical and statistical) methods wereused to characterize these sediments. Anthropogenic heavy metals coexist with coarse-grained magneticspherules. It can be demonstrated that the Pollution Load Index of industrial heavy metals (Fe, V, Cr, Mo,Zn, Pb, Cd, Cu) and the logarithm of saturation isothermal remanent magnetization, a proxy for magneticconcentration, are significantly correlated. The distribution heavy metal pollution in the Lianshui River iscontrolled by surface water transport and deposition. Our findings demonstrate that magnetic methodshave a useful and practical application for detecting and mapping pollution in and around modernindustrial cities.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Heavy metal pollution of aquatic ecosystems is becominga growing global problem as population increase, and urbanizationand industrialization expand. A consequence of the growth of heavyindustry has been the addition of high concentrations of heavymetals originating from anthropogenic inputs including industrialwastewater discharges, sewage wastewater, fossil fuel combustionand atmospheric deposition (Birch et al., 1996; Linnik and Zubenko,2000; Lwanga et al., 2003; Chaparro et al., 2004; Jordanova et al.,2004; Spiteri et al., 2005; Rijal et al., 2010; Sekabira et al., 2010).Trace amounts of heavy metals are always present in water, andsome elements may be immobilized within stream sediments andthus could be involved in absorption, co-precipitation, complexformation, and co-adsorbedwith Fe oxides and hydroxides, or otherparticulate forms (Awofolu et al., 2005; Okafor and Opuene, 2007;Mohiuddin et al., 2010). The development of sensitive tools andprocedures for detecting and monitoring water quality and thusavoiding heavy metal poisoning is a pressing task.

Heavy metal concentrations in soil and stream sediments can beused to reveal the intensity and range of local and regional pollu-tion. However, traditional geochemical methods (e.g. AAS, ICP-MS)

ng).

All rights reserved.

are relatively complex, time-consuming and expensive, and aretherefore not suitable for performing mapping or monitoring oflarge-scale pollution. In recent decades, environmental magneticmethods have been widely applied as proxy indicators of envi-ronmental pollution because they are simple, rapid, and have low-cost and non-destructive characteristics (Hoffmann et al., 1999;Petrovský et al., 2000; Gautam et al., 2004; Zhang et al., 2008).Measurements of magnetic susceptibility of soil, street dust, sedi-ments etc. have been used to map areas polluted by industrialemissions, such as coal-burning power plants, lead ore smelters,and roadside pollution due to automotive traffic and other atmo-spheric pollution (Thompson and Oldfield, 1986; Morris et al., 1995;Hay et al., 1997; Strzyszcz and Magiera, 1998; Kapi�cka et al., 1999,2001; 2008; Petrovský et al., 2001; Zhang et al., 2006; Blahaet al., 2008a, 2008b). The relationship between magnetic parame-ters and heavy metals has been investigated on the proxyassumption that sources of magnetic particles and heavymetals aregenetically related. High correlation coefficients between certainpollutants and magnetic susceptibility or SIRM (saturationisothermal remanence) have sometimes been reported (Helleret al., 1998; Petrovský et al., 2001; Chaparro et al., 2004;Jordanova et al., 2004) whilst different correlations betweencertain magnetic parameters and heavy metals like Pb, Zn, Cu havebeen detected at single emission sites and can be used for identi-fying potential sources and discriminating between them(Banerjee, 2003; Jordanova et al., 2003; Al-Khashman, 2004; Lu and

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e30703058

Bai, 2006; Wang and Qin, 2006; Maher et al., 2008; Lu et al.,2009a,b; Bijaksana and Huliselan, 2010; Bu�cko et al., 2010; Yanget al., 2010). In recent years the use of magnetic parameters asproxies for quantifying the contents of certain contaminants suchas heavy metals in street dust, atmospheric particles or soil havebeen demonstrated (Shu et al., 2001; Kim et al., 2007, 2009; Duanet al., 2010). Clear correlations were found between PLI andmagnetic concentration parameters for topsoil, stream and marinesediments (Chan et al., 2001; Wang and Qin, 2006; Lu et al., 2007,2009a,b; Yang et al., 2007; Chaparro et al., 2008; Canbay et al.,2010). These studies prove that quantified relationships betweenmagnetic parameters and heavy metals can be constructed basedon appropriate indexes. However, few results have been reportedfor the detection of heavy metal pollution by surface water trans-port and deposition in river sediments. This has prompted thepresent evaluation of whether magnetic parameters can be used asa proxy for such scenarios.

In this paper, we compare the results obtained from magneticparameters and geochemical analyses of heavy metals by inte-grating results from a spectrum of methods including environ-mental magnetic, geochemical, electron microscopy and EDXanalyses, and incorporating multivariate statistical analysisincluding of principal component analysis and cluster analysis. Themain objectives have been to discriminate between the contribu-tions of different pollution sources and quantifying relationshipsbetween magnetic parameters and heavy metal contaminationalong the Lianshui River near Loudi City in the Hunan Province ofChina. The study is thus a contribution to the potential use ofmagnetic measurements and their application for evaluating thescale of industrial pollution caused by surface water transport anddeposition in an urban environment.

2. Sampling and laboratory measurements

Loudi is a fast-developing industrial city located near the middleof Hunan Province and covers about 426 km2. The major industryoperating for fifty years has been a Fe-smelting plant; this mainpollution source is located at the north-western side of the city. Thelandscape of the region comprises low foothills and the Quaternarycover is represented by red loam, with a parent rock of limestone.The east side of the city is a wide plain with altitudes lower than inthe West. Lianshui River enters the city from the north-westernside, passes by the outside wall of the industrial plant, and thentraverses the city center to leave the city at the north-eastern side(Fig. 1a). At the outside of the Fe-smelting plant (sampling sitesLDR6-LDR10) there are three wastewater outlets (Fig. 1); OL1 is thewastewater outlet of the Fe-smelting plant. OL2 is the outlet of thepower supply unit of the Fe-smelting plant and includes the sin-tering net ringwater system overflowwater as well as effluent fromsmall blast furnace water and boiler workshops; the remainingoutlet OL3 comes from the cleaning ground and road wastewaters.The rate of wastewater flow at OL2 is about 400e500 m3/h.

River sediments were collected along the Lianshui River duringJuly 21e23, 2009, using a gravity sampler. Samples were taken fromthe upstream section (before entering the city) to the downstreamregion after leaving the city area; the total sampling distance alongthe river is about 20 km (Fig. 1a,b). At each sample site, two or threecores were collected and the material was sub-sampled at aninterval of 2 cm. In total, 19 sample sites were studied comprising260 river sediment samples (Fig. 1a).

In the laboratory all samples were freeze-dried with lyophilizerand mechanically homogenized and sieved through a 1 mm meshto remove small stones. Plastic boxes 2 � 2 � 2 cm in size werefilled with individual samples for magnetic measurements. Allsample susceptibility values were measured using a Bartington

MS2 susceptibility meter with 470 Hz operating frequency, andmass-specific susceptibility (c) values were calculated. Anhyste-retic remanent magnetizations (ARMs) were imparted in a peakalternating field (AF) of 80 mT with a superimposed direct current(DC) bias field of 0.05 mT parallel to the AF. Isothermal remanentmagnetization (IRM) experiments were performed using animpulse magnetizer and the IRM acquired in a field of 1.0 T wasregarded as saturation IRM (SIRM). Remanences were measuredwith a 2G-760 U-channel system. Temperature-dependence of thelow-field magnetic susceptibility curves of some samples wereconducted using a KLY-3 Kappabridge equipped with a CS-3 high-temperature furnace (sensitivity of 1 � 10�8 SI, AGICO Ltd., Brno,Czech Republic) in air atmosphere. All measurements were per-formed from room temperature up to 700 �C with a measurementinterval of 2 �C and a heating and cooling rate of about 9 �C perminute.

