Journal of Geochemical Exploration 140 (2014) 96–103

Contents lists available at ScienceDirect

Journal of Geochemical Exploration

j ourna l homepage: www.e lsev ie r .com/ locate / jgeoexp

Catchment basin analysis of stream sediment geochemical data:Incorporation of slope effect

Mehdi Abdolmaleki a, Ahmad Reza Mokhtari a, Somaieh Akbar a,⁎,Masood Alipour-Asll b, Emmanuel John M. Carranza c

a Department of Mining Engineering, Isfahan University of Technology, Isfahan 8415683111, Iranb Geology Department, Faculty of Earth Sciences, Shahrood University, Shahrood 3619995161, Iranc School of Earth and Environmental Sciences, James Cook University, Townsville, Queensland 4811, Australia

⁎ Corresponding author. Tel.: +98 9132527392.E-mail address: somaieh.akbar@mi.iut.ac.ir (S. Akbar).

http://dx.doi.org/10.1016/j.gexplo.2014.02.0290375-6742/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 23 December 2013Accepted 20 February 2014Available online 1 March 2014

Keywords:Stream sediment sampleCatchment basin approachSlope effectIran

This study examines catchment basin analysis of stream sediment samples considering slope effect by incorpo-rating 3D surfaces of catchments and lithologic units within the Rudbar 1:100,000 scale geological map(1:50,000 scale Mahin topographic sheet) in Northern Iran. In this region, 174 stream sediment samples werecollected in 625 km2 of survey area and were analyzed by ICP-OES for trace elements. Background values dueto upstream lithologic and dilution effects were calculated using 2D and 3D modeling. In each case, backgroundconcentration for every element due to lithology was estimated by weighted average method, and then geo-chemical residuals were determined and used for dilution effect correction. To identify the areas with possiblemineralization, dilution-corrected values in both 2D and 3D models were processed further separately usingprincipal component analysis. Then appropriate principal components (PCs) were integrated by fuzzy OR oper-ator to obtain a mineral favorability map per model. Rock samples, collected over the area, were used to validatethe results. Both 2D and 3Dmodels have good agreementwith the validation samples, but the 3Dmodelwas bet-ter. In other words, the use of 3D surfaces allows better representation of anomalies in the study regions. In ad-dition, validation against rock sample analyses demonstrated that using 3D surfaces improves the delineation ofpromising catchment basins. The effectiveness of incorporating slope effect in catchment basin modeling ofpromising areas was observed in dilution correction of background and in multivariate analysis of dilution-corrected residuals. Non-parametric significance test also confirmed that results using 2D and 3D surfaces aredifferent.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Stream sediment sampling remains an effective method in regionalgeochemical exploration. The results of such activity provide efficientguides for identifying regional geochemical patterns and locating areasof high potential for further mineral exploration. Various techniqueshave been developed and used for analysis and interpretation of region-al geochemical exploration data in order to extract the underlyingpatterns.

The sample catchment basin approach is a widely employed tech-nique for processing and analysis of regional stream sediment geo-chemical exploration data. The catchment basin of each streamsediment sampling point includes a region that hydrologically and, inturn, geochemically affects the chemical composition of stream sedi-ments at the sampling point. In other words, in this method, catchmentbasins are limited by the spill points, which are the stream sediment

sample locations. The chemical composition of stream sediments thatmigrate along drainage system resulted from weathering and erosionof upstream sources. A significant proportion of variations in elementconcentration in stream sediments are due to upstream lithology;therefore, catchment basin lithology can be used to evaluate geochem-ical background (Rose et al., 1970). Other properties of catchment basincan be applied in modeling of geochemical variations to predict anom-alous basins (Carranza and Hale, 1997; Sanford et al., 1993).

Based on definition of a model, predictive modeling involves de-scribing, representing and predicting an indirectly observable and com-plex real-world system by analyzing relevant data quantitatively(Carranza, 2009). For modeling of geochemical anomalies in samplecatchment basins, factors that influence variations in chemical composi-tion of geochemical samples should be recognized and taken intoaccount for processing and analysis of the data. Because lithology hasgreat influence on element content in stream sediment samples, back-ground concentrations of every element can be estimated as weightedaverage element content due to lithology using areal proportions of lith-ologic units in every sample catchment basin (Bonham-Carter et al.,

97M. Abdolmaleki et al. / Journal of Geochemical Exploration 140 (2014) 96–103

1987; Carranza andHale, 1997). In general, certain chemical contents ofstream sediments have positive relationships with areas of lithologicunits in a catchment basin and have negative relationships with totalarea of a catchment basin, and these relationships have been used tomodel background concentrations and dilution effect, respectively(Carranza, 2009).

