GY

GJa

b

c

d

a

ARRAA

KSESTE

1

i2ei

A

s

h0

Ecological Engineering 94 (2016) 306–314

Contents lists available at ScienceDirect

Ecological Engineering

jo ur nal home p ag e: www.elsev ier .com/ locate /eco leng

IS-mapping spatial distribution of soil salinity for Eco-restoring theellow River Delta in combination with Electromagnetic Induction

uangming Liua,∗∗, Jinbiao Lia, Xuechen Zhanga, Xiuping Wangb, Zhenzhen Lva,ingsong Yanga,∗∗, Hongbo Shaoc,d,∗, Shipeng Yua

State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, ChinaInstitute of Coast Agriculture, Hebei Academy of Agriculture and Forestry Sciences, Caofeidian, 063200, ChinaInstitute of Agro-biotechnology, Jiangsu Academy of Agricultural Sciences, Nanjing 210014, ChinaYantai Institute of Coastal Zone Research,Chinese Academy of Sciences,Yantai 264003,China

r t i c l e i n f o

rticle history:eceived 16 June 2015eceived in revised form 23 February 2016ccepted 12 May 2016vailable online 18 June 2016

eywords:oil salinitylectromagnetic inductionpatial distributionhe Yellow River deltaco-restoration

a b s t r a c t

Soil salinization is one of serious ecological problems around the world, which seriously restricts thestability of ecosystem and the economic development of agriculture. Mapping and monitoring spatialdistribution of soil salinity is important for management of ecology and agriculture. This study wascarried out to explore the spatial distribution of soil salinity in the Yellow River Delta using the portabledevice EM38-MK2 with geostatistical analysis. Apparent soil electrical conductivities were measuredunder four kinds of measurement modes (0.5H, 0.5 V, 1.0H and 1.0 V, respectively). The results revealedthat electrical conductivity of 1:5 soil to water extract (ECe) varied from 0.965 to 1.872 dS m−1 for allsampled soil profiles, and the salinity of topsoil was the highest, which indicated that soil soluble saltsaccumulated to the surface. The salinity in the top layer showed strong spatial variability while salinitiesof other layers were moderate. Soil salinity displayed a significant zonal distribution, gradually decreasingwith the increase of distance to the coastline. The regions with high ECa values were located in the northand the east of the study area, whereas regions with low ECa values were located in the south and the westparts. The correlation coefficient showed that salinities of adjacent two soil layers reached a significantlevel of 0.01, and gradually decreased with increasing soil depth. The linear interpretation models withECa as independent variables and ECe as dependent variables for each depth were with R2 between 0.828and 0.919. The interpretation models, taking ECa and ECe of 0- 15 cm depth as independent variables,

2

and ECe of each layer in 15–100 cm depth as dependent variables, were with higher R between 0.930and 0.953. The mean error (ME) showed that there was small positive deviation in 40–100 cm whereasa high positive deviation in the topsoil (0–40 cm). The scatter plots also indicated that the models hadbetter accuracy of salinity estimation in the top soil layers (0–80 cm). The results provides solid basis foreco-restoring in the Yellow River delta.

© 2016 Elsevier B.V. All rights reserved.

. Introduction

Salt salinity is a common problem around the world, especiallyn arid and semi-arid areas (Wichelns and Qadir, 2014; Singh et al.,

013; Wang et al., 2008; Li et al., 2015) with low rainfall and highvaporation. Apart from the natural factors resulting in soil salin-zation, long-term irrigated agriculture with poor drainage system

∗ Corresponding author at: Institute of Agro-biotechnology, Jiangsu Academy ofgricultural Sciences, Nanjing 210014, China.

∗∗ Corresponding authors.E-mail addresses: gmliu@issas.ac.cn (G. Liu), jsyang@issas.ac.cn (J. Yang),

haohongbochu@126.com (H. Shao).

ttp://dx.doi.org/10.1016/j.ecoleng.2016.05.037925-8574/© 2016 Elsevier B.V. All rights reserved.

contributes to secondary salinization (Ouni et al., 2013; Qadir andOster, 2004; Kitamura et al., 2006). Salt accumulation has detrimen-tal effects on soil physical and chemical properties (Zhang et al.,2014; Shukla et al., 2011) and on enzyme activities and microbialand biochemical activities (Rietz and Haynes, 2003; Yuan et al.,2007; Karlen et al., 2008), thus inhibiting agriculture productiv-ity (Rady, 2011; Ouni et al., 2014). Excessive Na+ may cause highpH, changes in soil solution ions and nutrients, and destabiliza-tion of soil structure (Li et al., 2015). Also, salt toxicity influencesplant growth significantly (Yan et al., 2013) from two aspects:

hyperosmotic stress and hyperionic stress (Tang et al., 2014; Jianget al., 2016). Saline soils are generally characterized by an electri-cal conductivity (EC) > 4 dS m−1, pH < 8, and exchangeable sodium

nginee

pp

hGs(omsrtrseciws

fSoEamwcscttpcwfieDsafcatltiysmeueA

rlc(Evsss

G. Liu et al. / Ecological E

ercentage (ESP) < 15%, whereas in alkaline soils with EC < 4 dS m−1,H > 8 and ESP > 15% (USSLS, 1969).