In addition, hysteresis loops of sediments from four samplingsites (LDR3, LDR11, LDR14 and LDR18) were measured at roomtemperature using a Model 3900 Micromag vibrating magnetom-eter. The saturation magnetization at 1 T (Ms), saturation rema-nence (Mrs, which is equal to SIRM), and the coercivity (Bc) wereobtained following subtraction of the paramagnetic contribution.Remanence coercivity (Bcr) was obtained by back-field demagne-tization curves. Low-temperature properties of representativesamples were measured using a Quantum Design Model XP-5Magnetic Properties Measurement System (MPMS XP-5, sensi-tivity 5.0� 10�10 Am2). Samples were cooled down from 300 to 5 Kin a zero magnetic field (ZFC), and saturation remanance acquiredin a 5-T field at 5 K (designated SIRM5 T@5 K hereafter) wasmeasured by warming from 5 to 300 K. To determine the SPcontents in these samples, temperature dependence of suscepti-bility was measured from 300 K to 5 K at frequencies of 1 Hz and1000 Hz. Frequency dependent cfd was calculated fromc1 Hz� c1000 Hz. All magnetic measurements were carried out in thePaleomagnetism and Geochronology Laboratory in Beijing.

Magnetic extracts were obtained from the middle stream riversediment (LDR1, depth of 8e12 cm) using a strong hand magnetsealed with a plastic bag. Afterward, the magnetic extracts werefixed by gum, and their surfaces were coatedwith gold for scanningelectron microscope (SEM). A LEO1450VP was used for SEMobservations of the morphology of the particles, and an INCAENERGY 300 energy dispersive X-ray spectrometer (EDX) wasapplied to determine elements and composition of the magneticextracts. The SEM and EDX studies were carried out in the labora-tories of the Institute of Geology and Geophysics Beijing.

Heavy metal (Be, V, Cr, Co, Ni, Cu, Zn, Rb, Mo, Cd, Cs, Ba, Nd, Pband Fe) analysis of river sediments from four sampling sites (LDR3,LDR11, LDR14, LDR18) was performed by inductively-coupledplasma-mass spectrometry (ICP-MS) using the DZ/T0223-2001method with HR-ICP-MS (Element I) Finningan MAT equipmentin the Analytical Laboratory of the Beijing Research Institute ofUranium Geology. Pearson’s correlation coefficient analysis, Prin-cipal Component Analysis (PCA) and Cluster Analysis (CA) wereperformed using the commercial statistics software package SPSSversion 13.0 (SPSS Inc.)

3. Results

3.1. Magnetic properties

The mean c values of surface river silts (0e2 cm) from the 19sampling sites show low values in the upstream region (typicallyless than c. 70 � 10�8 m3 kg�1); strongly enhanced intensities(c. 200e660 � 10�8 m3 kg�1) in the middle part near thesurroundings of the Fe-smelting plant and lower values again

Fig. 1. (a) Sketch map of the study area (Loudi) and sampling sites; (b) Histogram of mass-specific low-field magnetic susceptibility (c) mean values of surface samples; (c) Isolineplot of c values with a function of distance and depth. Hexagons (OL1, OL2, OL3) are the positions of wastewater outlets.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e3070 3059

further downstream (c. 150 � 10�8 m3 kg�1) (Fig. 1b). Lateral andvertical changes of c values along Lianshui River are more clearlydisplayed in Fig. 1c which plots the data as a function of distanceand depth. Maximum c values are located below 10 cm depth in themiddle section.

Vertical profiles of c, SIRM, Ms and cARM/c at four sample sites(LDR3, LDR11, LDR14, LDR18) are selected to identify the differencesamong upstream, middle and downstream sections (Fig. 2). Thesediment profile from LDR11 (middle stream section, polluted riversediments) exhibits the largest enhancement of c as its values arefour to sixteen times higher than those (<70 � 10�8 m3 kg�1) fromLDR3 (upstream section, unpolluted river sediments), especially, inthe depth of 10e20 cm. Sediment profiles from LDR14 and LDR18show c values higher than those from LDR3 and lower than thosefrom LDR11. In the downstream section LDR18, lower c values(100e300 � 10�8 m3 kg�1) appear again, however, higher thanLDR3. This indicates that these sediments accumulated a high inputof anthropogenic material particularly from the Fe-smelting plant

and the city. Most maximum values are found below 15e20 cmdepth. The variation trends of SIRM and Ms are similar to c

(Fig. 2a,b,c), and high positive Pearson correlation coefficients(P > 0.95) are found between each other, indicating that ferri-magnetic minerals are controlling c in these samples (Oldfield,1991). cARM/c values in profile of LDR3 are a little higher than inother profiles suggesting a higher portion of smaller single domain(SD) to small pseudo-single domain (PSD) particles in LDR3(Oldfield, 1991).

Narrow hysteresis loops that close below w250 mT (Fig. 3a)indicate that low coercivity ferrimagnetic minerals are dominant inthese river sediments. A Day-plot of hysteresis parameters of allsamples from sites LDR3, LDR11, LDR14 and LDR18 shows that allsamples locate within the PSD range (Day et al., 1977; Dunlop,2002) (Fig. 3b). In the Day diagram, it seems that LDR3(upstream) values plot more to the left than others indicatinga more consistent grain size in the less polluted LDR3. The distri-bution of middle stream samples (LDR11 and LDR14) is closer to the

Fig. 2. The vertical profiles of (a) c, (b) SIRM, (c) Ms and (d) cARM/c for all samples from LDR3, LDR11, LDR14 and LDR18, different symbols refer to different sampling sites.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e30703060

multidomain (MD) range than in the upstream (LDR3) and down-stream (LDR18) samples, suggesting that magnetic particles in themiddle-stream sediments have relatively coarse-grained PSD sizes.IRM acquisition and back-field demagnetization curves for theabove representative samples are shown in Fig. 3c. IRM acquisitioncurves rapidly reach saturation below 250 mT and are independentof sampling site whilst back-field demagnetization curves displaysoft behavior with Bcr values of 30e40 mT. All room temperaturemagnetic measurements indicate that magnetic mineralogy isdominated by low coercivity ferrimagnetic minerals.

Magnetic minerals in river sediments can be identified bytemperature-dependent susceptibility (c-T) cycles in air (Fig. 4).The increased susceptibility below 250 �C and the peaks around250e300 �C may indicate ultrafine particles reaching the SD tosuperparamagnetic (SP) transition at these temperatures (Ochesand Banerjee, 1996; Deng et al., 2004; Liu et al., 2005); the

susceptibility increases sharply to just below the Curie temperatureof magnetite (585 �C), and shows a Tc of w580 �C revealingmagnetite as the major contributor to c, in cooling curves c ishigher than in heating curves for temperatures < 580 �C, whichdemonstrates new formation of magnetite during heating (Zhanget al., 2010). Magnetic enhancement upon heating of LDR3 isobviously higher than others (Fig. 4).