The following equation shows the relationship between sample ele-ment concentration in a catchment and assumed anomaly and back-ground values (Hawkes, 1976):

YiAi ¼ YaAa þ Y ′i Ai−Aað Þ ð1Þ

where Yi and Ai are element concentration of sample i and correspondingcatchment basin area, respectively. Ya represents element concentrationdue to anomalous sources occupying an area Aa, Y'i represents elementconcentration due to background sources occupying an Ai − Aa (Jones,2002). Eq. (1) can be re-arranged as:

YaAa ¼ Ai Yi−Y0

i

� �þ Y

0

iAa ð2Þ

Local background content due to lithology in every sample catch-ment basin can be estimated in two steps. Firstly, a weighted averageelement concentration Mj (j = 1, 2… m) for the jth lithologic unitscan be calculated as:

Mj ¼Xn

i¼1YiXij=

Xni¼1

Xij ð3Þ

where Xij is area of the jth (j = 1, 2… m) lithologic unit in samplecatchment basin i (i = 1, 2… n), and the sum term in the denominator

Fig. 1. Lithologic mapAdapted from Rudbar

is total area of lithology j (Carranza, 2009). Then, the local backgroundconcentration of element (Y'i) due to lithology can be estimated as (4):

Y ′i ¼

Xmj¼1

MjXij=Xm

j¼1Xij ð4Þ

where the sum term in the denominator is total area of sample catch-ment basin i.

Aside from considering effects of lithology, it is important to consid-er dilution of concentrations along drainages within catchment basins.The term Y'iAa in Eq. (2) can be disregarded if Ai is much larger than Aa(Rose et al., 1979; Spadoni, 2006). It is assumed that exposed anomaloussources occupy a small unit area, e.g. Aa = 0.01 km2. Then, Eq. (5) isadapted for dilution correction of element concentrations in streamsed-iments (Carranza and Hale, 1997):

Ya ¼ 100 � Ai Yi−Y0

i

� �ð5Þ

Positive or negative geochemical residuals (Yi − Y'i) can beinterpreted as enrichment or depletion, respectively, of element con-centration in stream sediments. But only positive values are processedfurther for dilution effect correction because they are of interest inmin-eral exploration. In summary, measured concentrations at each streamsediment sample locationwere corrected for background concentrationdue to upstream lithology and for the effect of downstream dilution.

1.1. Problem definition

It should be noted that the area factor applied in Eq. (5) is a horizon-tal (2D) projection of the area whereas in reality chemical compositionof stream sediments is influenced by surfaces of lithologic units, whichvary in three-dimensional (3D) space, as do the surfaces of samplecatchment basins. If topographic slope is 0, the projected 2D area is

of the study area.1:100,000 scale geological map, Geological Survey of Iran.

Table 1Descriptive statistics of elements analyzed for stream sediment samples. Concentrationsare in ppm unless mentioned otherwise.

Element Min Max Mean St. dev. Skew.

Au (ppb) 0.75 22.00 1.63 2.44 6.75Al (%) 2.34 9.32 7.45 1.02 −1.88As 5.00 482 25.14 40.5 8.72Ba 203 2370 538.06 226.6 3.46Be 0.80 4.40 1.76 0.73 1.64Bi 0.08 3.00 0.26 0.35 4.50Ca (%) 0.56 27.4 4.96 3.77 3.08Ce 24.1 264 75.54 37.27 2.28Co 5.00 100 18.74 8.51 5.28Cr 20.00 270 61.38 39.10 2.09Cs 1.60 29.70 6.48 4.20 2.35Cu 10.00 408 51.33 34.01 6.94Fe (%) 1.70 27.5 6.28 3.06 4.07K (%) 0.75 4.48 2.40 0.54 0.22La 12.00 130 37.06 17.36 2.34Li 9.00 46.00 25.96 7.01 0.30Mg (%) 0.51 3.68 1.46 0.63 1.34Mn 316 6700 1420.7 684.9 3.61Mo 0.08 9.00 1.89 1.18 2.47Na (%) 0.45 3.43 1.74 0.48 0.20Nb 4.50 135 18.66 14.03 4.20Ni 10.00 130.00 31.93 14.13 2.32P (%) 0.05 0.37 0.14 0.05 1.46Pb 11.00 457 37.56 46.96 5.61Rb 35.80 214 100.5 30.99 0.98S (%) 0.01 8.22 0.17 0.73 8.53Sb 0.20 9.20 2.31 1.73 1.61Sc 5.00 36.00 15.43 4.92 0.93Sn 0.15 15.00 2.25 1.39 4.79Sr 111 1050 376.05 145.3 1.37Th 2.40 129 14.98 16.37 3.53Ti (%) 0.18 2.37 0.64 0.24 3.00Tl 0.20 2.60 0.46 0.26 4.16U 0.80 23.9 3.17 2.72 4.10V 60.00 1030 200.70 117.9 3.73W 0.08 24.50 2.27 2.16 6.77Y 10.80 71.60 26.80 10.36 1.81Zn 44.00 912 132.5 93.26 4.80Zr 26.00 229 102.90 47.11 0.39