Soil properties, such as cation exchange capacity (CEC), water-olding capacity and soil fertility are influenced by the soil texture.eostatistical analysis has been one common method to depictpatial dependence and variability of soil properties. Cao et al.2011) described the characteristics and spatial variability of soilrganic matter and organic carbon around Qinghai Lake using theethod called geostatistics. Cemek et al. (2007) also used the geo-

tatistical method to assess spatial variability of soil propertieselated to soil salinity and alkalinity and to discuss the spatial dis-ribution patterns in the Black Sea coastal region of Samsun. Theesults reported that there was strong spatial dependence in top-oil, whereas moderate spatial dependence in subsoil. Elbasiounyt al. (2014) conducted a survey about the spatial variability of soilarbon and nitrogen pools in north Nile Delta, which was signif-cant to obtain a better understanding of biological process and

as crucial to make good strategies and managements to developustainable agriculture by using ordinary Kriging method.

Apparent soil electrical conductivity(ECa)has been widely usedrom 1970s (Rhoades and Ingvalson, 1971; Rhoades and vanchilfgaarde, 1976). ECa is an instant, easy and reliable way tobtain spatial characterization of soil salinity. Many factors affectCa, such as water content, cation exchange capacity(CEC), sodiumdsorption ratio (SAR) and texture (Friedman, 2005). The electro-agnetic induction (EMI) is one of the fast methods to measure ECa,hich has been applied in different regions. Farzamian et al. (2015)

ombined electrical resistivity tomography (ERT) and EM38 to mea-ure moisture content variations and estimate saturated hydrauliconductivity in natural conditions. Sudduth et al. (2005) used ECao reflect soil physical and chemical properties. Researchers foundhat electromagnetic methods are fast and reliable to delineate soilroperties. This was the case that Heil and Schmidhalter (2012)haracterized soil texture variability in the Tertiary upland hillsith a digital terrain model and get the content of clay, silt and sand

rom ECa with EM38. The EMI method is widely used in monitor-ng soil salinity. Akramkhanov et al. (2014) used EMI system as anfficient and reliable method to monitor soil salinity in Uzbekistan.ing and Yu (2014) conducted a survey to monitor and evaluate the

patial and seasonal changes of soil salinity using remote sensingnd EMI, and found that spatial distribution of soil salinity dif-ered from small horizontal or vertical distances, and soluble saltsoncentrated in the surface through capillary movement whichggravated the salt variability in arid and semiarid areas. It is impor-ant to monitor soil salinity in an effort to manage the irrigatedand well. Herrero et al. (2011) conducted a survey with EMI sys-em to get the detailed baseline data for soil salinity in Flumenrrigation district where the land had been irrigated more than 50ears, showing that the EMI system was suitable to evaluate soilalinity in irrigated areas. The information of ECa survey is used forapping spatial variability of soil properties that are invaluable in

cosyetem and agriculture for assessing soil quality, planning landse, and determining the suitability of cropping patterns (Lescht al., 2005; Corwin and Lesch, 2005; Yao et al., 2007; Urdanoz andraguees, 2011).

Coastal districts are vulnerable to climatic changes, such asising temperature caused by several environmental factors: seaevel rises, seawater intrusion, changes in upstream river dis-harges, cyclones and erosion of coastal embankment contractionsRawlani and Sovacool 2011). With the application of portableM38, Barbiéro et al. (2001) studied salt distribution in the Senegalalley by obtaining the electromagnetic soil conductivity to under-

tand variability and spatial arrangement. The results showed thataline areas were distributed as strips. EMI is also used to map soilalinity of individual plots (Amezketa, 2006).

ring 94 (2016) 306–314 307

Yellow River Delta (YRD), the largest delta in China, which cov-ers the places alongside the Lower Yellow River, especially in theestuary formed by large amount of sediments carried by the Yel-low River (Cui et al., 2009). The YRD plays an important role inglobal ecosystem because it provides an indispensable staging, win-tering and breeding site for birds around Pacific migration route(Fan et al., 2011; Fan et al., 2012). However it is suffered primaryand secondary salinization because the shallow saline groundwaterand strong evaporation as well as human activities. The salinity ofthis area threatens food production and environment (Zhang et al.,2011; Fang et al., 2005; Ye et al., 2004). Mapping spatial distributionof soil salinity is important for ecology and agriculture arrangementand management as it reflects the use and the dynamics of soil andwater resources, which provides basic knowledge for researches onrestoration of saline land and farmland sustainability assessment(Adam et al., 2012).

The objectives of this paper are (1) to obtain optimal interpreta-tion models between ECa and ECe for the reconstruction of profilesoil salinity across the study area; (2) to map the three-dimensionalvariation of soil salinity using ordinary kriging approach based on3D scatter data and 3D Mesh model in the study area for restoringthe Yellow River Delta.

2. Materials and methods

2.1. Study area

The study area was the field of Dongying (including 5 counties),Shandong Province, China, which is the main field of lower Yel-low River Delta (Fig. 1). The climate is characterized by continentalmonsoon in the North Temperate Zone with seasonal fluctuationsin precipitation and temperature. The annual average precipita-tion is 580 mm, and the annual average evaporation/precipitationis 3.22. The soils are naturally saline due to very saline groundwater(average salinity of 30.1 g L−1) and shallow depth (average depth of1.2–2.4 m). Sandy loam is the predominant soil texture in this area.