Further proof of the presence of magnetite in surface and deepsamples is the clear Verwey transition atw120 K (Fig. 5a,d,g,j). Thebehavior of frequency-dependence of c0 between 5 and 300 K(Fig. 5b,c,e,f,h,i,k,l) shows that in-phase c0 is different for bothfrequencies above 120 K. This indicates that SP magnetite particlesare present in these river sediments (Kosterov, 2003). An obviouscfd peak between 5 and 80 K (Fig. 5e,f,h,j,l) could be caused byrelaxation of domain walls associated with MD magnetite particles(Moskowitz et al., 1993, 1998; Kosterov, 2003; Ba1anda et al., 2005)

Fig. 3. (a) Magnetic hysteresis loops for representative river sediments following before (dash line) and after (solid line) subtraction of the paramagnetic contribution; (b) Day-plotof the ratios Mrs/Ms and Bcr/Bc for all samples in LDR3, LDR11, LDR14 and LDR18, grain size boundaries for SDePSDeMD are according to Dunlop (2002); (c) IRM acquisition andback-field demagnetization curves of respective samples.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e3070 3061

and supporting the dominance of MD magnetite particles in themiddle (LDR11, LDR14) and downstream (LDR18) sections. Thesharp decrease in remanence and c during warming from 5 to 50 K(Fig. 5aec,j,k) may relate to the main presence of SP particles andparamagnetic minerals in the upstream (LDR3) and at the surface ofthe downstream (LDR18) sediments (Coey, 1988; Dunlop andÖzdemir, 1997).

Magnetic results (Figs. 3e5) show that PSDeMD magnetite isthe dominant ferrimagnetic phase in the middle stream and

Fig. 4. Temperature-dependence of magnetic susceptibility (ceT) heating (solid line) and cosampling sites. Each curve was normalized with its corresponding magnetic susceptibility

downstream sediments, especially at depths >20 cm, whereas, PSDwith SP magnetite is predominant in the upstream sediments; thisdifferences of magnetic mineral phases between different sectionsof Lianshui River shows that enhancements of magnetic concen-tration values are controlled by coarse magnetic particles in thesediments. One can conclude that anthropogenic magnetite addsa different grain size fraction to the natural magnetite contentwhich leads to a shift of data points toward “right-down” from therange occurring for uniform grain sizes in Day-plot (Fig. 3b).

oling (dashed line) curves of representative samples in LDR3, LDR11, LDR14 and LDR18at room temperature c0.

Fig. 5. Low-temperature magnetic measurements for representative samples in LDR3, LDR11, LDR14 and LDR18, respectively; (a, d, g and j) accompanying thermal warming ofSIRM5 T@5 K from 5 to 300 K and after cooling in zero field (ZFC) for surface (heavy solid line) and deep (dashed line) sediments, respectively; (b, c, e, f, h, i, k and l) Low-temperature-dependence of magnetic susceptibility at 1 Hz (light hollow line) and 1000 Hz (heavy solid line) curves (c0eT) and low-temperature cfdeT curves (dashed line),respectively.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e30703062

3.2. SEM and EDX

Samples from 8 to 12 cm depths at LDR11 (representing therange c of highest anthropogenic input according to maximum c)have been observed and analyzed by SEM and EDX. Results areshown in Fig. 6; magnetic spherules with diameters of 9e140 mm

(Fig. 6aec) and irregular-shaped particles (Fig. 6d) are observed andidentified as iron-oxides by EDX analysis. The EDX of spherulessurfaces show dominant peaks of Fe, along with smaller peaks of Sior Al (Fig. 6aec, spectrum 1), whereas the EDX of broken sides onspherules show that Si and Al are predominant with small amountsof Fe, Mg, K and Ca etc (Fig. 6bec, spectrum 2). The irregular-shaped

Fig. 6. SEM images and EDX analysis results of extractor from 8 to 12 cm depths at LDR11.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e3070 3063

particles (Fig. 6a, d, spectrum 2) have variable Fe contents (13e70%)with different contents of Si, Al, Mg, Ca, Mn, K etc, and smallamounts of Zn are also present in some particles (Fig. 6d). SEM andEDX results from the magnetic spherules are consistent withprevious studies of fly ashes and smelting slag (Hoffmann et al.,1999; Xie et al., 2001; Goddu et al., 2004; Jordanova et al., 2004;Gautam et al., 2005; Zhang et al., 2006; Blaha et al., 2008b; Kimet al., 2009; Chaparro et al., 2010; Huliselan et al., 2010;Rosowiecka and Nawrocki, 2010).

3.3. Heavy metals concentrations

Concentrations of 15 heavy metals and their mean, standarddeviation, minimum and maximum values in different sectionsof the Lianshui River are given in Table 1. The middle streamsection LDR11 contains the highest mean concentrations: Fe(74.19 g/kg), Pb (831 mg/kg), Zn (2528 mg/kg), Cu (84 mg/kg), Cd(13 mg/kg), V (178 mg/kg), Mo (3 mg/kg) and Cr (122 mg/kg).The lowest mean concentrations of Fe (40 g/kg), Pb (58 mg/kg),

Table 1Summary of heavy metals in river sediments from four different sections.

Heavy metals LDR3(upstream)(20)

LDR11(middle stream)(19)

LDR14(middle stream)(13)

LDR18(downstream)(12)

All (64)

Fe (g/kg) Range 34.06e44.97 62.31e92.62 56.67e68.91 43.84e55.96 34.06e92.62Mean � Sd 40.08 � 2.72 74.19 � 7.97 61.42 � 4.39 50.96 � 4.15 56.58 � 14.76

Pb (mg/kg) Range 46e73 360e1735 246e821 95e294 46e1735Mean � Sd 58 � 8 831 � 450 546 � 183 194 � 59 412 � 413

Zn (mg/kg) Range 199e400 958e5349 742e2182 426e928 199e5349Mean � Sd 284 � 52 2528 � 1461 1613 � 468 704 � 147 1299 � 1236

Cu (mg/kg) Range 42e66 63e99 63e99 61e88 42e99Mean � Sd 50 � 6 84 � 9 79 � 12 75 � 9 71 � 17

Ni (mg/kg) Range 100e304 57e100 61e94 70e120 57e304Mean � Sd 148 � 47 76 � 11 79 � 9 93.65 � 16.93 102 � 42

V (mg/kg) Range 98e146 109e282 97e139 120e158 97e282Mean � Sd 115 � 11 178 � 52 118 � 13 137 � 12 138 � 40

Co (mg/kg) Range 51e103 21e70 33e56 30e88 21e103Mean � Sd 73 � 13 38 � 11 45 � 7 60 � 17 55 � 19

Rb (mg/kg) Range 72e99 60e85 74e119 92e136 60e136Mean � Sd 88 � 8 73 � 7 91 � 11 112 � 12 89 � 16

Mo (mg/kg) Range 1.3e2.2 2.1e3.4 1.7e2.6 1.2e2.9 1.2e3.4Mean � Sd 1.7 � 0.2 2.8 � 0.4 2.0 � 0.3 2.1 � 0.6 2.2 � 0.6

Cd (mg/kg) Range 1.9e25 7.0e23.7 4.52e14.60 2.50e6.82 1.85e25.20Mean � Sd 7.4 � 5.6 13.0 � 3.7 9.73 � 3.26 5.11 � 1.33 9.11 � 4.92