98 M. Abdolmaleki et al. / Journal of Geochemical Exploration 140 (2014) 96–103

equal to the 3D area of the catchment; but, as slope increases, the ratioof 3D area to the projected 2D area increases as well. Therefore, topog-raphy affects the relative areas of various lithologic units and, thus, theircontributions to background concentrations in stream sediments. Inaddition, topographic slope and 3D area of sample catchment basinsaffect the intensity of elemental dilution in stream sediments.

It is hypothesized that by calculating and applying 3D surface areainstead of projected 2D area in Eqs. (3) to (5), topographic effects onvariations of stream sediment compositions can be taken into account.The new equations are as follows:

M 3Dð Þ j ¼Xn

i¼1YiX 3Dð Þij=

Xni¼1

X 3Dð Þij ð6Þ

Y ′3Dð Þi ¼

Xmj¼1

M 3Dð Þ jX 3Dð Þij=Xm

j¼1X 3Dð Þij ð7Þ

Y 3Dð Þa ¼ 100 � A 3Dð Þi Yi−Y ′3Dð Þi

� �ð8Þ

where X(3D)ij represents 3D surface area of lithologic unit j in samplecatchment basin i and A(3D)i is 3D surface area of sample catchmentbasin i.

To demonstrate the hypothesis associated with Eqs. (6) to (8), partof the 1:100,000 scale geological map of Rudbar (Iran) and stream sed-iment geochemical data available in the regionwere used in the presentresearch.

2. Study area

The study area is located west of the Alborz Mountains (northernIran), part of Talysh and Tarom Heights, and comprises the southwest-ern quarter of the 1:100,000 scale Rudbar geological map. It is situatedbetween 49°00′ and 49°15′ N latitudes and between 36°30′ and36°45′ E longitudes.

The study area includes various rock units from Paleozoic to Cenozoic(Fig. 1). The oldest units are volcanic rocks with basic to intermediatecomposition that have been metamorphosed severely and alteredslightly. Granitoids include granite, monzonite and aplite, which exhibitsevere alteration at and near the surface. Magnetite and tourmalineveins and lenses are observed widely in some sections of the area. Inthe study area, hydrothermal fluids sourced from intrusive bodies areassumed to be responsible for the formation of polymetallic vein-typemineralization of mainly Cu-Au-Pb-Zn. Argillic, propylitic, chlorite andepidote alterations are also observed near the veins at different loca-tions (Geological Survey of Iran, 2008). A wide range of Mesozoic sedi-mentary rocks and Quaternary sediments are other lithologic units inthe study area.

2.1. Sampling and analysis

In this study, we used data for 174 stream sediment samples werecollected from the southwest part of the Rudbar 1:100,000 scale geolog-ical map (corresponding to the 1:50,000 scaleMahin topographicmap).The samples were collected at a density of one sample per about3 km2and each sample consists of 25 sub-samples collected over30–50 m of the active part of a stream channel. The samples were pre-pared in the field by sieving stream sediments through 177 μm screenand then the b177 μm fraction was collected for chemical analysis.The samples were digested in HNO3 + HCl (aqua regia) and then ana-lyzed for multi-elements using Varian 735-ES inductively coupledplasma-optical emission spectrometry (ICP-OES) at the Zar-Azma labo-ratory in Tehran, Iran (Geological Survey of Iran, 2008). For measuringAu, fire assay method was employed and the final aliquot was analyzedusing Perkin-Elmer 5300 Atomic Absorption Spectrometer (AAS) at thesame laboratory. Elemental concentrations below the detection limits

were replaced by 0.75 of detection limit of the instrument (Sanfordet al., 1993). To monitor the analytical precision using sub-sample du-plicates (2×RSD), the procedure of Thompson and Howarth (1976)was used. The elements mostly showed precision better than ±10%.The descriptive statistics of the analytical data for the stream sedimentsamples are shown in Table 1.