2.2. Acquisition of data

2.2.1. Apparent electrical conductivityIn this study, the electromagnetic conductivity meter EM38-

MK2 was used to measure apparent soil electrical conductivity.The instrument containing two coils, 0.5 and 1.0m, respectively, isbased on the principle of electromagnetic induction. In Saline areas,apparent soil electrical conductivity is mainly associated with saltcontent whose contribution is generally greater than 80% and it willbe higher with the increase of soil salinity.

In this study, we used mechanical distribution method withthe grid of 4 km × 4 km and arranged a total of 259 measure-ment points. We measured the apparent soil electrical conductivityunder four measurement modes (0.5H, 0.5 V, 1.0H and 1.0 V, respec-tively) at the measurement position, corresponding to three soildepths (0–0.6, 0–1.2 and 0–1.5 m, respectively). The apparent soilelectrical conductivities were presented as EC0.5H, EC0.5V, EC1.0Hand EC1.0V, respectively. The coordinates of each measurementposition were determined by global positioning system, samplingdate in May 2014.

2.2.2. Soil sample analysis259 soil samples of 0–15 cm at the measurement point were

collected. In order to obtain more detailed information about soil

salinity, 84 random profiles among the 259 measurement pointswere selected, and 420 soil samples were collected in accordancewith 0–15, >15–40, >40–60, >60–80 and >80–100 cm at the mea-surement positions. These soil samples were air dried, crashed and

308 G. Liu et al. / Ecological Engineering 94 (2016) 306–314

nd dis

pT

2

v

1

Ea>ow

TD

Fig. 1. Location of sampling area a

assed through 2 mm sieve in the laboratory for further analysis.he 1:5 soil to water extracts were prepared for measuring ECe.

.3. Interpreting and mapping methods

The method used to interpret and map the three-dimensionalariation of soil salinity involved two main steps:

) Regression

First, regression models were made takingCa(EC0.5H EC0.5V EC1.0H and EC1.0V) as independent vari-

bles and ECe for each depth (0–15, >15–40, >40–60, >60–80 and80–100 cm)as dependent variables using the data of 84 profilebtained from the survey (Table 3). Then, linear regression modelsere fitted by using ECa(EC0.5H EC0.5V EC1.0H and EC1.0V) and ECe

able 1escriptive statistics of soil ECe and ECa0.143.

Layer (cm-cm) N Min Max Average Stan

ECe(0–15)(dS m-1) 259 0.143 8.970 1.872 2.14ECe(>15–40)(dS m-1) 84 0.083 4.000 0.965 0.89ECe(>40–60)(dS m-1) 84 0.083 4.400 0.996 0.84ECe(>60–80)(dS m-1) 84 0.082 4.250 0.984 0.82ECe(>80–100)(dS m-1) 84 0.082 4.170 0.970 0.75EC0.5H(mS m-1) 259 6 1980 233 306EC0.5V(mS m-1) 259 15 1816 248 269EC1.0H(mS m-1) 259 18 1886 246 277EC1.0V(mS m-1) 259 26 1395 241 211

tributions of soil sampling points.

of 0–15 cm as independent variables, and ECe of each other layerin 15- 100 cm as dependent variable (Table 4).

2) Mapping 3D soil electrical conductivity

According to the linear regression models shown in Table 4, wegot soil electrical conductivity of 259 points in the study area (atotal of 1259 points of five layers), and these data were used asdata base for mapping the three-dimensional distribution of ECe.

Using the GMS software, we built 3D Scatter Data model of soilelectrical conductivity (Fig. 3), which was as a basis for building3D Mesh Data Model in the study area and formatting the three-

dimensional structure model (Fig. 4). We used interpolation Krigingmethod of GMS software for three-dimensional soil salinity esti-mates. For better visualization, all 3D maps were expanded 20,000times in the vertical.

dard deviation Coefficient of variation Skewness Kurtosis

7 1.15 1.619 2.0351 0.92 1.339 1.0845 0.85 1.478 2.2875 0.84 1.587 2.5425 0.78 1.519 2.983

1.31 9.04 1.31 1.08 8.20 1.08 1.12 9.24 1.12 0.87 5.56 0.87

G. Liu et al. / Ecological Engineering 94 (2016) 306–314 309

trical

2

frfem

M

R

Fig. 2. Spatial distribution of apparent soil elec

.4. Validation

In order to validate the regression model, 32 points from the sur-ace soil (0–15 cm) and 16 points from other layers were selectedandomly. Three different criteria were used to evaluate the per-ormance of regression interpretation models of ECe, namely, meanrror (ME), root mean square error (RMSE), and coefficient of deter-ination (R2) and they are equated below as:Mean error:

E = 1m

N∑i=1

[z∗ (si) − z (si)] (1)

Root-mean-square error:

MSE =

√√√√ 1N

N∑i=1

[z∗ (si) − z (si)]2 (2)

conductivity (ECa) under different EMI modes.

R2 =

⎡⎢⎢⎢⎢⎣

∑Ni=1(z(si) − z(si)ave)(z∗(si) − z∗(si)ave))

N∑i=1

(z(si) − z(si)ave)(z∗(si) − z∗(si)ave)2

⎤⎥⎥⎥⎥⎦

2

wherez∗ (si)is the interpreted value, z (si)is the observed value,z(si)aveand z∗(si)ave shows the average of observed and interpretedvalues, N is the number of data.