Cs (mg/kg) Range 7.7e13.0 10.5e17.4 12.4e19.8 10.9e21.5 7.7e21.5Mean � Sd 10.4 � 1.4 13.8 � 2.1 16.2 � 2.5 16.9 � 3.4 13.8 � 3.4

Ba (mg/kg) Range 361e808 306e464 355e575 270e486 270e808Mean � Sd 464 � 102 394 � 44 413 � 67 382 � 62 418 � 79

Nd (mg/kg) Range 44e61 42e73 40e64 51e70 40e73Mean � Sd 53 � 5 50 � 7 50 � 7 60 � 6 53 � 7

Be (mg/kg) Range 1.9e4.2 2.34e7.3 3.1e6.4 3.7e6.7 1.9e7.3Mean � Sd 2.9 � 0.7 4.2 � 1.0 4.2 � 0.8 4.9 � 0.9 3.9 � 1.1

Cr (mg/kg) Range 65e86 80e160 71e100 80e124 65e160Mean � Sd 73 � 6 121 � 22 85 � 7 99 � 13 95 � 24

PLI Range 1.85e2.58 3.20e4.06 2.45e3.74 2.19e3.18 1.85e4.06Mean � Sd 2.22 � 0.21 3.46 � 0.24 3.14 � 0.40 2.84 � 0.31 2.89 � 0.58

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e30703064

Zn (284 mg/kg), Cu (50 mg/kg), V (115 mg/kg), Mo (1.7 mg/kg)and Cr (73 mg/kg) are present in the upstream section LDR3.Mean concentrations of Ni, Co and Ba in LDR3 are higher than inother sampling sites (Table 1). The highest mean concentrationsof other heavy metals (Rb, Cs, Nd, Be) in Table 1 are present inLDR18.

The lateral and vertical changes of typical heavy metals (Fe, Pb,Zn, Cu, Ba, Nd) along the Lianshui River, with sufficient potentialto reveal possible pollution sources, are given in Fig. 7 andFig. 8, respectively. In upstream (LDR3) to downstream (LDR18),concentrations of Fe, Pb, Zn and Cu are observed in the followingorder: LDR11 > LDR14 > LDR18 > LDR3, however, obviousdifferences are not found in concentrations of Ba and Nd (Fig. 7).Concentrations of Fe, Pb, Zn in the vertical profiles show the sameorder with lateral characteristics when depths of less than 20 cmare considered. However, two high concentration peaks of Pb andZn are found at w6 cm and w12 cm in the sediment profileLDR11; when depths >20 cm are considered, concentrations of Pband Zn are more stable in LDR11 although they are still higherthan those in LDR3 (Fig. 8aec). From the top to 22 cm, concen-trations of Cu in LDR11, LDR14 and LDR18 are similar to each otherand are all higher than in LDR3, increasing gradually with depth(Fig. 8d). In contrast, concentrations of Ba and Nb in differentsampling sites seem to be more stable and are essentially inde-pendent of the depth (Fig. 8e,f).

The pollution load index (PLI) proposed by Tomlinson et al.(1980) has been used obtained from concentration factorsCF ¼ Cmetal/Cbackgroud value for selected metals. The PLI is thencalculated by the n-root from the product of the n CFs of themetals included: PLI ¼ nO(CF1 � CF2 � CF3 � .CFn) (Angulo,

1996); the lowest concentration value for each element wasused as background value Cbackgroud value. According to Singh et al.(2003) PLI values vary from 0 (unpolluted) to 10 (highly polluted)as follows: PLI ¼ 0 background concentration; 0 < PLI � 1unpolluted; 1 < PLI � 2 moderately to unpolluted; 2 < PLI � 3moderately polluted; 3 < PLI � 4 moderately to highly polluted;4 < PLI � 5 highly polluted; PLI > 5 very highly polluted. As shownin Table 1, including all 15 heavy metals, the highest mean value is3.46 in LDR11, followed by LDR14 (3.14), LDR18 (2.84) and LDR3(2.22). A more detailed discussion of the PLI index is given insection 4.

3.4. Correlations of heavy metal concentrationsand magnetic parameters

Pearson’s correlation coefficients (P) between heavy metalconcentrations andmagnetic parameters are listed in Table 2.Ms, c,SIRM and cARM have significant positive correlations with V, Cr, Cu,Zn, Mo, Cd, Pb, Fe and PLI (of all 15 metals). They show significantnegative correlationwith Co, Ni, Rb and Nd. However, no significantcorrelations between magnetic parameters and Be as well as Cs arefound; c and SIRM correlate negatively with Ba (Table 2). Since theP between SIRM and heavy metals are highest among all themagnetic parameters, we select SIRM to quantify the relationshipbetween magnetic parameters and heavy metals. The scatter plotsof SIRM versus Fe, Pb, Zn, Cu, Ba and Nd are given in Fig. 9 includingthe coefficient of determination R2 values. All heavy metals exceptBa (R2 ¼ 0.10) and Nd (R2 ¼ 0.09) show a good correlation withSIRM i.e., Fe (R2 ¼ 0.90), Pb (R2 ¼ 0.88), Zn (R2 ¼ 0.86), Cu

Fig. 7. Box plots of (a) Fe, (b) Pb, (c) Zn, (d) Cu, (e) Ba and (f) Nb for all samples from LDR3, LDR11, LDR14 and LDR18.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e3070 3065

(R2 ¼ 0.71). The differences in the correlations between magneticparameters and heavymetals are likely dependent on their sources.

Principal component analysis (PCA) was applied to identify thesource of heavy metals in the river sediments by applying Varimaxrotation with Kaiser normalization (Wu, 2004). The results showthat there are four factors that explain 85.6% of the total variancein the river sediments (Table 3). As shown in Table 3, the firstfactor explains 34.9% of the total variance and is dominated by Ms,c, SIRM, cARM, V, Cr, Mo and Fe; Factor 2 accounts for 26.4% of thetotal variance and loads in Cu, Zn, Cd and Pb; Factor 3 is domi-nated by Be, Rb, Cs and Nd, and accounts for 14.8% of the totalvariance; Factor 4 is dominated by Co, Ni and Ba and accounts for9.5% of the total variance. The relationships between the heavymetals and the magnetic parameters based on the first threeprinciple components of the sediments are illustrated in Fig. 10a in3D space.

Cluster analysis (CA) was performed to evaluate further thesimilarities of the heavy metal sources by applying Ward’s method(Wu, 2004). The CA results for magnetic parameters and heavymetals are shown in Fig. 10b as a dendrogram. Fig. 10b displays fourclusters: (I) c, SIRM, Ms, Fe, cARM, V, Cr, Mo; (II) Zn, Pb, Cd; (III) Cu;(IV) Be, Cs, Rb, Nd, Co, Ni, Ba. As shown in this figure the clusters I, IIand III join together at a relatively high level, likely implyinganthropogenic activities, however, they differ from cluster IV. TheCA results are in agreement with the PCA results listed in Table 3and illustrated in Fig. 10a.

4. Discussion

4.1. Sources of magnetic minerals and heavy metals

Magnetic, SEM and EDX results (Figs. 3e6) reveal that PSDeMDmagnetite often occurring as spherules is the dominant ferrimag-netic phase with small amounts of SP magnetic particles in themiddle stream and downstream sediments (at depth <20 cm);large particles input obviously accounts for the high c values inthose sections. In contrast, PSD and SP magnetite particles arepresent in the upstream sediments corresponding to lower c

values. The SP magnetite particles are likely to come from thecontribution of pedogenic processes in this subtropical region (Lu,2000).