In addition to stream sediment samples, we collected 76 rock chipsamples from veins, alterations and mineralized zones, dykes as wellas intrusive and volcanic rocks. These samples were collected afterthe stream sediment sampling in order to check some of the samplecatchment basins associated with high elemental concentrations. Therock chip samples were crashed and pulverized to 200 μm size, andthe b200 μm fraction was analyzed for total elemental content. Analyt-ical data for the rock chip samples were later employed in this study toevaluate the efficiency of the developed method.

3. Methodology

In order to calculate element background concentrations, digitizedgeology map and digital elevation model (DEM) of the study areawere utilized. The sample catchment basins (SCBs) were outlinedbased on the DEM and stream sediment sampling locations (used asspill points). Each SCB corresponds to the area upstream of each sampleextending to the next sample location upstream.

Variations in topographic slopes were divided into three classes:(1) b25°, (2) 25–30° and (3) N30°. The study area has mainly moderateslopes, although 13% of the SCBs showed an average slope greater than30° (Fig. 2).

Fig. 2. Sample catchment basins of the study area. Left: 2D view, average slope variations in the basins, black dots are sampling locations and white lines are catchment boundaries. Right:3D view of sample catchment basins.

99M. Abdolmaleki et al. / Journal of Geochemical Exploration 140 (2014) 96–103

To calculate background concentration per element due to lithology,it is required to determine the contribution of every lithologic unit ineach SCB. As this research aims to find out the lithologic contributionbased on real surface area (i.e., in 3D) and to compare the resultsbased on surface projection of lithologic boundaries (i.e., in 2D), wemeasured the surfaces of each SCB in 2D and 3D. Therefore, 2D and 3Dsurfaces were calculated and 2D areas were used in Eqs. (3) and (4)for 2D modeling whereas 3D areas were used in Eqs. (6) and (7) to in-corporate slope effect.

4. Results and discussion

The elemental background value (Mj), in Eq. (4) for every lithologyunit was calculated and part of the results is shown in Tables 2 and 3.From these results, it can be inferred that maximum background con-centrations of Au and Cu pertain to the QF2 unit (young gravel fan)whereas maximum Pb, Zn and Mn background values pertain to the Et

Table 2Weighted average content of some elements in each lithologic unit (Mj) calculated using 2D ar

2D Area Au Ba Ca Co Cr Cu

QT1 13.75 1.3 567.0 4.4 17.4 81.1 52.3E2T 37.01 1.1 575.7 3.7 23.2 70.6 54.5E1TV 75.87 1.2 537.4 4.6 16.0 57.5 44.5E2TV 18.53 1.9 878.1 1.4 15.0 39.6 48.6G1 123.38 1.2 451.8 3.8 14.4 60.2 46.4E2V 12.09 1.2 452.2 3.8 18.2 65.5 41.0QF2 1.10 2.0 597.4 1.9 17.8 51.9 58.2G2 38.64 1.4 427.7 3.6 17.0 59.1 50.1QC 11.26 1.1 507.4 4.6 16.2 72.6 47.8ETV 124.70 1.0 572.7 4.4 18.4 60.5 45.8ET 30.42 1.3 546.5 6.3 15.3 50.7 45.9EV 11.27 1.1 503.4 5.1 17.8 57.5 40.8ETA 26.54 1.0 525.7 4.1 19.6 58.5 44.6NG1 4.41 0.8 651.2 4.5 16.0 53.3 35.7NG 6.16 0.9 599.0 4.5 21.1 76.8 50.1NG2 3.10 0.9 585.6 5.2 15.2 77.7 30.9NG2CS 0.32 1.0 528.0 4.7 15.0 90.0 52.0NG2M 1.11 0.9 736.6 4.1 26.6 76.3 52.0E1S 5.36 1.3 580.8 6.1 16.6 64.0 53.6

unit (pyroclastic). Formost of the elements, there is no considerable dif-ference between background concentrations using 2D and 3D areas.

Comparisons of dilution-corrected values using 2D and 3D areasdemonstrate that statistically themedian equality of the twopopulationsis rejected at 95% confidence level based on sign test (non-parametric).In otherwords, the hypothesis of the current research is proved by incor-porating the 3D surface areas of catchments and lithologies into thepresent formulations.

