JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 25 May 2007
Published in J Environ Qual 36:935-942 (2007)
DOI: 10.2134/jeq2006.0184
© 2007 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lu, P.
Right arrow Articles by Wu, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lu, P.
Right arrow Articles by Wu, J.
Agricola
Right arrow Articles by Lu, P.
Right arrow Articles by Wu, J.
Related Collections
Right arrow Watershed and Landscape Processes
Right arrow Nitrogen
Right arrow Phosphorus
Right arrow Ecological Risk Assessment
Right arrow Soil Fertility and Productivity

TECHNICAL REPORTS

Ecological Risk Assessment

Geostatistical Analysis and Risk Assessment on Soil Total Nitrogen and Total Soil Phosphorus in the Dongting Lake Plain Area, China

Peng Lua, Yirong Sub, Zheng Niua and Jinshui Wub,*

a The State Key Lab. of Remote Sensing Science, Jointly Sponsored by the Inst. of Remote Sensing Applications, Chinese Academy of Sciences and Beijing Normal Univ., Beijing 100101, P.R. China
b The Inst. of Subtropical Agriculture, Chinese Academy of Sciences, Changsha 410125, P.R. China

* Corresponding author (jswu{at}isa.ac.cn)

Received for publication May 7, 2006.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Nonpoint-source pollution and water body eutrophication have become increasing concerns for scientists and policymakers. Nitrogen and phosphorus affect environmental pollution, especially lake eutrophication. To assess the environmental risk of soil total nitrogen (TN) and total phosphorus (TP) pollution, a typical ecological unit of Dongting Lake plain was selected as the experimental site. To verify the stationary of the data, a moving windows technique was adopted. Our results showed that Box-Cox transformation achieved normality in the data set and dampened the effect of outliers. The best theoretical model for semivariogram of TN and TP was a spherical model. The ordinary kriging estimates of TN and TP concentrations were mapped. The integrative comparisons of semivariogram parameters with different trends to the kriging prediction errors of TN and TP indicated that the two-order trend is preferable. Kriging SDs provided valuable information that will increase the accuracy of TN and TP mapping. The probability kriging method is useful to assess the risk of N and P pollution by providing the conditional probability of N and P concentrations exceeding the threshold concentrations of 3.2 and 1.05 g/kg, respectively. The probability distribution of TN and TP at different levels will be helpful to conduct risk assessment, optimize fertilization, and control the pollution of N and P.

Abbreviations: KSD, kriging SD • TN, soil total nitrogen • TP, soil total phosphorus


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
DURING the last 50 yr, global nitrogen (N) fertilizer applications have increased steadily, rising almost 20-fold to the present rate of ~1011 kg yr–1 (Glass, 2003). Because of the annual increase in the usage of N and phosphorous (P)-based fertilizers, soil total N (TN) and total soil phosphorus (TP) contents are exceeding plant growth requirements, leading to regular surpluses of fertilizer. High N and P fertilization rates generally result in low N- and P-use efficiency and high N loss (Li et al., 1999). On a global basis, the apparent N recovery rate is as low as 33% (Raun and Johnson, 1999), and P-use efficiency is much lower (~10%) (Shenoy and Kalagudi, 2005). For example, in Hunan province, N and P surpluses are 104% and 72%, which enhances the risk of soil N and P entering water bodies and seriously threatening the water quality in drainage areas. Liu (1987) reported that 84 to 87% of TN entering water bodies came from agriculture, and approximately 3.0 to 4.0 million tons of P2O5 entered rivers or lakes.

Therefore, worldwide attention has focused on the ecological effects of excessive N and P and on optimizing N and P fertilization. Because N and P are main factors of nonpoint-source pollution and lake eutrophication, spatial distribution of TN and TP at different regional scales, especially at the landscape scale, plays an important role in revealing the questions of environment and agriculture. Nonpoint-source pollution in agriculture is also of great concern to the government and scientists. Understanding the spatial variation of TN and TP at the landscape scale is important to increase production, to increase our understanding of the beginning and growth of nonpoint-source pollution, and to reduce its potential harmful effects.