The heavy metals can be separated into two groups based onthe multivariate statistical analysis results (Tables 2 and 3,Fig. 10). Group I (including Be, Cs, Rb, Nd, Co, Ni, Ba) revealsneither correlation with magnetic parameters nor any obviousdifferences between upstream, middle stream and downstreamsections (Table 2, Figs. 7e10). Noting that the main source of Rb,Cs, Ba, Nd and Be etc is from soil (Pan and Yang, 1988; Wang andWei, 1995), we infer that the contents of these heavy metalscomes from pedogenic sources in the catchment region upstreamof Loudi city. Group II includes Fe, V, Cr, Mo, Zn, Pb, Cd and Cuwhich show significant correlations with magnetic concentrationparamenters Ms, c, SIRM and cARM suggesting that magnetic

Fig. 8. Vertical profiles of (a) Fe, (b) Pb, (c) Zn, (d) Cu, (e) Ba and (f) Nb for all samples from LDR3, LDR11, LDR14 and LDR18. The different symbols refer to different sampling sites.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e30703066

particle contents and these heavy metals stem from anthropo-genic activities. Zn, Cd and Cu are common in iron ores (Wongand Au, 1984) and Mo, V and Cr are typically used as additivesin the Fe-smelting industry (Farmaki and Thomaidis, 2008); Pband Zn are used in steel wire heat-treatment (Kientzl andDobránszky, 2007). Each of these heavy metals could enter intothe environment with the emission of slag, dust and wastewater

Table 2Pearson’s correlations matrix for the heavy metals concentrations and magneticparameters (n ¼ 64).

PearsonCorrelation (P)

Ms clf SIRM cARM

Be 0.155 0.186 0.184 0.174V 0.697a 0.679a 0.669a 0.706a

Cr 0.800a 0.785a 0.789a 0.721a

Co �0.728a �0.763a �0.767a �0.564a

Ni �0.559a �0.601a �0.601a �0.460a\

Cu 0.652a 0.716a 0.739a 0.557a

Zn 0.684a 0.708a 0.721a 0.428a

Rb �0.568a �0.588a �0.570a �0.470a

Mo 0.799a 0.776a 0.783a 0.681a

Cd 0.522a 0.532a 0.543a 0.417a

Cs 0.144 0.203 0.235 0.063Ba �0.243 �0.262b �0.254 b �0.110Nd �0.407a �0.419a �0.412a �0.297 b

Pb 0.721a 0.755a 0.770a 0.483a

Fe 0.924a 0.933a 0.934a 0.748a

PLI 0.741a 0.782a 0.799a 0.615a

a Correlation is significant at the 0.01 level (2-tailed).b Correlation is significant at the 0.05 level (2-tailed).

by the Fe-smelting activities in Loudi. The magnetic properties(Figs. 3e5) and SEM and EDX (Fig. 6) results reveal that most ofmagnetic spherules with larger dimension (9e140 mm) at depthsof 8e12 cm stem from fly ash of the coal-fired Fe-smelting plant.Maximum values of heavy metals in Group II at depths of8e14 cm (Fig. 8), suggest that most of the spherical particles andthe heavy metals Fe, V, Cr, Mo, Pb, Zn, Cd and Cu are concomitantand may originate from Fe-smelting plant.

4.2. Environmental implications of magnetic parametersin river sediments

As mentioned above Group II heavy metals (Fe, V, Cr, Mo, Zn, Pb,Cd, Cu) show a good positive correlation withMs, c, SIRM and cARMand suggest that these magnetic concentration parameters areefficient indicators of the anthropogenic heavy metals input intothe river sediments. Among all the magnetic concentrationparameters, correlation coefficients between SIRM and heavymetals are highest (Table 2), moreover SIRM values exclude thecontribution of pedogenic SP particles. Hence SIRM is considered asthe optimum proxy parameter for the detection of heavy metalpollution in the river sediments of Loudi.

To evaluate accurately the contamination degree of anthropo-genic activities-related heavy metals, the Tomlinson PLI includingFe, V, Cr, Mo, Zn, Pb, Cd and Cu was calculated for all samples andtermed as PLIanthro. We note that the correlation coefficientbetween the PLIanthro and magnetic parameters (P ¼ 0.869 for Ms,P¼ 0.900 for c, P¼ 0.912 for SIRM and P¼ 0.679 for cARM) is higherthan for the PLI of all heavymetals. Scatter plots of log(SIRM) versus

Fig. 9. Relationship between SIRM and (a) Fe, (b) Pb, (c) Zn, (d) Cu, (e) Ba and (f) Nb for all samples in LDR3, LDR11, LDR14 and LDR18. Equations and coefficient of determination (R2)between SIRM and heavy metals contents are listed on the figures.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e3070 3067

PLI and PLIanthro are shown in Fig. 11; the coefficient of determi-nation (R2 ¼ 0.83) with PLIanthro is higher than with PLI (R2 ¼ 0.77)based on a linear equation (Fig. 11); standard errors of linearregression are �0.09 and �0.68, respectively. It reinforces thefinding that the magnetic concentration parameter log(SIRM) isreasonably proportional to the concentration of the sum of Group IIheavy metals.

Table 3Rotated Component Matrix (a) for data of typical river sediments (PCA loading >0.4are shown in bold).

Components

1 2 3 4

Ms 0.82 0.48 �0.17 �0.20c 0.80 0.52 �0.14 �0.24SIRM 0.7 0.54 �0.12 �0.22cARM 0.83 0.26 �0.15 �0.12Be 0.30 �0.05 0.76 �0.27V 0.94 �0.07 0.05 0.06Cr 0.90 0.17 0.18 �0.12Co �0.41 �0.65 0.07 0.42Ni �0.31 �0.53 �0.25 0.56Cu 0.57 0.54 0.50 �0.16Zn 0.25 0.89 �0.04 �0.16Rb �0.37 �0.25 0.81 0.16Mo 0.84 0.33 0.20 0.12Cd 0.25 0.74 �0.09 0.13Cs 0.01 0.51 0.76 �0.14Ba �0.03 �0.08 �0.01 0.91Nd 0.00 �0.52 0.70 0.23Pb 0.30 0.89 �0.06 �0.16Fe 0.76 0.55 0.10 �0.18Eigenvalues 6.6 5.0 2.8 1.8% of Variance 34.9 26.4 14.8 9.5Cumulative % 34.9 61.3 76.1 85.6

Extraction Method: Principal Component Analysis, Rotation Method: Varimax withKaiser Normalization.

a Rotation converged in 5 iterations.

The degree of heavy metal pollution is clearly outlined as anisoline plot of the PLIanthro values showing the data as a function ofdistance and depth (Fig. 12). PLIanthro values are in the range of 0e2before the river reaches the city identifying the upstream section asunpolluted to moderately polluted. The PLIanthro values increasegradually from 2 to 5 after entering into the city indicating that theriver is moderately to highly polluted by anthropogenic activityhere. The PLIanthro reach values vary from 5 to 6 near and beyondthe Fe-smelting plant and increase to almost 8 at a distanceof w2.5 km (LDR11) downstream from the plant. It then againdecreases gradually to w5 near to the city boundary (LDR15); thisshows that the river is highly polluted in this city region. PLIanthrovalues of w5e2.5 are sustained within a distance of w7 km afterthe river leaves the city indicating that this downstream section ofthe river remains moderately to highly polluted.