Recognition of anomalous catchments based on univariate analysisof elemental data can be useful for delineation of promising areas for ex-ploration. However, it is proven that multivariate techniques can bemore effective in separating anomalous and background samples. Inthis study, principal component analysis (PCA)was used to define inter-relationships between dilution-corrected stream sediment elementcontents and to recognize anomalies in the study area.

Because PCA requires a set of variables with normal distribution andas geochemical data invariably do not exhibit normal distribution, thenon-parametric approach to PCA was used. That is the reason that PCA

eas.

Mo Ni Pb S Sb Sn U Zn

2.0 30.4 29.6 632.9 1.9 2.3 3.3 117.71.6 33.2 26.5 557.5 1.4 2.0 2.8 123.51.8 28.2 29.2 612.7 2.1 2.1 2.7 111.61.4 25.1 28.8 252.0 1.0 2.0 2.3 123.22.4 23.3 37.7 761.0 2.6 2.9 4.1 132.41.6 30.4 31.6 462.8 1.7 2.1 2.8 138.91.2 45.0 25.3 268.4 0.8 2.0 2.2 128.92.3 26.1 38.1 552.6 2.6 2.8 3.9 128.72.3 25.9 32.2 938.5 2.4 2.9 4.1 115.71.5 30.3 23.6 392.2 1.4 1.7 2.3 105.31.7 28.1 41.8 674.7 2.1 1.7 2.8 151.92.1 30.6 26.6 701.8 2.7 2.7 3.3 98.51.5 25.6 28.9 779.5 1.9 2.5 3.2 109.51.5 29.3 23.5 458.3 1.5 2.2 2.2 86.92.4 33.6 28.4 1131.9 3.0 2.7 3.4 116.01.6 22.8 20.0 312.9 1.6 2.2 2.1 87.13.0 24.0 35.0 1250.0 3.4 3.0 5.1 122.03.5 34.8 39.4 2944.4 3.2 3.2 4.1 141.21.4 28.6 38.5 419.5 2.1 2.0 2.3 151.0

Table 4Results of PCA of 2D and 3D modeling datasets.

F1 F2 F3 F4 F5 F6

2D modelingVariability (%) 27.919 13.381 7.894 6.688 6.132 5.529Cumulative % 27.919 41.301 49.195 55.883 62.015 67.544Al 0.261 0.039 0.650 −0.214 −0.480 −0.133As 0.451 −0.095 0.335 0.376 0.403 −0.165Au −0.001 −0.207 −0.072 0.450 0.073 −0.113Ba −0.022 0.027 0.495 −0.093 0.067 0.670Be 0.828 0.143 0.018 0.054 −0.172 −0.151Co −0.268 0.706 −0.020 0.157 −0.007 −0.074Cr −0.360 0.713 0.025 0.095 0.194 0.159Cu −0.270 0.408 0.354 0.285 −0.099 −0.101Fe 0.079 0.867 −0.060 −0.111 0.329 −0.017K 0.411 −0.114 0.588 −0.165 −0.324 −0.001Li −0.018 0.329 −0.061 0.678 −0.251 −0.106Mn 0.395 0.302 −0.285 0.262 −0.190 0.169Mo 0.528 0.098 −0.035 0.260 0.344 0.305Ni −0.259 0.416 0.038 0.464 −0.096 0.305Pb 0.544 0.092 −0.126 0.165 −0.101 0.478S −0.159 −0.315 0.236 −0.016 0.671 −0.016Sb 0.651 −0.109 0.261 0.256 0.275 −0.175Sn 0.749 0.212 −0.180 0.063 0.023 0.057U 0.738 0.057 −0.052 −0.049 0.391 −0.170W 0.810 0.010 −0.179 0.114 0.141 0.017Zn 0.374 0.334 −0.058 0.301 −0.155 0.456