When using the test model, root mean squared error (RMSE)is always used to make statistical analysis of conformity betweensimulated values and measured values. The smaller the RMSE value,

the better the consistency and the smaller the deviation betweensimulated values and measured values, and thus the more accurateand reliable of the model. RMSE can reflect well the interpretabilityof simulated values of the model.

310 G. Liu et al. / Ecological Engineering 94 (2016) 306–314

0.000

1.000

2.000

3.000

4.000

5.000

6.000

7.000

8.000

9.000

0.000 2.000 4.00 0 6.000 8.00 0

Mea sured ECe (dS/m)

Pre

dict

ed E

Ce

(dS

/m) ECe (0-15 cm)

R2=0.9159

0.000

0.500

1.000

1.500

2.000

2.500

3.000

3.500

0.000 0.500 1.000 1.500 2.000 2.500 3.000 3.500

Measured ECe (dS/m)

Pre

dict

ed E

Ce

(dS/

m)

ECe (15 -40cm)

R2=0.91 73

0.000

0.500

1.000

1.500

2.000

2.500

3.000

3.500

0.00 0 0.500 1.00 0 1.500 2.00 0 2.500 3.00 0

Mea sured ECe (dS/m)

Pre

dict

ed E

Ce

(dS

/m) ECe (40 -60 cm)

R2=0.938 3

0.00 0

0.50 0

1.00 0

1.50 0

2.00 0

2.50 0

3.00 0

3.50 0

0.000 0.50 0 1.000 1.50 0 2.00 0 2.500 3.000

Mea sured ECe (dS/m)

Pre

dict

ed E

Ce

(dS

/m) ECe (60-80 cm)

R2=0.931 2

a) b)

c) d)

0.000

0.500

1.000

1.500

2.000

2.500

0.00 0 0.50 0 1.00 0 1.50 0 2.000 2.500

Mea sured ECe (dS/m)

Pre

dict

ed E

Ce

(dS

/m) ECe (80-100cm)

R2=0.899 4

e)

Fig. 3. Scatter plots of the measured versus interpreted ECe: (a) 0–15 cm, (b) >15–40 cm, (c) >40–60 cm, (d) >60–80, and (e) >80–100 cm using regression model based onvalidation data set.

Table 2Pearson correlation coefficients of soil salinity at various depths (n = 84)0–15 cm.

0–15 cm >15–40 cm >40–60 cm >60–80 cm >80–100 cm

0–15 cm 1>15–40 cm 0.831** 1>40–60 cm 0.741** 0.933** 1>60–80 cm 0.635** 0.895** 0.956** 1

3

3

eE

Table 3Soil electrical conductivity interpretation models for different soil layers taking ECaunder different EMI modes as independent variabless.

soil layer (cm-cm) ECe = a + bEC0.5Ha + cEC0.5Va + d EC1.0Ha + e EC1.0Va

a b c d e R2

0–15 0.626 0.014 0.000 −0.009 0.000 0.833>15–40 0.255 0.000 0.003 0.000 0.000 0.828>40–60 0.104 0.000 0.000 0.000 0.004 0.861

indicated that soluble salts accumulated to soil surface. Notably,

>80–100 cm 0.604** 0.844** 0.932** 0.966** 1

. Results

.1. Descriptive statistics of the ECe and ECa

Soil salinity (ECe) at different depths (0–100 cm) and soil appar-nt electrical conductivity (ECa) readings under four modes ofM38-MK2 were displayed in Table 1.

>60–80 −0.009 −0.001 0.000 0.000 0.005 0.914>80–100 0.014 −0.001 0.000 0.000 0.006 0.919

The mean of soil salinity varied from 0.965 to 1.872 dS m−1 in soilprofiles, and the soil salinity of the topsoil was the highest which

profile ECe varied from 0.082 dS m−1 to 8.970 dS m−1. Coefficientof variation of soil salinity at each soil 1ayer was from 0.78 to1.15, exhibiting the moderate spatial variability except for salin-

G. Liu et al. / Ecological Engineering 94 (2016) 306–314 311

Fig. 4. The distribution of 3D scatter data and soil electrical conductivity in the study area.

Table 4Soil electrical conductivity interpretation models for different soil layers taking ECa under different EMI modes and ECe of 0–15 cm as independent variabless.

soil layer (cm- cm) ECe = a + bEC0.5Ha + c EC0.5Va + d EC1.0Ha + e EC1.0Va + f EC0-15

a b c d e f R2

0–15 0.626 0.014 0.000 −0.009 0.000 0.000 0.833>15–40 −0.028 −0.006 0.004 0.005 0.000 0.308 0.931>40–60 −0.052 −0.009 0.004 0.007 0.000 0.246 0.930>60–80 −0.055 −0.007 0.000 0.008 0.002 0.132 0.935

.000 0.007 0.003 0.146 0.953

i1

w0Tns

i0cd

3

svstEbs

dvw

Table 5Results of model evaluation criteria for ECe for the five standard depths based on avalidation data setSoil layers (cm- cm).