It is also interesting to note that the highest PLIanthro values arenot found directly at the Fe-smelting plant (sites LDR7eLDR9), butare present in the bend of the river (LDR11) within a distanceof w2.5 km to the plant; at the same time, the highest valueappears at a depth of w10e20 cm and not at the surface (Fig. 12).This is likely to result from the change of the hydraulic flowconditions along this section of the channel: pollutants will beretained in suspension where the width of the river is w80 m andthe course is relatively straight as it is at the plant. After site LDR10where the river makes a strong bend and the width is expandingto w150 m (LDR11), the flow velocity decreases and the solidparticles precipitate from the suspension and deposit by helicalwater flow. Due to rapid flow and three wastewater outlets onlylittle sediment is present directly at the plant (with the exception ofgravels in the vicinity of LDR7eLDR9). In contrast sediment accu-mulation is high after the strong river bend and long sedimentcores (such as LDR11) can be recovered here. High flow velocitiyleads to strong fluvial abrasion at the boundary between the waterand the riverbedwhich can remove and transport coarser materialslike the large anthropogenic 9e140 mm spherules. This coarser

Fig. 10. (a) PCA results of LDR3, LDR11, LDR14 and LDR18 samples in the three-dimensional space: plots of loading of the first three principal components (different symbols meandifferent cluster-groups); (b) Dendrogram results of all samples from Ward method of hierarchical cluster analysis for 15 elements and 4 magnetic parameters.

Fig. 11. The relationship between (a) SIRM and PLI, (b)SIRM and PLIanthro for all samples in LDR3, LDR11, LDR14 and LDR18. Equations and coefficient of determination (R2) betweenlog(SIRM) and PLI, PLIanthro are listed on the Figure. PLIanthro refers to the Tomlinson PLI of Fe, V, Cr, Mo, Zn, Pb, Cd and Cu in river sediments originating from anthropogenic activities.

Fig. 12. Isoline plot of PLIanthro values as a function of distance and depth. PLIanthro ¼ 0: background concentration; 0 < PLIanthro � 1: unpolluted; 1 < PLIanthro � 2: moderately tounpolluted; 2< PLIanthro� 3:moderately polluted; 3< PLIanthro� 4:moderately to highly polluted; 4< PLIanthro� 5: highly polluted; PLIanthro> 5: very highly polluted (Singh et al., 2003).

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e30703068

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e3070 3069

fraction then precipitates after the river bend where the flowvelocity becomes lower, whilst the fine particles are transportedfurther downstream.

5. Conclusions

We have shown that the coexistence of magnetic particles andheavy metals with significant correlations between them, make itpossible to use magnetic methods as a tool for assessment of heavymetal pollution in industrial regions by surface water runoff.Specifically we find that the enrichment of coarse PSDeMDmagnetite particles is the main reason for enhancement ofmagnetic concentration parameters (Ms, c, SIRM, cARM) in the riversection downstream of a Fe-smelting plant. Antrhopogenic spher-ules are found by SEM and most likely originate from the plant. Thecorrelation of elevated c and SIRM values with heavy metal (Fe, V,Cr, Mo, Zn, Pb, Cd, Cu) concentrations from industrial activities(mainly the Fe-smelting plant) prove that magnetic concentrationparameters can be used as proxies for the anthropogenic contri-bution of pollutants to the river sediments. In contrast, very weakcorrelations with Be, Cs, Rb, Nd, Co, Ni and Ba, indicate that thesemetals stem from soil in the catchment region. The Tomlinsonpollution load index (PLI) shows significant correlations withmagnetic concentration parameters; most notably PLIanthro(including Fe, V, Cr, Mo, Zn, Pb, Cd, Cu) shows a strong positivecorrelation with SIRM values and a linear relationship with log(-SIRM) can be established. Therefore, for the studied river sedimentsat Loudi, it can be concluded that log(SIRM) mapping is best suit-able for quantitative assessment of heavy metal pollution. Thedegree and range of Lianshui River sediment pollution is controlledby surface water transport and deposition by precipitation of solidparticles from suspension. Most of earlier works on magneticproxies have considered pollution of soil and sediments by atmo-spheric input of fly ash. Our results demonstrate that also fortransport by surface water the use of environmental magneticmethods in conjunction with auxiliary geochemical analysis canprovide a fast and non-destructive tool for assessment of heavymetals pollution.

Acknowledgments

Qingsong Liu and two reviewers (Erwin Appel and Eduard Pet-rovský) are gratefully acknowledged for their constructivesuggestions. This work was financially supported by the NationalNature Science Foundation of China (Grants 40804014, 20677059,40525013 and 40821091).

References

Al-Khashman, O.A., 2004. Heavy metal distribution in dust, street dust and soilsfrom the work place in Karak Industrial Estate, Jordan. Atmospheric Environ-ment 38, 6803e6812.

Angulo, E., 1996. The Tomlinson pollution load index applied to heavy metal“Mussel-Watch” data: a useful index to assess coastal pollution. The Science ofthe Total Environment 187, 19e56.

Awofolu, O.R., Mbolekwa, Z., Mtshemla, V., Fatoki, O.S., 2005. Levels of trace metalsin water and sediments from Tyume River and its effects on an irrigatedfarmland. International Journal of Environmental Science and Technology 7 (3),435e446.

Ba1anda, M., Wieche�c, A., Kim, D., Kakol, Z., Koz1owski, A., Niedziela, P., Sabol, J.,Tarnawski, Z., Honig, J.M., 2005. Magnetic AC susceptibility of stoichiometricand low zinc doped magnetite single crystals. European Physical Journal B 43,201e212.

Banerjee, A.D.K., 2003. Heavy metal levels and solid phase speciation in street dustsof Delhi, India. Environmental Pollution 123, 95e105.

Bijaksana, S., Huliselan, E.K., 2010. Magnetic properties and heavy metal content ofsanitary leachate sludge in two landfill sites near Bandung, Indonesia. Envi-ronmental Earth Science 60, 409e419.

Birch, L., Hanselmann, K.W., Bachofen, R., 1996. Heavy metal conservation in LakeCadagno sediments: historical records of anthropogenic emissions in a mer-omictic alpine lake. Water Research 30 (3), 679e687.

Blaha, U., Appel, E., Stanjek, H., 2008a. Determination of anthropogenic boundarydepth in industrially polluted soil and semi-quantification of heavy metal loadsusing magnetic susceptibility. Environmental Pollution 156, 278e289.

Blaha, U., Sapkota, B., Appel, E., Stanjek, H., Rösler, W., 2008b. Micro-scale grain-sizeanalysis and magnetic properties of coal-fired power plant fly ash and itsrelevance for environmental magnetic pollution studies. Atmospheric Envi-ronment 42, 8359e8370.

Bu�cko, M.S., Magiera, T., Pesoen, L.J., Janus, B., 2010. Magnetic, geochemical, andmicrostructural characteristics of road dust on roadsides with different trafficvolumes-case study from Finland. Water Air Soil Pollution 209, 295e306.

Canbay, M., Aydin, A., Kurtulus, C., 2010. Magnetic susceptibility and heavy-metalcontamination in topsoils along the Izmit Gulf coastal area and IZAYTAS(Turkey). Journal of Applied Geophysics 70, 46e57.

Chan, L.S., Ng, S.L., Davis, A.M., Yim, W.W.S., Yeung, C.H., 2001. Magnetic propertiesand heavy-metal contents of contaminated seabed sediments of Penny’s bay,Hong Kong. Marine Pollution Bulletin 42 (7), 569e583.