3D modelingVariability (%) 27.743 13.205 7.800 6.353 6.082 5.616Cumulative % 27.743 40.948 48.748 55.101 61.183 66.799Al 0.257 0.054 0.600 0.135 0.568 −0.140As 0.469 −0.124 0.421 −0.186 −0.430 −0.161Au 0.002 −0.234 0.021 0.208 −0.372 −0.124Ba −0.046 0.045 0.474 −0.111 0.052 0.672Be 0.806 0.141 0.024 0.216 0.017 −0.169Co −0.258 0.693 0.012 0.140 −0.150 −0.091Cr −0.365 0.705 0.048 −0.042 −0.239 0.155Cu −0.257 0.392 0.417 0.179 −0.062 −0.124Fe 0.088 0.859 −0.083 −0.329 −0.110 0.006K 0.403 −0.109 0.545 0.085 0.405 −0.014Li 0.008 0.281 0.081 0.645 −0.380 −0.144Mn 0.426 0.269 −0.205 0.337 −0.079 0.176Mo 0.523 0.073 0.036 −0.105 −0.389 0.285Ni −0.243 0.384 0.144 0.410 −0.320 0.303Pb 0.547 0.084 −0.073 0.202 −0.085 0.472S −0.187 −0.287 0.204 −0.550 −0.415 0.003Sb 0.640 −0.125 0.313 −0.103 −0.305 −0.186Sn 0.757 0.196 −0.152 0.012 −0.048 0.063U 0.720 0.076 −0.076 −0.351 −0.234 −0.160W 0.810 0.006 −0.145 −0.024 −0.196 0.027Zn 0.401 0.305 0.024 0.272 −0.076 0.457

Table 3Weighted average content of some elements in each lithologic unit (Mj) calculated using 3D areas.

3D Area Au Ba Ca Co Cr Cu Mo Ni Pb S Sb Sn U Zn

QT1 13.98 1.3 566.8 4.4 17.4 81.2 52.3 2.0 30.4 29.6 631.9 1.9 2.2 3.3 117.9E2T 38.41 1.1 576.6 3.7 23.2 70.7 54.4 1.6 33.2 26.6 559.9 1.4 2.0 2.8 123.6E1TV 83.06 1.2 541.4 4.5 15.9 57.1 44.7 1.8 28.4 30.2 604.8 2.1 2.1 2.7 114.3E2TV 19.91 1.9 881.6 1.3 14.9 39.1 48.4 1.4 24.9 28.8 251.2 1.0 2.0 2.3 123.1G1 137.18 1.2 452.8 3.8 14.4 60.7 46.4 2.4 23.4 37.8 760.1 2.6 2.9 4.1 132.9E2V 13.14 1.2 456.4 3.9 18.1 66.1 41.3 1.6 30.5 31.8 461.2 1.6 2.1 2.8 139.8QF2 1.10 2.0 597.3 1.9 17.8 51.9 58.2 1.2 45.0 25.3 268.3 0.8 2.0 2.2 128.9G2 42.85 1.4 428.6 3.6 17.0 59.8 50.3 2.4 26.2 38.2 553.7 2.5 2.8 3.9 129.3QC 11.81 1.1 507.1 4.6 16.2 72.7 47.8 2.3 25.9 32.3 939.9 2.4 2.9 4.1 115.8ETV 132.07 1.0 575.0 4.4 18.4 60.4 45.9 1.5 30.3 23.9 392.8 1.4 1.7 2.3 105.9ET 33.03 1.3 548.0 6.3 15.2 50.8 46.0 1.7 28.1 42.2 673.2 2.1 1.7 2.7 153.1EV 12.48 1.1 504.3 5.0 17.8 58.1 40.9 2.1 30.4 26.6 702.2 2.7 2.7 3.4 98.8ETA 27.90 1.0 525.4 4.1 19.5 58.4 44.6 1.5 25.6 28.9 780.3 1.9 2.5 3.2 109.7NG1 4.51 0.8 654.4 4.5 16.0 53.1 35.6 1.5 29.1 23.4 457.2 1.5 2.2 2.2 86.9NG 6.34 0.9 598.4 4.5 21.1 76.7 50.1 2.4 33.7 28.4 1120.7 3.0 2.6 3.4 115.8NG2 3.29 0.9 583.7 5.2 15.2 77.3 30.9 1.6 22.7 20.1 318.4 1.6 2.3 2.2 87.2NG2CS 0.32 1.0 528.0 4.7 15.0 90.0 52.0 3.0 24.0 35.0 1250.0 3.4 3.0 5.1 122.0NG2M 1.12 0.9 737.0 4.1 26.6 76.3 52.0 3.5 34.8 39.4 2946.3 3.2 3.3 4.1 141.2E1S 5.74 1.3 582.2 6.1 16.6 64.2 53.5 1.4 28.6 39.0 421.0 2.1 2.0 2.3 152.2

