This article is an openaccess article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).
The objectives of this study were to investigate the degree of spatial variability and variance structure of salinization parameters using classical and geostatistical method in Songnen Plain of China, which is one of largest salinesodic areas in the World, and to analyze the relationship between salinization parameters, including soil salinity content (SC), electrical conductivity (EC), sodium adsorption ratio (SAR), and pH, and seven environmental factors by Pearson and stepwise regression analysis. The environmental factors were ground elevation, surface ponding time, surface ponding depth, and soil moistures at four layers (0–10 cm, 10–30 cm, 30–60 cm, and 60–100 cm). The results indicated that SC, EC, and SAR showed great variations, whereas pH exhibited low variations. Four salinization parameters showed strongly spatial autocorrelation resulting from the compound impact of structural factors. The empirical semivariograms in the four parameters could be simulated by spherical and exponential models. The spatial distributions of SC, EC, SAR and pH showed similar patterns, with the coexistence of high salinity and sodicity in the areas with high ground elevation. By Pearson analysis, the soil salinization parameters showed a significant positive relationship with ground elevation, but a negative correlation with surface ponding time, surface ponding depth, and soil moistures. Both correlation and stepwise regression analysis showed that ground elevation is the most important environmental factor for spatial variation of soil sanilization. The results from this research can provide some useful information for explaining mechanism of salinization process and utilization of salinesodic soils in the Western Songnen Plain.
Soil salinity and sodicity have become an increasingly acute problem in Songnen Plain in northeast China, one of three largest salinesodic areas in the World [
Saline and sodic soils often display high spatial variability in soil salinity and sodicity at a field scale [
The study on spatial variation of soil salinity and sodicity in Songnen Plain has been focusing on qualitative description and classical statistics analyses [
The objective of this research was to apply geostatistical approaches to determine the degree of spatial variability and variance structure of salinization parameters and evaluate the effects of environmental factors such as micro topography and hydrological elements. The results from this research are of great importance for a better understanding of the mechanism of soil salinization and rational utilization of salinesodic land resources.
The experiment is conducted at the Da’an Sodic Land Ecological Experiment Station of the Northeast Institute of Geography and Agroecology, Chinese Academy of Sciences. This region lies in 123º50′27″–123º51′31″ of east longitude and 45º35′58″–45º36′28″ of north latitude. The study area features a temperate continental monsoon climate, with an average annual precipitation of 413.7 mm and an average annual evaporation of 1,696.9 mm. The average annual temperature is 4.7 centigrade with the lowest temperature of −17.7 centigrade in January. Soil is generally frozen from late November and will thaw completely until late May or early June of the next year. The experimental station is surrounded by a low flood plain. Currently, the moderate and mild salinesodic soils are gradually degraded into moderate and severe salinesodic soils.
A representative salinesodic area measuring 100 × 100 m was selected within the experimental station for soil sampling. A border (40 cm in height and 40 cm in width) was constructed along the boundary of the experiment area to collect the surface runoff after rainfall events. The experimental area was divided into 10 × 10 m grid squares, and 40 grids were randomly selected as sampling points (
The field experiment was conducted during 1 May to 1 October for three consecutive years (2005 to 2007). The experiment was carried out during May to October because the evapotranporation is the largest in the summer months. Due to the large evapotransporation, the ascending motion of capillary water is generally greater than the descending motion. This facilitates the salt in soil pedon and shallow groundwater to build up on the soil surface [
The surface ponding time and surface ponding depth were observed every five days for the norainfall period, and every day after the rainfall. At the end of May, July and September, the soil water contents were measured. Soil water contents were observed at four depths of soil layers: 0–10 cm, 10–30 cm, 30–60 cm, and 60–100 cm. Ground elevation was measured by DSN232 balance level.
At the end of September, soil samples were collected to analyze the ions. Soil samples were collected at the depth of 0–10 cm at the 40 sampling points for laboratory analysis. The measured physical and chemical parameters including pH, EC, Na^{+}, K^{+}, Ca^{2+}, Mg^{2+}, CO_{3}^{2−}, HCO_{3}^{−},Cl^{−}, SO_{4}^{2−}. All soil samples were airdried and then passed a 1mm roundhole sieve for chemical analyses. Soluble salt estimates were based on 1:5 soilwater extracts. The pH and EC of the extracts were determined using a pH meter and a conductivity meter, respectively [
Four soil salinization parameters were analyzed for identifying the outliers and carrying out different transformations such as log normal and square root to ensure a normal distribution. Then semivariogram parameters for each theoretical model such as spherical, exponential, linear, and Gaussian were generated. Selection of the bestfitting model was based on regression statistics such as minimum Residual Sums of Squares (RSS) and maximum determination coefficient (R^{2}). The corresponding sill, nugget, and range values of the bestfitting theoretical model were calculated. After selection of the suitable theoretical model and the corresponding semivariogram parameters, spatial variability maps were generated for these four parameters of soil salinization using ordinary kriging. The Pearson correlation and stepwise regression analysis of the data was based on values for salinization parameters and water content of three years.