R2 ME (dS m−1) RMSE (dS m−1) N

0–15 0.916 0.337 0.420 32>15–40 0.917 0.211 0.301 16>40–60 0.938 0.152 0.196 16>60–80 0.931 0.152 0.204 16

>80–100 −0.044 −0.007 0

ty of topsoil. Likewise, the variation of ECa was from 6 mS m−1 to980 mS m−1.

The map of ECa under four modes (0.5H, 0.5 V, 1.0H and 1.0 V)as produced using the ordinary kriging. The map of the ECa under

.5H, 0.5 V, 1.0H or 1.0 V mode was shown as Fig. 2a–d, respectively.hese maps clearly illustrated that high ECa were located in theorth and the east of the study area, whereas low ECa were in theouth and the west parts.

Pearson correlation coefficients of soil salinity were presentedn Table 2. The correlation coefficients ranged from 0.604 to.966 and reached the significant level of 0.01. Correlation coeffi-ients between adjacent 2 soil layers were higher, which graduallyecreased with depth increasing.

.2. Regression model of ECe and the validation

First, linear regression models were fitted with ECa of four mea-urement modes as independent variables and ECe as dependentariables for each depth. Consequently, five equations for the wholeoil profile were derived and shown in Table 3. This Table showedhat ECa of four modes had a significant linear correlation withCe of each layer, and R2 varied from 0.828 to 0.919. This mighte attributed to the fact that ECa were mainly influenced by soilalinity in saline area (Rhoades et al., 1990).

Then, linear regression models, taking ECa and ECe of 0–15 cmepth as independent variables and ECe of each layer as dependentariables, were fitted (Table 4). Four equations for the 15–100 cmere derived. Table 4 showed that ECa and ECe of 0–15 cm had

>80–100 0.899 0.199 0.199 16

significant correlation with ECe of other layers from 15 to 100 cm,and R2 ranged from 0.930 to 0.953. Therefore, it could be concludedthat equations in Table 4 can be used directly for the reconstructionof the profile soil salinity across the study area.

Table 5 showed the validation results for ECe at each depth. R2

ranged from 0.899 to 0.938, which indicated that the models inTable 4 had strong interpretation function upon profile soil salinity.

The best interpretation models were between 40- 80 cm depthwith the RMSE ranging between 0.196 and 0.204 dS m−1. The MEshowed that there was a very small positive deviation (or underes-timation) for interpreting ECe values in 40–100 cm depth, whereasthere was a larger positive deviation in 0–40 cm depth.

The scatter plots of the measured and interpreted ECe for eachdepth were given in Fig. 3, which indicated interpreted ECe of the

top four layers (Fig. 3a–d) were with better accuracy, whereas thatof 80–100 cm depth was with good accuracy (Fig. 3e). The resultimplied that the interpretation models shown in Table 4 were opti-

312 G. Liu et al. / Ecological Engineering 94 (2016) 306–314

Fig. 5. 3D Mesh model of the study area.

f soil e

mt

3

iss(tt

t(d

Fig. 6. Three dimensional distribution o

al in interpreting soil salinity of different depth from the surfaceo 100 cm.

.3. Mapping the three-dimensional variation of soil salinity

ECe of the whole profile was reconstructed by models shownn Table 4. Fig. 4 showed the distribution of 3D scatter data andoil electrical conductivity in the study area. Spatial variety of soilalinity (Fig. 4) is similar with the ECa map for four EMI modesFig. 2). High ECe points were located in the north and the east ofhe study area, whereas points with low ECe were in the west andhe south parts.

3D Mesh model of the study area (Fig. 5) was built by 3D scat-er data. Then three dimensional distribution of ECe in study areaFig. 6) was made using the ordinary kriging based on 3D scatterata and 3D Mesh model.

lectrical conductivity in the study area.

Fig. 6 showed that regions with high ECe were located in thenorth and the east of the study area, especially in the corner ofthe east (red region of Fig. 6), whereas regions with low ECe werein southwest part. On the basis of spatial variety of ECe in thewhole study area, spatial distribution of ECe for the three typicalcross sections created by Create cross section tool in GMS software,were shown in Fig. 7. These three typical sections were A–A′, B–B′

and C–C′, and the specific locations of the 3 sections were shownin Fig. 7. A–A′ section displayed that ECe decreased at first, thenincreased from west to east. B-B’ section showed that ECe grad-ually increased from west to east. The region with high ECe wasin the north part of C–C′, whereas the region with low ECe was inthe south part of C–C′. The change of profile soil salinity allover

the study area was consistent with ECe of the top depth, whereECe of the top depth was high and that of the low depth was alsohigh.

G. Liu et al. / Ecological Engineering 94 (2016) 306–314 313

ns of 3

4

4

aroacrcrisar

sposrasAtt

4

wtttphom

ui

Fig. 7. Typical cross sectio

. Discussion

.1. Accuracy of the interpretation models

We constructed one less complicated and higher accuratepproach for assessing regional scale soil salinity compared toemote sensing (Singh, 2011; Malone et al., 2009), on the basisf the significant relationship between ECa measured with EMInd soil salinity (ECe) for an entire region. Most researches indi-ated that digital soil mapping with remote sensing was not easy toeach a good accuracy with R2> 70%, especially in interpreting soilharacteristics for deep depth. Taghizadeh-Mehrjardi et al. (2014)eported the model of the vertical and lateral variation of soil salin-ty using a combination of regression tree analysis and equal-areamoothing splines in a 72,000 ha area located in the central Irannd modeled validation of the interpretation models at each layer,esulting in R2 ranging from 78% (0–15 cm) to 11% (60–100 cm).