100 M. Abdolmaleki et al. / Journal of Geochemical Exploration 140 (2014) 96–103

was conducted using the Spearman correlationmatrix (non-parametricstatistic) for the dilution-corrected residuals. Table 4 shows the resultsof the PCA of dilution-corrected residuals obtained by using 2D and3D areas. The PCAwas carried out only for the SCBs that yielded positivedilution-corrected residuals for Au or Cu or Pb. The loading for the firstsix PCs are given in Table 4 for both of 2D and 3D approaches. As thestudy region is favorable for Au, Cu and polymetallic mineralization,the related pathfinder elements in addition to major controlling ele-ments are considered for selection of the appropriate PCs. PC1 displaysconsiderable loading values for elements of interest (e.g. As, Mo, Pband Sb). The second component (PC2) is ignored as it shows the highimpact of iron on this component and it may reflect iron oxide scaveng-ing effects. There is no significant contribution from the elements of in-terest in the 3rd PC and also high loadings in this component belong toAl and K, probably showing the impact of chemical composition of lith-ologic units (mainly felsic rocks). However, some commodity elementsare pronounced in PC4 and PC5 and additionally these componentsshow low coefficient loading for elements like Al and K reflecting lowcontribution fromclaymaterials.Moreover, Fe andMnhave insignificantloading values in PC4 andPC5 reflecting that variations are not due to thescavenging effects of iron and manganese oxides. As a result, PC1, PC4and PC5 are selected as the components more likely related to mineral-ized zones and are employed for further processing in this research.

The results of the two approaches, using 2D and 3D surface areas, arepresented in Figs. 3 and 4, respectively. Each of these figures includesthree potential maps of the study area, corresponding to PC1, PC4 andPC5 of each dataset, and a fuzzy integrated potential map of the areagenerated by fusing the three PCmaps into a singlemap. For themap fu-sion,minimum tomaximumPC valueswere transformed linearly to therange from 0 to 1, and then the fuzzy OR operator was used to integratethe transformed PC maps.

4.1. Validation of results

Table 5 lists the descriptive statistics of some elements in rock chipsamples, for which threshold values were calculated as median+ 2MAD, and the number of anomalous samples per element is alsoshown in the table (Carranza, 2009). To facilitate comparison of theSCB approaches and to examine the efficiency of the models, results ofanalysis of the rock chip sample data are highlighted in Figs. 3 and 4using different symbols. The maps are annotated with Au, Cu or Pb atrock chip sample locations if values of one element are above threshold;if values of at least two of these elements are above the threshold, differ-ent symbols are used.

Fig. 3.Maps of PC1, PC4 and PC5 obtained by 2D modeling and map obtained by fuzzy integration (black symbols are anomalous rock chip samples and red points are non-anomalousones).

101M. Abdolmaleki et al. / Journal of Geochemical Exploration 140 (2014) 96–103

The most favorable catchment basins correspond to the acidicvolcanic units particularly tuff and pyroclastic units. The centraland northwestern parts of the study area are defined as promisingareas based on comparison with analysis of rock chip samples, andparticularly in the northwestern part there is a known copper potential.As shown in Fig. 2, these promising areas are characterized by highlysteep slopes.

Comparison of the results of the 2D and 3D approaches reveals thatincorporation of slope effect in sample catchment basin analysis ofstream sediment samples improves delineation of promising areas forexploration. The PC1 maps obtained through each approach displayalmost the same results, but the PC4 map obtained through the 3Dapproach has successfully highlighted the central, northwest and south-west parts of the study area where anomalous rock chip samples arepresent but these parts of the study area are not promising areas accord-ing to results of the 2D approach. Moreover, this issue is more pro-nounced in the PC5 map. Comparison of the 2D and 3D approachesfurther shows that most of favorable catchments are missed by using2D surface areas but they are delineated by using 3D surface areas.The most considerable deviations of the 2D approach from the valida-tion data can be seen in the fuzzy integratedmap. Comparison of the re-sults of the 2D and 3D approaches with respect to the rock chip samplesdemonstrates that 2D modeling has failed to highlight the promising

areas or it has classified some favorable catchments as lower priorityclass compared to 3D classification. In general, it can be concludedthat 3D modeling, to account for slope effect, delivers more accurateand reliable results especially in regions with highly steep topography.In order to statistically test the hypothesis in multivariate context(PC results), the sign test of Lehmann (1975) was applied to a map ob-tained by fuzzy integration of the PC maps obtained and it has shownthat the null hypothesis that the results of 2D and 3D approaches arethe same is rejected at 95% confidence level.