Geostatistics aims at providing quantitative descriptions of natural variables distributed in space and time [
The experimental semivariogram γ(h) can be fitted by different theoretical models such as spherical, exponential, linear, or Gaussian to determine three semivariogram parameters: the nugget (C_{0}), the sill (C_{0} + C), and the range (A_{0}) [
Ordinary kriging was used to generate the spatial distribution of these four salinization parameters. The parameter values at the unsampled grids were estimated based on the values at the 40 sampling grids by the ordinary kriging method, which provides the best linear unbiased estimate of a regionalized variable at an unsampled location. Ordinary kriging assumes that the mean of the process is constant and invariant within the spatial domain. A linear combination of available sample values is used for ordinary kriging estimation. Weights, the coefficients of this linear combination, are dependent on two factors: the distance between the sample point and the estimated point and the spatial structure of the variable [
The descriptive statistics of the spatial distributions of EC, SAR, SC, and pH, are given by
Frequency distributions and the Kolmogorv–Smirov test for normality showed that EC, SC, and SAR were not normally distributed (
A semivariogram for each soil salinization parameter was developed to quantify the spatial variation of soil salinization. The nugget, sill, and range values of the bestfit theoretical models for soil salinization parameters are given in
The ratio of nugget to total semivariance, expressed as a percentage, was used to classify spatial dependence: a ratio of <25% indicated strong spatial dependence; between 25% and 75% indicated moderate spatial dependence; and >75% indicated weak spatial dependence [
Soil salinization parameters exhibited strip and block patterns (
In the study area, high saline and sodic areas codistribute with nonor low saline and sodic areas, based on Chinese salinzation classification [
Pearson correlation analysis was conducted to evaluate the effects of environmental factors on these four soil salinization parameters. The investigated environmental factors included ground elevation (H), surface ponding time (d), surface ponding depth (h), and soil moistures at four layers [0–10 cm (W_{0–10}), 10–30 cm (W_{10–30}), 30–60 cm (W_{30–60}), and 60–100 cm (W_{60–100})]. The results of correlation analysis are given by
The results of stepwise regression analysis also revealed that ground elevation is the most important factor affecting the spatial distribution of salinity and socidity at field scale (
Minor changes in ground elevation can cause large variations in soil salinity. The higher salinity and soidicity occurred in the grids located in higher elevation, and
High levels of sodium restrict waterholding capacity and prevent flocculation of soil particles. The process of soil clay particles gathering together into small aggregates is called flocculation, which allows water to penetrate between the groups of soil particles and provide moisture at deeper depths. During wet conditions the individual clay particles will overlap each other randomly when sodium levels are high enough to prevent flocculation. This will prevent water penetration through the high sodium layer [
A geostatistical approach was applied to the Western Songnen Plain to investigate the spatial distributions of salinization parameters and the relationship between salinization parameters and environmental factors. The calculated coefficients of variation for EC, SAR, and SC ranged from 72% to 197%, indicating high spatial variation patterns, whereas pH showed a pattern of low spatial variations with a coefficient of variation from 7% to 9%. SC, EC, SAR and pH showed a strong spatial autocorrelation resulting from the compound impact of structural factors. The empirical semivariograms of the four parameters could be simulated by spherical and exponential models.
Correlation and stepwise analysis showed that soil salinization parameters were related in a positive linear fashion to ground elevation, but negatively linearly related to surface ponding time, surface ponding depth, and soil moisture. Ground elevation is the key factor affecting the spatial distribution of soil salinity and sodicity through change hydrological process at the field scale.
This research is supported by the National Natural Science Foundation of China (No. 41071022) and the Knowledge Innovation Project (KIP) of Chinese Academy of Sciences (CAS) (No. KZCX2YWQ0623). We also wish to acknowledge the support of the Da’an Sodic Land Experimental Station.
Distribution of measuring points and ground elevation in the experiment site. The numbers in red indicate the randomly selected sampling points. The unit of ground elevation is cm.
Bestfit semivariograms of soil salinization parameters from 2005 to 2007 year. SC is soil salinity content (mg·kg^{−1}); EC is electrical conductivity (μs·cm^{−1}); and SAR is sodium adsorption ratio.
Maps of kriging estimations of four salinization parameters. SC is soil salinity content (mg·kg^{−1}); EC is electrical conductivity (μs·cm^{−1}); and SAR is sodium adsorption ratio.
Saline accumulate in subtly undulating landscape.
Descriptive statistics on salinization patameters from 2005 to 2007 year.
0.81  pH  2005  7.87  10.70  9.51  0.09  −0.44  −1.03  N  0.212 
0.63  2006  8.06  10.90  9.88  0.06  −0.50  0.23  N  0.411  
0.70  2007  8.25  10.60  9.48  0.07  0.10  −1.38  N  0.332  
 