Chaparro, M.A.E., Bidegain, J.C., Sinito, A.M., Jurado, S.S., Gogorza, C.S.G., 2004.Relevant magnetic parameters and heavy metals from relatively pollutedstream sediments e vertical and longitudinal distribution along a cross-citystream in Buenos Aires Province, Argentina. Studia Geophysica et Geodaetica48 (3), 615e636.

Chaparro, M.A.E., Chaparro, M.A.E., Marinelli, C., Sinito, A.M., 2008. Multivariatetechniques as alternative statistical tools applied to magnetic proxies forpollution: a case study from Argentina and Antarctica. Environmental Geology54, 365e371.

Chaparro, M.A.E., Marié, D.C., Gogorza, C.S.G., Navas, A., Sinito, A.M., 2010. Magneticstudies and scanning electron microscopy e X-ray energy dispersive spec-troscopy analyses of road sediments, soils and vehicle-derived emissions. Stu-dia Geophysica et Geodaetica 54, 633e650.

Coey, J.M.D., 1988. Magnetic Properties of Soil Iron Oxides and Clay Minerals. ReidelPublishing, Dordrecht.

Day, R., Fuller, M., Schmid, V.A., 1977. Hystersis properties of titanomagnetites: grainsize and compositional dependence. Physics of the Earth and Planetary Interiors13, 260e267.

Deng, C.L., Zhu, R.X., Verosub, K.L., Singer, M.J., Vidic, N.J., 2004. Mineral magneticproperties of loess/paleosol couplets of the central loess plateau of China overthe last 1.2 Myr. Journal of Geophysical Research 109 (B01103), 1e13.

Duan, X.M., Hu, S.Y., Yan, H.T., Blaha, U., Rösler, W., Appel, E., Sun, W.H., 2010.Relationship between magnetic parameters and heavy element contents ofarable soil around a steel company. Nanjing. Science China (Earth Sciences) 53(3), 411e418.

Dunlop, D.J., 2002. Theory and application of the Day plot (Mrs/Ms versus Hcr/Hc) 1.Theoretical curves and tests using titanomagnetite data. Journal of GeophysicalResearch 107 (B3), 2056. doi:10.1029/2001JB000486.

Dunlop, D.J., Özdemir, Ö, 1997. Rock Magnetism: Fundamentals and Frontiers.Cambridge University Press, Cambridge, UK.

Farmaki, E.G., Thomaidis, N.S., 2008. Current status of the metal pollution of theenvironment of Greece e a review. Global NEST Journal 10 (3), 366e375.

Gautam, P., Blaha, U., Appel, E., 2005. Magnetic susceptibility of dust-loaded leavesas a proxy of traffic-related heavy metal pollution in Kathmandu city, Nepal.Atmospheric Environment 39 (12), 2201e2211.

Gautam, P., Blaha, U., Appel, E., Neupane, G., 2004. Environmental magneticapproach towards the quantification of pollution in Kathmandu urban area,Nepal. Physics and Chemistry of the Earth 29 (13e14), 973e984.

Goddu, S.R., Appel, E., Jordanova, D., Wehland, F., 2004. Magnetic properties of roaddust from Visakhapatnam (India) e relationship to industrial pollution and roadtraffic. Physics and Chemistry of the Earth 29 (13e14), 985e995.

Hay, K.L., Dearing, J.A., Baban, S.M.J., Loveland, P., 1997. A preliminary attempt toidentify atmospherically-derived pollution particles in English topsoils frommagnetic susceptibility measurements. Physics Chemistry of the Earth 22(1e2), 207e210.

Heller, F., Strzyszcz, Z., Magiera, T., 1998. Magnetic record of industrial pollution inforest soils of Upper Silesia, Poland. Journal of Geophysical Research 103 (B8),17767e17774.

Hoffmann, V., Knab, M., Appel, E., 1999. Magnetic susceptibility mapping of roadsidepollution. Journal of Geochemical Exploration 66 (1e2), 313e326.

Huliselan, E.K., Bijaksana, S., Srigutomo, W., Kardena, E., 2010. Scanning electronmicroscopy and magnetic characterization of iron oxides in solid waste landfillleachate. Journal of Hazardous Materials 179, 701e708.

Jordanova, D.V., Hoffmann, V., Fehr, K.T., 2004. Mineral magnetic characterization ofanthropogenic magnetic phases in the Danube river sediments (Bulgarian part).Earth and Planetary Science Letters 221, 71e89.

Jordanova, N.V., Jordanova, D.V., Veneva, L., Yorova, K., Petrovský, E., 2003. Magneticresponse of soils and vegetation to heavy metal pollution e A case study.Environmental Science & Technology 37 (19), 4417e4424.

Kapi�cka, A., Jordanova, N.V., Petrovský, E., Ustjak, S., 2001. Effect of different soilconditions on magnetic parameters of power-plant fly ashes. Journal of AppliedGeophysics 48 (2), 93e102.

Kapi�cka, A., Petrovský, E., Fialová, H.P., Podrázská, V., Dvo�rák, I., 2008. Highresolution mapping of anthropogenic pollution in the Giant MountainsNational Park using soil magnetometry. Studia Geophysica et Geodaetica 52,271e284.

C. Zhang et al. / Environmental Pollution 159 (2011) 3057e30703070

Kapi�cka, A., Petrovský, E., Ustjak, S., Machá�cková, K., 1999. Proxy mapping of fly-ashpollution of soils around a coal-burning power plant a case study in the CzechRepublic. Journal of Geochemical Exploration 66, 291e297.

Kientzl, I., Dobránszky, J., 2007. Production and examination of double composites.Materials Science Forum 537e538, 191e197.

Kim, W., Doh, S.J., Park, Y.H., Yun, S.T., 2007. Two-year magnetic monitoring inconjunction with geochemical and electron microscopic data of roadside dustin Seoul, Korea. Atmospheric Environment 41, 7627e7641.

Kim, W., Doh, S.J., Yu, Y.J., 2009. Anthropogenic contribution of magnetic particu-lates in urban roadside dust. Atmospheric Environment 43, 3137e3144.

Kosterov, A., 2003. Low-temperature magnetization and AC susceptibility ofmagnetite: effect of thermomagnetic history. Geophysical Journal International154 (1), 58e71.

Linnik, P.M., Zubenko, I.B., 2000. Role of bottom sediments in the secondarypollution of aquatic environments by heavy metal compounds, lakes andreservoirs. Lake & Reservoirs: Research & Management 5 (1), 11e21.

Liu, Q.S., Deng, C.L., Yu, Y.J., Torrent, J., Jackson, M.J., Banerjee, S.K., Zhu, R.X., 2005.Temperature dependence of magnetic susceptibility in an argon environment:implications for pedogenesis of Chinese loess palaeosols. Geophysical JournalInternational 161, 102e112.

Lu, S.G., 2000. Magnetic properties of subtropical ferrisols and its magneticmineralogical study. Chinese Journal of Geophysics 43 (4), 498e504.

Lu, S.G., Bai, S.Q., 2006. Study on the correlation of magnetic properties and heavymetals content in unban soils of Hangzhou City, China. Journal of AppliedGeophysics 60, 1e12.

Lu, S.G., Bai, S.Q., Xue, Q.F., 2007. Magnetic properties as indicators of heavy metalspollution in urban topsoils: a case study from the city of Luoyang, China.Geophysical Journal International 171, 568e580.