Since the 1970s, geostatistics have provided an advanced methodology that allows spatial interpolation and facilitates the quantification of the spatial features of soil parameters (Burgess and Webster, 1980). Subsequently, kriging techniques have been widely applied in soil, ecological, and environmental sciences to analyze the spatial patterns and variability of soil nutrients. Recent examples of kriging in soil nutrients include studies by Burrough (1983), Zhang et al. (1992), Cambardella et al. (1994), Yanai et al. (2003), Gallardo (2003), Ouyang et al. (2006). These reports primarily focused on the application of ordinary kriging. However, the probabilities of unknown points and their spatial distribution at a certain threshold value have been rarely reported. McGrath et al. (2004) used probability kriging to estimate the risk of soil lead in Ireland. Zhang et al. (2004) used this approach to map soil organic carbon in a rehabilitating ecosystem. In addition, geostatistics and the Geographic Information System (GIS) have become useful tools to study spatial uncertainty and risk assessment (Goovaerts, 2001).

Based on field survey and data analysis, we studied the spatial variability of TN and TP in a typical area of Dongting Lake plain, applying geostatistics combined with GIS methods. Geostatistic and GIS techniques were used to explore the spatial structure of the studied variables (e.g., soil organic carbon, TN, TP, and soil lead) and to provide a basis for risk assessment. By mapping the distribution of TN and TP and risk assessment, the objectives of the present study were to provide a guide for optimizing fertilization in agricultural production, reduce the pollution of underground water and lakes by ground runoff and infiltration, and provide the theoretical basis for agricultural and eco-environmental management for the Dongting Lake plain area.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Characteristics of Study Area and Sampling and Measurements
The study was conducted in a typical area in Dongting Lake plain, the "Hometown of fish and rice," located in the north of Hunan Province, covering approximately 4 km2. It has a humid monsoonal climate with an annual average air temperature of 16.5°C and annual average precipitation of 1313 mm. This site is an alluvial plain with paddy soil and fluvo-aquic soil developed from lake sediments and river alluvium. The soil quality in the area has had a decrease due to excessive fertilization in recent years (Li et al., 1991).

In March 2004, about 651 soil samples were collected from the surface layer (0–20 cm) at the study site, according to land use types, such as paddy fields and dry lands. Detailed records were made on the coordinates of the sampling sites by global positioning system. Sampling site geo-positions were transformed into ".shp" format, which is compatible with ArcMap, and underwent ArcInfo projection transformation into plane coordinates, based on meters. The sampling map (Fig. 1 ) was used for geostatistical analysis. Each sample consisted of 15 soil cores (0–20 cm depth, 400 g soil) as a mixture sample taken from within a field whose central point can provide the position of the sample. All the samples were air-dried at room temperature and sieved through a 0.25-mm mesh for sample testing. The soil samples were digested by HClO4–H2SO4, and TP concentrations were measured by the molybdenum blue colorimetric method. Total soil N concentrations were measured with an Element Auto-Analyzer (Vario MAX CN, Elementar, Germany).


Figure 1
View larger version (27K):
[in this window]
[in a new window]

  		Fig. 1. Sampling diagram of soil nutrients in Dongting Lake area in subtropical area, China.

 
Data Transformation and Trend Analysis
A calculation of the semivariances without checking the statistical properties of data should be avoided, and a preliminary data check is necessary (Delcourt et al., 1996). We applied logarithmic transformation and Box-Cox transformation (Box and Cox, 1962; Zhang et al., 1998) to calculate parameters such as skewness, kurtosis, and the significance level of Kolmogorov-Smirnov test and to examine their normality.

The Box-Cox transformation is given by

Formula 1[1]
where y is the transformed value, and x is the value to be transformed. For a given data set (x1, x2, ... xn), {lambda} is estimated based on the assumption that the transformed values (y1, y2, ... yn) are normally distributed. When {lambda} = 0, the transformation becomes the logarithmic transformation (McGrath et al., 2004).

We used the ratio of anisotropy in different directions to describe the characteristic of anisotropy

Formula 2[2]
where K(h) is the ratio of anisotropy in different directions, and {gamma}(h,{theta}1) and {gamma}(h,{theta}2) are the semivariograms in both directions.

Being affected by soil-forming factors, the spatial distribution of soil properties displayed obvious characteristics and anisotropic distribution structures. In general, surface trends are categorized into three types: 0-order, nonsurface trend; 1-order, linear variation; and 2-order, multinomial variation. ArcMap software with the function of trend analysis facilitated the acquisition of characteristic anisotropic parameters of TN and TP and their characteristic surface trends.