400

EC (μs·cm^{−1})  2005  82.8  2710  392  1.02  5.22  30.50  LN  0.596 
411

2006  78  2776  400  1.02  5.18  30.20  LN  0.591  
919  2007  123  4730  660  1.39  3.51  12.80  LN  0.764  
 
3,833  SC (mg·kg^{−1})  2005  588  13,616  5,212  0.74  5.22  30.50  N  0.596 
3,556  2006  1,134  14,687  5,366  0.72  5.18  30.20  N  0.591  
5,105  2007  874  22,826  4,068  1.25  2.93  8.07  LN  0.759  
 
12.60  SAR  2005  3.09  86.90  10.60  1.19  5.98  37.00  LN  0.248 
13.40  2006  3.91  91.10  12.00  1.12  5.51  32.90  LN  0.385  
70.90  2007  5.55  306.00  35.90  1.97  3.26  9.66  LN  0.060 
Note: EC is electrical conductivity; SC is soil salinity content; SAR is sodium adsorption ratio; Min is the minimum value; Max is the maximum value; Mean is the average value; SD is the standard deviation; CV is the calculated coefficient of variation; Ske is the calculated skewness; Kur is the calculated Kurtosis; KS is the coefficient of KolmogoravSmirnow; DP is the distribution pattern; N is normal distribution; and LN is normal distribution after logarithmic transformation.
Summary of bestfit models for salinization parameters.
pH  2005  Spherical  0.00001  0.007  43.6  0.200  0.875  0.0000007 
pH  2006  Spherical  0.00001  0.004  44.0  0.300  0.891  0.0000015 
pH  2007  Spherical  0.00107  0.005  76.2  20.000  0.888  0.0000018 
LnEC  2005  Exponential  0.00001  0.010  150.0  0.100  0.906  0.0000044 
LnEC  2006  Exponential  0.00032  0.011  124.0  0.100  0.884  0.0000061 
LnEC  2007  Exponential  0.00000  0.026  212.0  0.000  0.965  0.0000083 
SC  2005  Exponential  0.00100  0.930  81.0  0.100  0.862  0.0870000 
SC  2006  Exponential  0.00100  0.761  64.8  0.100  0.860  0.0510000 
LnSC  2007  Exponential  0.00001  0.019  285.0  0.100  0.960  0.0000048 
LnSAR  2005  Spherical  0.00013  0.017  39.6  0.800  0.606  0.0001059 
LnSAR  2006  Spherical  0.00000  0.021  29.9  0.000  0.756  0.0006431 
LnSAR  2007  Spherical  0.00010  0.053  58.4  0.002  0.914  0.0002494 
Note: SC is soil salinity content (mg·kg^{−1}); EC is electrical conductivity (μs·cm^{−1}); SAR is sodium adsorption ratio; R^{2} is the determination coefficient; and RSS is residual sums of squares.
Pearson analysis between salinization parameters and environmental factors.
EC  0.811 
−0.46 
−0.486 
−0.273  −0.387 
−0.510 
−0.379 
pH  0.592 
−0.559 
−0.572 
−0.532 
−0.633 
−0.468 
−0.529 
SAR  0.744 
−0.324 
−0.353 
−0.165  −0.249  −0.400 
−0.250 
SC  0.688 
−0.638 
−0.624 
−0.589 
−0.557 
−0.559 
−0.351 
Note: SC is soil salinity content (mg·kg^{−1}); EC is electrical conductivity (μs·cm^{−1}); SAR is sodium adsorption ratio; H is ground elevation(cm); h is surface ponding depth(cm); d is surface ponding time(day); W_{0–10}, W_{10–30}, W_{30–60} and W_{60–100} are the soil moistures at 0–10, 10–30, 30–60, and 60–100 cm, respectively;
p < 0.01;
p < 0.05.
Regression model between salinization parameters and environmental factors.
EC = 60.809H − 108.559  0.658  0.000 
EC = 82.821H + 13.710d − 613.38  0.713  0.000 
SC = 375.272H + 2000.542  0.446  0.000 
SC = 279.3H − 32716.3W_{60–100} + 10779.1  0.531  0.000 
SAR = 1.266H − 3.199  0.542  0.000 
SAR = 2.037H + 0.481d − 20.896  0.687  0.000 
pH = −7.227W_{10–30} + 11.361  0.401  0.000 
pH = 0.034H − 5.108W_{10–30} + 10.51  0.498  0.000 
Note: SC is soil salinity content (mg·kg^{−1}); EC is electrical conductivity (μs·cm^{−1}); SAR is sodium adsorption ratio; R^{2} is the determination coefficient, and Sig is the statistic significance level.