Lu, S.G., Wang, H.Y., Bai, S.Q., 2009a. Heavy metal contents and magnetic suscep-tibility of soils along an urban-rural gradient in rapidly growing city of easternChina. Environmental and Monitoring and Assessment 155, 91e101.

Lu, X.W., Wang, L.J., Lei, K., Huang, J., Zhai, Y.X., 2009b. Contamination assessment ofcopper, lead, zinc, manganese and nickel in street dust of Baoji, NW China.Journal of Hazardous Materials 161, 1058e1062.

Lwanga, M.S., Kansiime, F., Denny, P., Scullion, J., 2003. Heavy metals in Lake George,Uganda with relation to metal concentrations in tissues of common fish specie.Hydrobiologia 499 (1e3), 83e93.

Maher, B.A., Moore, C., Matzka, J., 2008. Spatial variation in vehicle-derived metalpollution identified by magnetic and elemental analysis of roadside tree leaves.Atmospheric Environment 42, 364e373.

Mohiuddin, K.M., Zakir, H.M., Otomo, K., Sharmin, S., Shikazono, N., 2010.Geochemical distribution of trace metal pollutants in water and sediments ofdownstream of an urban river. International Journal of Environmental Scienceand Technology 7 (1), 17e28.

Morris, W.A., Versteeg, J.K., Bryant, D.W., Legzdins, A.E., Mccarry, B.E., Marvin, C.H.,1995. Preliminarycomparisons betweenmutagenicity andmagnetic susceptibilityof respirable airborne particulate. Atmospheric Environment 29 (23), 3441e3450.

Moskowitz, B.M., Frankel, R.B., Bazylinski, D.A., 1993. Rock in magnetic criteria forthe detection of biogeic magnetite. Earth and Planetary Science Letters 120(3e4), 283e300.

Moskowitz, B.M., Jackson, M., Kissel, C., 1998. Low-temperature magnetic behaviorof titanomagnetites. Earth and Planetary Science Letters 157 (3e4), 141e149.

Oches, E.A., Banerjee, S.K., 1996. Rock-magnetic proxies of climate change fromloess-paleosol sediments of the Czech Republic. Studia Geophysica et Geo-daetica 40, 287e300.

Okafor, E.C., Opuene, K., 2007. Preliminary assessment of trace metals and polycyclicaromatic hydrocarbons in the sediments. International Journal of Environ-mental Science and Technology 4 (2), 233e240.

Oldfield, F., 1991. Environmental magnetism-A personal perspective. QuaternaryScience Reviews 10, 73e85.

Pan, Y.M., Yang, Z.G., 1988. The Background Value of Soil in Hunan and ResearchMethods. China Environmental Science Press, Beijing.

Petrovský, E., Kapi�cka, A., Jordanova, N., Bor�uvka, L., 2001. Magnetic properties ofalluvial soils contaminated with lead, zinc and cadmium. Journal of AppliedGeophysics 48, 127e136.

Petrovský, E., Kapi�cka, A., Jordanova, N., Knab, M., Hoffmann, V., 2000. Low-fieldmagnetic susceptibility: a proxy method of estimating increased pollutionof different environmental systems. Environmental Geology 39 (3e4),312e318.

Rijal, M.L., Appel, E., Petrovský, E., Blaha, U., 2010. Change of magnetic propertiesdue to fluctuations of hydrocarbon contaminated groundwater in unconsoli-dated sediments. Environmental Pollution 158, 1756e1762.

Rosowiecka, O., Nawrocki, J., 2010. Assessment of soils pollution extent insurroundings of ironworks based on magnetic analysis. Studia Geophysica etGeodaetica 54, 185e194.

Sekabira, K., Oryem Origa, H., Basamba, T.A., Mutumba, G., Kakudidi, E., 2010.Assessment of heavy metal pollution in the urban stream sediments and itstributaries. International Journal of Environmental Science and Technology 7(4), 435e446.

Shu, J., Dearing, J.A., Morse, A.P., Yu, L.Z., Yuan, N., 2001. Determining the sources ofatmospheric particles in Shanghai, China, from magnetic and geochemicalproperties. Atmospheric Environment 35 (15), 2615e2625.

Singh, A.K., Hasnain, S.I., Banerjee, D.K., 2003. Grain size and geochemicalportioning of heavy metals iin sediments of the Danodar River e a tributary ofthe lower Ganga, India. Environmental Geology 39, 90e98.

Spiteri, C., Kalinski, V., Rösler, W., Hoffmann, V., Appel, E., 2005. Magnetic screeningof a pollution hotspot in the Lausitz area Eastern Germany correlation analysisbetween magnetic proxies and heavy metal contamination in soils. Environ-mental Geology 49, 1e9.

Strzyszcz, Z., Magiera, T., 1998. Magnetic susceptibility and heavy metals contami-nation in soils of Southern Poland. Physics and Chemistry of the Earth 23(9e10), 1127e1131.

Thompson, R., Oldfield, F., 1986. Environmental Magnetism. Allen & Unwin, London.Tomlinson, D.L., Wilson, J.G., Harüs, C.R., Jeffrey, D.W., 1980. Problems in the

assessment of heavy metal levels in estuaries and the formation of a pollutionindex. Helgoländer Meeresunters 33, 566e575.

Wang, X., Qin, Y., 2006. Magnetic properties of urban topsoils and correlation withheavy metals: a case study from the city of Xuzhou, China. EnvironmentalGeology 49 (6), 897e904.

Wang, Y., Wei, F.S., 1995. Environmental Chemistry of Elements in Soil. ChinaEnvironmental Science Press, Beijing.

Wong, M.H., Au, K.C., 1984. Contents of Cd, Fe, Mn and Zn in the tissue of Katelysiahiantina collected from Tolo Harbour, an almost land-locked sea. Hydrobio-logical Bulletin 18 (2), 95e101.

Wu, X.Z., 2004. Statistics: From Data to Conclusions. China Statistics Press, Beijing.135e147.

Xie, S.J., Dearing, J.A., Boyle, J.F., Bloemendal, J., Morse, A.P., 2001. Associationbetween magnetic properties and element concentrations of Liverpool streetdust and its implications. Journal of Applied Geophysics 48, 83e92.

Yang, T., Liu, Q.S., Chan, L.S., Cao, G.D., 2007. Magnetic investigation of heavy metalscontamination in urban topsoils around the East Lake, Wuhan, China.Geophysical Journal International 171, 603e612.

Yang, T., Liu, Q.S., Li, H.X., Zeng, Q.L., Chan, L.S., 2010. Anthropogenic magneticparticles and heavy metals in the road dust magnetic identification and itsimplications. Atmospheric Environment 44, 1175e1185.

Zhang, C.X., Huang, B.C., Li, Z.Y., Liu, H., 2006. Magnetic properties of highroad-sidepine tree leaves in Beijing and their environmental significance. ChineseScience Bulletin 51 (24), 3041e3052.

Zhang, C.X., Huang, B.C., Piper, J.D.A., Luo, R.S., 2008. Biomonitoring of atmosphericparticulate matter using magnetic properties of Salix matsudana tree ring cores.Science of the Total Environment 393, 177e190.

Zhang, C.X., Liu, Q.S., Huang, B.C., Su, Y.L., 2010. Magnetic enhancement uponheating of environmentally polluted samples containing haematite and iron.Geophysical Journal International 181, 1381e1394.