Taking anisotropy into consideration, we compared the errors of ordinary kriging interpolation for surface trend parameters (0-order, 1-order, and 2-order). The fitness of the semivariogram model and its parameters were judged according to the following criteria: The absolute value of mean error is closest to 0 viz. the minimum of errors, MSE is closest to 0, the smallest RMSE is the best, and the average SE is closest to RMSE.

Probability Kriging
Probability kriging is a special form of the cokriging procedure wherein only one variable, the indicator transformation, is estimated using two spatial variables, the indicator and uniform transforms (Carr and Mao, 1993). For probability kriging, the indicator code I(x) is assigned as the main variable and the other variable, and the uniform value u(x) is assigned as the auxiliary variable in the cokriging estimator. The uniform value, also called the standardized rank, was reported in detail by Journel and Deutsch (1997) and is defined as

Formula 3[3]
where r denotes the rank of the rth order statistic z(r) located at x, and n is the total number of observations (Goovaerts, 1997; Juang and Lee, 2000).

Then, the semivariogram of I(x) and u(x) is

Formula 4[4]

Formula 5[5]
where h is the distance between both locations, xi and xi + h, and N(h) is the number of pairs for xi and xi + h.

The probability kriging estimator is defined by

Formula 6[6]
where Ic is the indicator transform for threshold c, and N is the number of the nearest neighbors used for estimation. To assure unbiasedness, a sufficient condition (one that is not a necessary condition and occasionally is poor) is to constrain the weights such that

Formula 7[7]
Therefore, the primary variable in this procedure is the indicator transform, and the uniform transform is the secondary variable (Carr and Mao, 1993).

Data Treatment with Computer Software
Raw data were analyzed with different software packages. The descriptive statistical parameters and normality tests were calculated with SPSS (version 13.0). A moving window statistic was investigated by Geostats 3-Plot98 4.60 (IBRAE, Russia). Box-Cox normality plots were computed with Dataplot software. The geostatistical analyses were performed with ILWIS (version 3.1). Maps were produced with ArcGIS8.3 ArcMap (ESRI, Redlands, CA) and its extension of Geostatistical analyst. We used ordinary kriging methods to estimate the unobserved points and to generate contour maps of the observed variables (TN and TP). We also used the probability kriging method to assess the risk of N and P contamination. The interpolated values were transmitted from ASCII data to a GRID graph, which was cut in accordance to the boundaries of the study area into a distribution map of TN and TP.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Moving Windows
The proportional effect is a particular form of heteroscedasticity where the local variance of data is related to their local mean (Goovaerts, 1997). For negatively skewed distributions, the local variance decreases with the local mean (inverse proportional effect). Stationarity of the data must be checked before proceeding into further geostatistical analysis. A moving window statistic can, however, examine the data for a proportional effect. If it is not feasible to separate out unambiguous subregions, a moving window technique can serve as a work-around (Cullmann and Saborowski, 2005).

Twelve non-overlapping 600 by 600 m windows were defined over the study area. The negatively skewed variables (TN) showed a direct relation between local mean and local variance. There was no significant proportional effect for TP (Fig. 2 ).


Figure 2
View larger version (12K):
[in this window]
[in a new window]

  		Fig. 2. Plots of local variances versus local means computed from 600 by 600 m moving windows overlaid across the study area. (A) Total soil nitrogen. (B) Total soil phosphorus.

 
Descriptive Parameters and Normality Examination
Even though normality may not be strictly required, serious violation of normality, such as higher skewness and outliers, can impair the viriogram structure and the kriging results (Gringarten and Deutsch, 2001; McGrath et al., 2004). To acquire more precise estimates in geostatistical analyses and assessments, the primary data need to be preprocessed to achieve normality. Therefore, Box-Cox transformation is often applied in environmental sciences because the data sets do not always follow the lognormal distribution (Zhang et al., 1998).