5. Conclusion

Elemental background concentrations in stream sediments and cor-rection for dilution effect can be performed using 2D surface areas oflithologic units and sample catchment basins. This study attempted toincorporate slope effect in catchment basin analysis of geochemicalanomalies by using 3D surface area in background calculation and dilu-tion correction. Slope effect seems to be an important factor of streamsediment element concentrations where the slope is greater than 30°.The difference between the 2D and 3D modeling approaches is mainlyin the dilution correction of element background concentrations. For agiven catchment, if topography does not contain steep slopes then thereal surface areas of lithologic units (i.e., in 3D) will be close to the

Table 5Results of rock sample analysis, all concentrations are in ppm except Au in ppb.

Min Max Median MAD Median + 2MAD Number of anomalous samples

Co 1.25 74.22 10.37 5.06 20.49 12Cr 0.00 167.10 12.52 6.60 25.72 11Ni 1.43 61.05 8.00 3.56 15.13 10Mn 41.00 11010 818.35 550.25 1918.85 8Fe 4658 292385 29440 14145 57730 11Zn 18.71 135000 162.45 102.32 367.08 12Mo 0.42 141.80 4.28 2.71 9.69 13As 1.14 30070 24.73 19.96 64.64 18Pb 4.00 19054.5 73.88 50.43 174.74 14Ag 0.05 254.40 0.71 0.45 1.60 11Au 1.30 760.00 3.25 1.40 6.05 13Cu 5.52 236052.5 85.29 63.62 212.52 17

Fig. 4.Maps of PC1, PC4 and PC5 obtained by 3D modeling and map obtained by fuzzy integration (black symbols are anomalous rock chip samples and red points are non-anomalousones).

102 M. Abdolmaleki et al. / Journal of Geochemical Exploration 140 (2014) 96–103

projected surface area (i.e., in 2D); therefore, if topographic slopes in asurvey area are low (e.g., b30°), the 3D and 2D approaches to catchmentbasin analysis will yield similar results. The case tested in this study, innorthern Iran, comprising stream sediment and rock geochemicaldata, demonstrates that use of real surface or 3D areas of lithologicunits in catchment basin analysis of geochemical anomalies, combinedwith application of multivariate and fuzzy integration techniques, canimprove delineation of really promising areas.

References

Bonham-Carter, G.F., Rogers, P.J., Ellwood, D.J., 1987. Catchment basin analysis applied tosurficial geochemical data, Cobequid Highlands, Nova Scotia. J. Geochem. Explor. 29,259–278.

Carranza, E.J.M., 2009. Geochemical anomaly and mineral prospectively mapping in GIS,Handbook of Exploration and Environmental GeochemistryFirst edition. Elsevier.

Carranza, E.J.M., Hale, M., 1997. A catchment basin approach to the analysis of reconnais-sance geochemical–geological data from Albay Province, Philippines. J. Geochem.Explor. 60, 157–171.

103M. Abdolmaleki et al. / Journal of Geochemical Exploration 140 (2014) 96–103

Geological Survey of Iran (Parspay Azma Consulting Co.), 2008. Geochemical-heavy min-eral exploration in 1:100000 Rudbar geological map (internal report-in Farsi).

Hawkes, H.E., 1976. The downstream dilution of stream sediment anomalies. J. Geochem.Explor. 6, 345–358.

Jones, R., 2002. Algorithms for using a DEM for mapping catchment areas of streamsediment samples. Comput. Geosci. 28, 1051–1060.

Lehmann, E.L., 1975. Nonparametric: Statistical Methods Based on Ranks. Holden Day,San Francisco.

Rose, A.W., Dahlberg, E.C., Keith, M.L., 1970. Multiple regression technique for adjustingbackground values in stream sediment geochemistry. Econ. Geol. 65, 156–165.

Rose, A.W., Hawkes, H.E., Webb, J.S., 1979. Geochemistry in Mineral Exploration, 2nd edn.Academic Press, London.

Sanford, R.F., Pierson, C.T., Crovelli, R.A., 1993. An objective replacement method for cen-sored geochemical data. Math. Geol. 25, 59–80.

Spadoni, M., 2006. Geochemical mapping using a geomorphologic approach based oncatchments. J. Geochem. Explor. 90, 183–196.

Thompson, M., Howarth, R.J., 1976. Duplicate analysis in geochemical practice. Part 1:theoretical approach and estimation of analytical reproducibility. Analyst 101,690–698.