The accuracy of soil salinity interpretation models for differentoil layers taking ECa measured by EMI and ECe of 0–15 cm as inde-endent variables was quite good, which was better or similar tother researchers’ findings (Corwin and Lesch, 2014). The regres-ion modeling approach performs well and represents a viableegional-scale calibration technique. Its simplicity is particularlyppealing. Corwin and Lesch (2014) reported a simplified regional-cale electromagnetic induction-salinity calibration model usingNOCOVA modeling techniques with good accuracy. Interpreta-

ion models of this study performed better for the topsoil (0–80 cm)han the deep depth (>80–100 cm).

.2. Spatial variation of soil salinity

The salinity map (Fig. 6) showed that most of the saline soilsere located in the southeast corner of the study area, namely,

he region of lower elevation. This is likely due to the fact thathe region is located in the fan vertex area of Yellow River delta;he low and flat terrain leads to high underground water level andoor drainage. Furthermore, soil salinity is closely related to theigh groundwater mineralization. The soils of the southeast cornerf this study area are coastal chloride saline soil with low organic

atter.In order to manage saline soils effectively, it is necessary to

nderstand the spatial variation of the vertical and lateral soil salin-ty. Consequently, three-dimensional kriging approach based on

D soil body in study area.

3D scatter data, 3D mesh model, and ECa from EMI survey, is asimple and useful mean to interpret the spatial variation of soilsalinity in the Yellow River delta. The result provided basic knowl-edge for restoration and sustainability assessment of saline landand ecosystem of the study area.

5. Conclusions

In this study, we attempted to investigate three-dimensionalvariation of soil salinity in the Yellow River delta. Interpretationmodels created on the basis of ECa under four operation modesof EM38-MK2, were used to build the map for each depth (0–15,>15–40, >40–60, >60–80 and >80–100 cm, respectively). Then, ordi-nary kriging approach based on 3D scatter data and 3D mesh modelwas used to establish the three dimensional distribution of soilsalinity in the Yellow River delta. The main conclusions were:

Soil salinity interpretation models for different depth, taking ECaunder different EMI modes and ECe of 0–15 cm as independent vari-ables, were reliable, and thus one simple but accurate method wasbrought out for regional soil salinity mapping. The interpretationmodels performed better for the topsoil (0–80 cm) than the deepdepth (>80–100 cm).

The three-dimensional ordinary kriging approach is a reliablemethod to interpret spatial variety of soil salinity across the lowerYellow River Delta. Regions with high soil salinity were located inthe north and the east of the study area, whereas regions with lowsalinity were in the south and the west. Soil salinity decreased withthe increasing distance to the coastline. Electrical conductivity of1:5 soil to water extract (ECe) varied from 0.965 to 1.872 dS m−1 forall sampled soil profiles, and the salinity of topsoil was the highestindicating that soil soluble salts accumulated to the soil surface viacapillary transporting.

Acknowledgements

The authors are grateful for the financial support of the Agricul-tural Science and Technology Autonomous Innovation Foundationof Jiangsu Province (No. CX(15)1005), the National Science and

Technology Basic Work Project (No. 2015FY110500), the Key Tech-nology R&D Program of Jiangsu Province (No. BE2013357) theNational Natural Science Foundation of China (No. 41171178,41171181) and the National Basic Research Program of China

3 nginee

(t

R

A

A

A

B

C

C

C

C

C

D

E

F

F

F

F

F

H

H

J

K

K

L

L

Zhang, T.T., Zeng, S.L., Gao, Y., Ouyang, Z.T., Li, B., Fang, C.M., et al., 2011. Assessing

14 G. Liu et al. / Ecological E

2013CB430403). We also acknowledge the valuable comments ofhe editors and anonymous reviewers.

eferences

dam, I., Michot, D., Guero, Y., Soubega, B., Moussa, I., Dutin, G., Walter, C., 2012.Detecting soil salinity changes in irrigated Vertisols by electrical resistivityprospection during a desalinisation experiment. Agric. Water Manage. 109,1–10.

kramkhanov, A., Brus, D.J., Walvoort, D.J.J., 2014. Geostatistical monitoring of soilsalinity in Uzbekistan by repeated EMI surveys. Geoderma 213 (SI), 600–607.

mezketa, E., 2006. An integrated methodology for assessing soil salinization, aprecondition for land desertification. J. Arid Environ. 67 (4), 594–606.

arbiéro, L., Cunnac, S., Mané, L., Laperrousaz, C., Hammecker, C., Maeght, J.L., 2001.Salt distribution in the Senegal middle valley: analysis of a saline structure onplanned irrigation schemes from N’Galenka creek. Agric. Water Manage. 46 (3),201–213.

ao, S.K., Chen, K.L., Cao, G.C., Zhang, L., Ma, J., Yang, L., et al., 2011. The analysis ofcharacteristic and spatial variability or soil organic matter and organic carbonaround Qinghai Lake. Procedia Environ. Sci. 10 (Part A), 678–684.

emek, B., Guler, M., Kilic, K., Demir, Y., Arslan, H., 2007. Assessment of spatialvariability in some soil properties as related to soil salinity and alkalinity inBafra plain in northern Turkey. Enviorn. Monit. Assess. 124 (1–3), 223–234.

orwin, D.L., Lesch, S.M., 2005. Characterizing soil spatial variability with apparentsoil electrical conductivity I. Survey protocols. Comput. Electron. Agric. 46(1–3), 103–133.

orwin, D.L., Lesch, S.M., 2014. Simplified regional-scale electromagneticinduction- salinity calibration model using ANOCOVA modeling techniques.Geoderma 230, 288–295.

ui, B.S., Yang, Q.C., Yang, Z.F., Zhang, K.J., 2009. Evaluating the ecologicalperformance of wetland restoration in the Yellow River Delta, China. Ecol. Eng.35 (7), 1090–1103.

ing, J., Yu, D., 2014. Monitoring and evaluating spatial variability of soil salinity indry and wet seasons in the Werigan–Kuqa Oasis China, using remote sensingand electromagnetic induction instruments. Geoderma 235, 316–322.

lbasiouny, H., Abowaly, M., Abu Alkheir, A., Gad, A.A., 2014. Spatial variation ofsoil carbon and nitrogen pools by using ordinary Kriging method in an area ofnorth Nile Delta, Egypt. Catena 113, 70–78.

an, X.M., Pedroli, B., Liu, G.H., Liu, H.G., Song, C.Y., Shu, L.C., 2011. Potential plantspecies distribution in the Yellow River Delta under the influence ofgroundwater level and soil salinity. Ecohydrology 4 (6), 744–756.

an, X., Pedroli, B., Liu, G., Liu, Q., Liu, H., Shu, L., 2012. Soil salinity development inthe yellow river delta in relation to groundwater dynamics. Land Degrad. Dev.23 (2), 175–189.

ang, H.L., Liu, G.H., Kearney, M., 2005. Georelational analysis of soil type, soil saltcontent, landform, and land use in the Yellow River Delta, China. Environ.Manage. 35 (1), 72–83.

arzamian, M., Santos, F.A.M., Khalil, M.A., 2015. Application of EM38 and ERTmethods in estimation of saturated hydraulic conductivity in unsaturated soil.J. Appl. Geophys. 112, 175–189.

riedman, S.P., 2005. Soil properties influencing apparent electrical conductivity: areview. Comput. Electron. Agric. 46 (1–3), 45–70.

eil, K., Schmidhalter, U., 2012. Characterisation of soil texture variability usingthe apparent soil electrical conductivity at a highly variable site. Comput.Geosci. U. K. 39, 98–110.

errero, J., Netthisinghe, A., Hudnall, W.H., Pérez-Coveta, O., 2011. Electromagneticinduction as a basis for soil salinity monitoring within a Mediterraneanirrigation district. J. Hydrol. 405 (3–4), 427–438.

iang, H., Bai, Y., Du, H., Hu, Y., Rao, Y., Chen, C., et al., 2016. The spatial and seasonalvariation characteristics of fine roots in different plant configuration modes innew reclamation saline soil of humid climate in China. Ecol. Eng. 86, 231–238.

arlen, D.L., Tomer, M.D., Neppel, J., Cambardella, A., 2008. A preliminarywatershed scale soil quality assessment in north central Iowa, USA. Soil TillageRes. 99 (2), 291–299.

itamura, Y., Yano, T., Honna, T., Yamamoto, S., Inosako, K., 2006. Causes offarmland salinization and remedial measures in the Aral Sea Basin-Research onwater management to prevent secondary salinization in rice-based croppingsystem in arid land. Agric. Water Manage. 85 (1–2), 1–14.

esch, S.M., Corwin, D.L., Robinson, D.A., 2005. Apparent soil electrical conductivitymapping as an agricultural management tool in arid zone soils. Comput.

Electron. Agric. 46 (1–3), 351–378.

i, X.B., Kang, Y.H., Wan, S.Q., Chen, X.L., Chu, L.L., Xu, J.C., 2015. First andsecond-year assessments of the rapid reconstruction and re-vegetationmethod for reclaiming two saline–sodic, coastal soils with drip-irrigation. Ecol.Eng. 84, 496–505.

ring 94 (2016) 306–314

Malone, B.P., McBratney, A.B., Minasny, B., Laslett, G.M., 2009. Mapping continuousdepth functions of soil carbon storage and available water capacity. Geoderma154 (1–2), 138–152.

Ouni, Y., Lakhdar, A., Scelza, R., Scotti, R., Abdelly, C., Barhoumi, Z., et al., 2013.Effects of two composts and two grasses on microbial biomass and biologicalactivity in a salt-affected soil. Ecol. Eng. 60, 363–369.

Ouni, Y., Albacete, A., Cantero, E., Lakhdar, A., Abdelly, C., Pérez-Alfocea, F., et al.,2014. Influence of municipal solid waste (MSW) compost on hormonal statusand biomass partitioning in two forage species growing under saline soilconditions. Ecol. Eng. 64, 142–150.

Qadir, M., Oster, J.D., 2004. Crop and irrigation management strategies forsaline-sodic soils and waters aimed at environmentally sustainableagriculture. Sci. Total Environ. 323 (1–3), 1–19.

Rady, M.M., 2011. Effect of 24-epibrassinolide on growth, yield, antioxidant systemand cadmium content of bean (Phaseolus vulgaris L.) plants under salinity andcadmium stress. Sci. Hortic. 129 (2), 232–237.

Rawlani, A.K., Sovacool, B.K., 2011. Building responsiveness to climate changethrough community based adaptation in Bangladesh. Mitig. Adapt. StrategyGlob. Change 16 (8), 845–863.

Rhoades, J.D., Ingvalson, R.D., 1971. Determining salinity in field soils with soilresistance measurements. Soil Sci. Soc. Am. Proc. 35 (1), 54–60.

Rhoades, J.D., van Schilfgaarde, J., 1976. An electrical conductivity probe fordetermining soil salinity. Soil Sci. Soc. Am. J. 40 (5), 647–650.

Rhoades, J.D., Shouse, P.J., Alves, W.J., Manteghi, N.A., Lesch, S.M., 1990.Determining soil salinity from soil electrical conductivity using differentmodels and estimates. Soil Sci. Soc. Am. J. 54 (1), 46–54.

Rietz, D.N., Haynes, R.J., 2003. Effects of irrigation-induced salinity and sodicity onsoil microbial activity. Soil Biol. Biochem. 35 (6), 845–854.

Shukla, S.K., Singh, K., Singh, B., Gautam, N.N., 2011. Biomass productivity andnutrient availability of Cynodon dactylon (L.) Pers. growing on soils of differentsodicity stress. Biomass Bioenergy 35 (8), 3440–3447.

Singh, K., Singh, B., Singh, R.R., 2013. Effect of land rehabilitation onphysicochemical and microbial properties of a sodic soil. Catena 109, 49–57.

Singh, A., 2011. Estimating long-term regional groundwater recharge for theevaluation of potential solution alternatives to waterlogging and salinisation. J.Hydrol. 406 (3–4), 245–255.

Sudduth, K.A., Kitchen, N.R., Wiebold, W.J., Batchelor, W.D., Bollero, G.A., Bullock,D.G., et al., 2005. Relating apparent electrical conductivity to soil propertiesacross the north-central USA. Comput. Electron. Agric. 46 (1–3), 263–283.

Taghizadeh-Mehrjardi, R., Minasny, B., Sarmadian, F., Malone, B.P., 2014. Digitalmapping of soil salinity in Ardakan region, central Iran. Geoderma 213 (SI),15–28.

Tang, X.L., Mu, X.M., Shao, H.B., Wang, H.Y., Brestic, M., 2014. Globalplant-responding mechanisms to salt stress: physiological and molecular levelsand implications in biotechnology. Crit. Rev. Biotechnol. Early Online, 1–13.

USSLS, 1969. Diagnosis and Improvement of Saline and Alkali Soils. AgricultureHandbook United States Salinity Laboratory Staff. US Government PrintingOffice, Washington, pp. 60.

Urdanoz, V., Araguees, R., 2011. Pre- and post-irrigation mapping of soil salinitywith electromagnetic induction techniques and relationships with drainagewater salinity. Soil Sci. Soc. Am. J. 75 (1), 207–215.

Wang, Y.G., Xiao, D.N., Li, Y., Li, X.Y., 2008. Soil salinity evolution and itsrelationship with dynamics of groundwater in the oasis of inland river basins:case study from the Fubei region of Xinjiang Province, China. Enviorn. Monit.Assess. 140 (1–3), 291–302.

Wichelns, D., Qadir, M., 2014. Achieving sustainable irrigation requires effectivemanagement of salts, soil salinity, and shallow groundwater. Agric. WaterManage. 157, 31–38.

Yan, K., Shao, H.B., Shao, C.Y., Chen, P., Zhao, S.J., Brestic, M., et al., 2013.Physiological adaptive mechanisms of plants grown in saline soil andimplications for sustainable saline agriculture in coastal zone. Acta Physiol.Plant. 35 (10), 2867–2878.

Yao, R.J., Yang, J.S., Liu, G.M., 2007. Calibration of soil electromagnetic conductivityin inverted salinity profiles with an integration Method1. Pedosphere 17 (2),246–256.

Ye, Q.H., Liu, G.H., Tian, G.L., Chen, S.L., Huang, C., Chen, S.P., Liu, Q.S., Chang, J., Shi,Y., 2004. Geospatial-temporal analysis of land-use changes in the Yellow RiverDelta during the last 40 years. Sci. China Ser. D Earth Sci. 47 (11), 1008–1024.

Yuan, B.C., Li, Z.Z., Liu, H., Gao, M., Zhang, Y.Y., 2007. Microbial biomass and activityin salt affected soils under arid conditions. Appl. Soil Ecol. 35 (2), 319–328.

impact of land uses on land salinization in the Yellow River Delta, China usingan integrated and spatial statistical model. Land Use Policy 28 (4), 857–866.

Zhang, H.S., Zai, X.M., Wu, X.H., Qin, P., Zhang, W.M., 2014. An ecologicaltechnology of coastal saline soil amelioration. Ecol. Eng. 67, 80–88.