The TN and TP concentrations were not normally distributed after logarithmic transformation (Table 1). When the data were transformed by Box-Cox, TN and TP concentrations passed the Kolmogorov-Smirnov test (p > 0.05), which suggested that normality had been achieved and could meet the requirements of statistical analysis. We used Box-Cox normality plots to investigate the optimal choice of {lambda} with Dataplot software. The regression coefficients of TN and TP were approximately maximal at {lambda} = 2.0 (Fig. 3 ). The Box-Cox curve of TN rose sharply from {lambda} = –2.0 (r2 = 0.56) until reaching its maximum (close to {lambda} = 2.0; r2 = 0.99). The TP graph appeared flatter, rising from {lambda} = –2.0 (r2 = 0.875) to the maximum at {lambda} = 2.0 (r2 = 0.99). Therefore, we identified the optimum transformation value at {lambda}TN = 2.1 and {lambda}TP = 1.9. Box-Cox transformed data were used in the subsequent kriging interpolation analysis.


View this table:
[in this window]
[in a new window]

  		Table 1. Descriptive statistics of total soil nitrogen and total soil phosphorus in topsoil.

 

Figure 3
View larger version (9K):
[in this window]
[in a new window]

  		Fig. 3. Box-Cox normality plots of (A) total soil nitrogen and (B) total soil phosphorus. The horizontal axis represents the value of the {lambda} transformation parameter. The vertical axis refers to the correlation coefficient.

 
The descriptive statistic analyses of Table 1 were capable of illustrating only the content characteristics of TN and TP rather than the regional characteristics of the whole landscape unit. This means that they cannot illustrate the spatial and structural characteristics and the randomicity of TN and TP. Consequently, it is necessary to apply geostatistical methods together with GIS to solve this problem.

Spatial Variation Analysis of Total Nitrogen and Total Phosphorus
The analysis of isotropic spatial variation is the prerequisite of kriging interpolation to calculate the real values of semivariograms. The semivariograms of TN and TP under isotropy exhibited excellent spatial structure, which could be well fitted with a spherical model. The spatial correlation distances of TN and TP were 322 and 338 m, respectively (Fig. 4 ). The nugget variances were small (C0 = 0.025 and 0.008), indicating that the sampling density could reveal the spatial structure of TN and TP in the study area (McGrath et al., 2004). These results provide a strong support for predicting spatial structure in the spatial interpolation map.


Figure 4
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 4. Isotropic semi-variogram of (A) total soil nitrogen and (B) total soil phosphorus.

 
The ratio of nugget value (C0) to sill value (C0 + C) is an important index for investigating the spatial structures of soil properties. It not only indicates the spatial heterogeneity of regionalized variables but also implies whether regional factors (e.g., natural factors, such as topography and geomorphology) or the nonregional factors (e.g., human factors, such as irrigation and fertilization) play the dominant role in spatial variation (Li et al., 2002). In Fig. 4, the ratio values of C0 to C0 + C of TN and TP were below 0.5, suggesting a strong spatial autocorrelation at the scale of the study, even though it was influenced by small-scale factors such as plant absorption and fertilization (Hu et al., 2004).

To investigate the effects of natural processes on TN and TP, we used a semivariogram/covariance cloud and automatic searching function to analyze the spatial directions of TN and TP. Finally, we calculated the parameters of the spatial variogram and their K(h) in four directions (NE25°, NE70°, NE115°, and NE160°).

The sill and range of TN varying in the four directions indicated a strip anisotropic structure (Table 2). In the directions of NE70° and NE160°, the anisotropy structure of TN was varied within different ranges (Fig. 5 ). The anisotropy ratio (K(h)) was below 1.0 from 0 to 400 m, suggesting that the semivariance of TN in the direction of NE70° was lower than that in NE160° at this range. However, the opposite results were observed at the range of 400 to 1400 m (i.e., the semivariance of TN in the direction of NE70° was greater than that in NE160°). The K(h) of TN was above the isotropy lines at most scales in the directions of NE25° and NE115°. These findings revealed a significant difference in the anisotropy structure of TN, probably due to the strip-shaped land-use structure in the direction of NE25°.


View this table:
[in this window]
[in a new window]

  		Table 2. Theory models and corresponding parameters of total soil nitrogen and total soil phosphorus in four directions.

 

Figure 5
View larger version (17K):
[in this window]
[in a new window]

  		Fig. 5. Anisotropic ratio of (A) total soil nitrogen and (B) total soil phosphorus at different directions. K(h), ratio of anisotropy in different directions.

 
The sill of TP in the direction of NE70° was the same as that in the direction of NE160°, whereas the ranges of the two directions were different, which suggested a geometric anisotropic structure. The sill and range of TP differed in the directions of NE25° and NE115°, displaying a strip anisotropic structure. The K(h) ratios of TP were situated on the isotropy lines at 0 to 1400 m with little fluctuation, exhibiting an isotropy for TP (Fig. 5). This also suggested that human factors, such as fertilization and cultivation management, reduce the spatial heterogeneity of TP.

According to the criteria of trend analysis, we found that TN and TP of Dongting Lake plain applied to 2-order multinomial variation in both directions. Therefore, the following kriging interpolation analysis was based on the semivariogram theory model of the anisotropy structure with 2-order trend parameters.

Spatial Distribution Patterns of Total Nitrogen and Total Phosphorus
Based on the Box-Cox transformed data set, we used ordinary kriging interpolation methods and acquired the spatial distribution maps (Fig. 6 ). We used the reverse process of Box-Cox transformation, taking anisotropy into consideration, with the parameters of the 2-order trend and the greatest estimated samples numbering 20. Using grids with the cell size of 5 by 5 m, we divided the study area into 440 ranks (NE115°, 2200 m long) and 560 rows (NE25°, 2800 m long). Kriging interpolation was subject to effects from the accuracy of the semivariogram simulation and the distribution of samples and the number of searching neighborhoods.


Figure 6
View larger version (35K):
[in this window]
[in a new window]

  		Fig. 6. Spatial distribution pattern of (A) total soil nitrogen (TN) and (B) total soil phosphorus (TP).

 
The interpolation map revealed remarkable differences in the size, shape, and spatial distribution pattern between TN and TP. Regardless of the direction, there was always a distribution gradient in nutrient contents. Based on the nutrient grading standards (Chien et al., 1997), we judged the flow of nutrients within a region and the potential of nonpoint-source pollution. The characteristic of nutrient spatial distribution has a strong relationship with various physical, chemical, and biological processes in the soil at different spatial positions (Robertson and Frekman, 1995).

The interpolation map showed that the directional variation of TN was relatively more evident than that of TP. The spatial distribution of TN in the direction of NE25° exhibited a strip-shaped structure with alternating light and shade and a wave structure that seemed to have been formed by water flow (Fig. 6). This spatial distribution was in accordance with the directional characteristics shown in the semivariogram of TN. The spatial distribution of TP was characterized by sporadically distributed areas of high concentration rather than by a strip-shaped distribution pattern, which helped to confirm the characteristic of inconspicuous direction shown in the semivariogram of TP (Fig. 4). The areas where TN and TP concentrations were higher were located in paddy fields, whereas the areas with lower concentration were in dry lands and newly reclaimed dry lands. These results demonstrated that different land use types can cause differences in spatial distribution between TN and TP.

Kriging Standard Deviations of Total Nitrogen and Total Phosphorus
Kriging SDs (KSDs) were considered as the SDs of the interpolated pixel values (McGrath et al., 2004). They also reflected the internal uncertainty of the SDs of the sample data and the interpolation results by sampling at a large scale. Therefore, KSD values can be used as a criterion to estimate the accuracy of kriging interpolation. The smaller the KSD values are, the more reliable the kriging results are.

Figure 7 represents the KSD of TN and TP, whose distribution patterns were cloud shaped, indicating the degree of model error. The estimated SDs of TN and TP varied within the range of 0.32 to 0.67 g/kg and 0.1052 to 0.1415 g/kg, respectively, and their mean values were 0.3971 and 0.1119 g/kg,respectuvely. The KSD values were relatively smaller in the neighborhood of sampling sites, whereas they were higher in the areas farther away from the sampling sites. Therefore, KSD values were related to the sampling density and the variogram structure (McGrath et al., 2004).


Figure 7
View larger version (49K):
[in this window]
[in a new window]

  		Fig. 7. Kriging SD (KSD) of (A) total soil nitrogen and (B) total soil phosphorus.

 
Risk Assessment of Total Nitrogen and Total Phosphorus
To further assess the risk of TN and TP on the environment, we applied probability kriging methods to conduct the distribution probability of TN and TP at different levels, choosing 2-order trend, anisotropic structure, and data of the sampling sites that had been transformed by Box-Cox. Box-Cox transformation was also applied in the determination of threshold value. If TN and TP concentrations are over certain thresholds, they threaten environmental quality, especially water quality. Lin et al. (2002) pointed out that the value of the fifth classes was selected as the threshold for soil properties. To obtain TP probability maps, which represented natural historic wetland conditions, Grunward et al. (2004) chose a threshold of 0.45 g/kg of TP. Goovaerts (1997) considered the sixth deciles as threshold and produced the indicator map exceeding the threshold. However, the standard values of best aerobic soils in China are 2.0 g/kg of TN and 1.0 g/kg of TP (Shen, 1998). The concentrations of TN and TP in the area (3.2 g/kg and 1.05 g/kg, respectively) were the sixth deciles and the fifth classes. Thus, the thresholds of TN and TP for declaring a hazardous or contaminated location were set to 3.2 and 1.05 g/kg, respectively.

In the present study, we regarded the low probability section (<10%) of the assessment as "safe" because it could not exceed the determined threshold value, whereas the highest probability section (>70%) was regarded as "dangerous" because it likely exceeded the determined threshold value (Fig. 8 ). We calculated the mean probability of TN content >3.2 g/kg to be 0.2916. The "safe" areas, where the probability was 0.0 to 0.1, dominated the main part of its distribution. These areas covered 1.2 km2, accounting for 33.2% of the whole area. The "dangerous" areas, where the probability was 0.7 to 1.0, covered 0.4 km2, accounting for 9.8% of the total. The mean probability of TP content >1.05 g/kg was 0.3263, with the "safe" section of 0.25 to 0.40 taking up the main part, which covered an area of 1.05 km2 (28.3% of the total). The "dangerous" section of 0.7 to 1.0 covered an area of 0.4 km2, accounting for 3.5% of the total. The "dangerous" sections would become the key regions for the prevention of nonpoint-source pollution. Therefore, the risk assessment of TN and TP not only helped to determine the potential problems of nutrient management in farmland soil at regional scales, but also facilitated the identification of key regions for controlling nonpoint-source pollution, which was of great significance for agricultural production and water environment management in the Dongting Lake plain.


Figure 8
View larger version (40K):
[in this window]
[in a new window]

  		Fig. 8. Probability distribution map of (A) total soil nitrogen (TN) and (B) total soil phosphorus (TP) exceeding the threshold value (TN, 3.2 g/kg; TP, 1.05 g/kg).

 
The study area covers approximately 4 km2, providing a relatively larger sampling density (two samples per hectare) and facilitating the attainment of an accurate interpolation mapping. The analysis method of this paper can provide an alternative way of judgment for effective differentiation between key region and control regions of nonpoint pollution and for conducting the corresponding assessment. Spatial heterogeneity is closely correlated with observation scale. Therefore, the future study of the spatial variation characteristics of soil nutrients in different scales (e.g, field, watershed, and region in the subtropical regions), which assesses the similarities and differences of spatial information of different scales, will be helpful for the extrapolation of the scales of spatial information.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
There was a deflection in the raw data of TN and TP. However, normal distributions can be obtained in data sets after Box-Cox transformations, which also served to reduce the negative effects of outliers. A moving windows technique was adopted to verify the stationarity of the data. The spatial correlation distances of TN and TP were 322 and 338 m, respectively. Different anisotropic structures were observed in different directions and at different scales. The 2-order trend was applied in kriging interpolation. The probability map of kriging interpolation provided powerful information for risk assessment of soil nutrients, controlling nonpoint sources of N and phosphorus P, and the decision of management strategy.


   ACKNOWLEDGMENTS
 
We thank financial support for this study from the Chinese Academy of Sciences (KZCX-YW-423; KZCX3-SW-338), the Ministry of Science and Technology of China (2002CB412503), and the National Science Foundation of China (No. 4057117).


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
C. Wu, J. Wu, Y. Luo, L. Zhang, and S. D. DeGloria
Spatial Estimation of Soil Total Nitrogen Using Cokriging with Predicted Soil Organic Matter Content
Soil Sci. Soc. Am. J., August 19, 2009; 73(5): 1676 - 1681.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Lu, P.
Right arrow Articles by Wu, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Lu, P.
Right arrow Articles by Wu, J.
Agricola
Right arrow Articles by Lu, P.
Right arrow Articles by Wu, J.
Related Collections
Right arrow Watershed and Landscape Processes
Right arrow Nitrogen
Right arrow Phosphorus
Right arrow Ecological Risk Assessment
Right arrow Soil Fertility and Productivity

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome