No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Spatial relationship between soil physical and hydraulic properties along a transect
Abstract
The relationship between soil properties may vary with their spatial separation. Understanding this
relationship is important in predicting hydraulic parameters from other soil physical properties. The objective of this study
was to identify spatially dependent relationships between hydraulic parameters and soil physical properties. Regularly
spaced (3 m) undisturbed soil samples were collected along a 384 m transect from a farm field at Smeaton, Saskatchewan.
Saturated hydraulic conductivity, the soil water retention curve, and soil physical properties were measured. The scaling
parameter, van Genuchten scaling parameter a (VGa), and curve shape parameter, van Genuchten curve shape parameter
n (VGn), were obtained by fitting the van Genuchten model to measured soil moisture retention data. Results showed that
the semivariograms of soil properties exhibited two different spatial structures at spatial separations of 20 and 120 m,
respectively. A strong spatial structure was observed in organic carbon, saturated hydraulic conductivity (Ks), sand, and
silt; whereas a weak structure was found for VGa and VGn. Correlation circle analysis showed strong spatially dependent
relationships of Ks and VGa; with soil physical properties, but weak relationships of us and VGn with soil physical
properties. The spatially dependent relationships between soil physical and soil hydraulic parameters should be taken into
consideration when developing pedotransfer functions
... The roles of the former two have been widely acknowledged (e.g., Arya and Paris, 1981;Rawls et al., 1998;Papanicolaou et al., 2015) and they are included in almost all the PTFs for K s (Wösten et al., 2001;Arrington et al., 2013;Patil and Singh, 2016). According to the phase relationships implied in the arrows in the regions of significant correlations, ln(K s ) was spatially negatively correlated with bulk density and clay content, but positively correlated with sand content (Fig. 5), which agrees with many previous studies (e.g., Biswas and Si, 2009;Bevington et al., 2016). On the other hand, topography as indicated by elevation and slope gradient specifically, might affect K s spatial pattern through its impact on sediment redistribution and hence on soil texture Sobieraj et al., 2002;Wang et al., 2013). ...
Core Ideas
K s was estimated using MLR‐type PTFs, ANN‐type PTFs and state‐space analysis.
State‐space modeling was scale‐sensitive in estimating K s in Loess Plateau.
Spatial correlations revealed in state‐space analyses were consistent with wavelet coherency.
Bulk density, clay content and topography dominated K s spatial distribution.
A precise description of saturated hydraulic conductivity ( K s ) and its spatial variability is required for modeling soil and water transport in the vadose zone. Nevertheless, the direct measurement of K s is expensive and laborious especially for large domains crossing hundreds of kilometers. The objective was to estimate K s from easily accessible soil properties and environmental factors using pedotransfer functions (PTFs) and state‐space analysis. Along an 860‐km south–north transect in the Loess Plateau of China, soil cores for K s measurements were collected at depths of 0 to 10, 10 to 20, and 20 to 40 cm at 10‐km intervals from 15 Apr. to 15 May 2013. Multiple linear regression (MLR) and artificial neural network (ANN) were used to derive PTFs for K s estimation. Based on the eight factors of bulk density, soil organic carbon, sand content, clay content, mean annual precipitation and temperature, slope gradient and elevation, the state‐space analysis appeared to outperform the PTFs in calibrating K s over the entire transect. The adjusted coefficients of determination ( R ² adj ) for the state‐space models were all greater than 0.9, whereas the corresponding R ² adj were much lower for the MLR‐ and ANN‐type PTFs (ranging from 0.398 to 0.880). However, the state‐space approach is quite scale‐sensitive, and overfitting occurred when it was cross‐validated with a leave‐one‐out procedure. It performed almost perfectly in calibration as implied in the R ² adj of ∼1 but rather poorly in validation with R ² adj typically >0.4. The ANN method exhibited the best K s estimations at all depths. Both wavelet coherency and state‐space modeling quantified the spatial correlations of K s with the eight factors investigated and manifested consistent results, that is, bulk density, clay content, and topography were the primary properties controlling K s distribution. These findings are critical for hydrological modeling and irrigation management in the Loess Plateau of China and possibly other arid and semi‐arid regions.
Spatial variation of the correlation among variables related to water flow and solute transport are important in the characterization of the spatial variability when performing uncertainty analysis and making uncertainty-qualified solute transport predictions. However, the spatial variation of the correlation between solute transport parameters and soil properties are rarely studied. In this study, the spatial correlation among laboratory-measured transport parameters dispersivity and coefficient of distribution of a reactive and a nonreactive solute and soil properties were studied at the scale of a few meters using a dense sampling design. In an area of 84 m² and a depth of 2 m, 55 undisturbed soil samples were taken to determine the soil properties. Column experiments were performed, and the transport parameters were obtained by fitting the experimental data to the analytical solution of the advection-dispersion equation using the computer program CFITM. Stepwise multiple linear regression (MLR) was performed in order to identify the statistically significant variables. The spatial correlation of the variables and between variables were determined using the Stanford Geostatistical Modeling Software. Soil properties presented a moderate coefficient of variation, while hydraulic conductivity and transport parameters were widely dispersed. The difference between its minimum and maximum value was quite large for most of the studied variables evidencing their high variability. Both dispersivity and retardation factor were higher than the expected and this result can be related to the preferential pathways and to the non-connected micropores. None of the physical soil property was strongly correlated to the transport parameters. Coefficient of distribution was strongly correlated to the cation exchange capacity and significantly correlated to mesoporosity and microporosity. Hydraulic conductivity presented significant positive correlation to the effective porosity and macroporosity. Stepwise multiple linear regression analysis indicated that further studies should be performed aiming to include other variables relevant for lateritic soils such as pH, electrical conductivity, the content of Al and Fe, CaCO3 and soil structure and microstructure. The study of the spatial correlation among transport parameters and soil properties showed that the codispersion among the variables is not constant in space and can be important in dictate the behavior of the combined variables. Our results also showed that some variables that were identified as explanatory in the MLR were not significant in the spatial analysis of the correlation, showing the importance of this kind of analyses for a better decision about the most relevant variables and their relations. The present study was a first attempt to evaluate the spatial variation in the correlation coefficient of transport parameters of a reactive and a nonreactive solute, indicating the more relevant variables and the ones that should be included in future studies.
Soil N2O emission measurements have high-spatial and temporal variability that results in poor field scale predictive relationships, and this leads to high uncertainty in estimates of flux. Correlation between variables may be location and scale dependent. This needs to be understood to identify at what scales or locations predictive relationships may be used to reduce uncertainty in estimates. The purpose of this study was to describe the scale and location dependency of N2O emission on soil water and temperature using wavelet coherency and their change through time. Measurements of N2O flux, soil water content, and temperature were made multiple times from a 128-point transect on a hummocky, agricultural landscape in the Dark Brown soil zone of Saskatchewan, Canada. Using three sampling dates in the spring of 2004, the local wavelet, cross-wavelet, and wavelet coherency spectra were generated. In late March, the spatial pattern of wavelet coherency between soil N2O flux and temperature was scale dependent. At the end of April, there was no significant relationship between N2O and temperature. Instead, a location-dependent spatial pattern existed between flux and water-filled pore space. By late June, there were no significant scale- or location-dependent relationships. The results indicate that N2O emission models for this landscape cannot rely on a single predictive relationship, but may have to adjust between scale- and location-dependent relationships at different times of the year.
No‐tillage (NT) practices can improve soil aggregation and change the distribution and retention of soil organic matter (SOM) compared with conventional tillage (CT), but the relationships between aggregates and SOM fractions are poorly known. The effects of long‐term (13‐yr) CT and NT management on water‐stable aggregates (WSA) and aggregate‐associated SOM were investigated on a Hiwassee sandy clay loam (clayey, kaolinitic, thermic Rhodic Kanhapludult). Samples were collected at two depths in replicated CT and NT plots and separated into five aggregate size classes by wet sieving. The stability of intact WSA was measured turbidimetrically. The C and N content of total, particulate (POM), and mineral‐associated organic matter was determined for each size class. Whole‐soil organic C was 18% higher in NT (30.7 Mg C ha ⁻¹ ) than in CT (26.1 Mg C ha ⁻¹ ). In CT, macroaggregates (>250 µm) were fewer and less stable than those of NT. The POM C made up ≈36% of whole soil C regardless of tillage, but the quantity of POM was nearly 20% higher in NT than in CT. The POM comprised a higher percentage of total aggregate N in surface soils of NT than in CT and values increased with increases in aggregate size. In NT, concentrations of total and mineral‐associated C and N were higher in the 106‐ to 250‐µm WSA than in the other size classes but, in CT, the concentrations were similar among size classes. An alternative view of aggregate organization is discussed in which microaggregates are formed in association with POM at the center of macroaggregates, helping to explain relationships between SOM storage and aggregate size distributions under different management practices.
The spatial structure of soil properties has been examined on a Typic Torrifluvent soil at The University of Arizona Experiment Station at Marana. Nine hundred samples from nine transects were collected in straight lines (100 locations for each transect), at 20-, 200-, and 2,000-cm intervals. All samples were at a 50-cm depth. Variables include 0.1 and 15 bar water con-tent, available water, surface area, particle size distribution, pH, EC, bulk density, and moisture content in the field seven days after irrigation. Autocorrelation functions were evaluated for each parameter and found to be correlated over space with patterns of three basic types: typical, random, or with a large zone of influence. Generalizations were difficult, but the cal-culated zone of influence was strongly dependent on distance between samples, with larger intervals tending to give greater values. In a few cases, this could partially be explained on the basis of larger standard deviations measured on longer tran-sects. Results indicate future difficulty in assigning scale lengths by parameter or soil.
A new and relatively simple equation for the soil-water content-pressure head curve is described. The particular form of the equation enables one to derive closed-form analytical expressions for the relative hydraulic conductivity, when substituted in the predictive conductivity models of N. T. Burdine or Y. Mualem. The resulting expressions contain three independent parameters which may be obtained by fitting the proposed soil-water retention model to experimental data. Results obtained with the closed-form analytical expressions based on the Mualem theory are compared with observed hydraulic conductivity data for five soils with a wide range of hydraulic properties.
Spatial variation in gilgai soil mas studied on a transect across typical gilgai terrain of the Bland Plain of New South Wales. The soil was sampled at 4 m intervals over a distance of 1.5 km at Caragabal (33°50¿S, 147°40¿E). Soil morphology was recorded to 1 m and samples taken for laboratory determination of pH, electrical conductivity and chloride content from 0-10, 30-40 and 80-90 cm. Bulk density was determined for the 0-10 cm depth also. All soil properties, when graphed against their sampling positions, appear very erratic; but when filtered by a five-point moving average show recurrent short range variation, associated to some extent with the visible elements of the gilgai pattern. Block variance increases steadily with increasing block size to approximately 32 m for conductivity, pH at 80-90 cm, colour hue and chroma, depth to gypsum and the first principal component. Thereafter the rate of increase is less. Spectral densities were computed from correlograms for each property. Spectra for the properties above showed more or less strong peaks at 0.12 cycles, equivalent to a wavelength of 33.3 m. Gilgai evidently recurs sufficiently regularly on the transect for spectral analysis to reveal a periodicity, and the technique is judged to be sufficiently sensitive for analysing records made directly by data logger.
variability and relationships were studied using multifractal and joint second-order statistical method, this technique provides multifractal techniques. Results indicate that for all the studied vari- poor characterization of the variability that occurs in ables 0.80 H 0.90, suggesting a certain degree of statistical scale- the presence of intermittent high and low data values invariance and long-range dependency. At the observation scale, the (Kravchenko et al., 1999; Seuront et al., 1999). The variability in Ks was significantly related to sand (SA) and silt (SI) necessary conditions of stationarity and 'quasi-gaussian' distribution (R 0.40 for SA and 0.39 for SI, P 0.01; n 128), distribution also cast certain doubts on direct applica- whereas,acrossawiderrangeofscales,thevariabilityinKswasrelated tions of geostatistical and spectral techniques in the only to clay (CL) and organic C (OC). The result indicates scale analysis of non-Gaussian and extremely variable prop- dependentrelationshipsbetweenKsandsoilphysicalproperties,which erties such as the K s. The use of simple scaling tech- implies that the success of predictive models such as pedotransfer niques such as the Miller and Miller (1956) scaling the- functions (PTFs) and Ks aggregation techniques depends largely on the correspondence between observation and implementation scales. ory and monofractal analysis is generally limited to specific cases where abrupt changes (i.e., small scale and erratic variability) in the spatial pattern of Ks are negligible and where exact self-similarity is a rule rather
The gap between data analysis and site-specific recommendations has been identified as one of the key constraints on widespread adoption of precision agriculture technology. This disparity is in part due to the fact that analytical techniques available to understand crop GIS layers have lagged behind development of data gathering and storage technologies. Yield monitor, sensor and other spatially dense agronomic data is often autocorrelated, and this dependence among neighboring observations violates the assumptions of classical statistical analysis. Consequently, reliability of estimates may be compromised. Spatial regression analysis is one way to more fully exploit the information contained in spatially dense data. Spatial regression techniques can also adjust for bias and inefficiency caused by spatial autocorrelation. The objective of this paper is to compare four spatial regression methods that explicitly incorporate spatial correlation in the economic analysis of variable rate technology: (1) a regression approach adopted from the spatial econometric literature; (2) a polynomial trend regression approach; (3) a classical nearest neighbor analysis; and (4) a geostatistical approach. The data used in the analysis is from a variable rate nitrogen trial in the Crdoba Province, Argentina, 1999. The spatial regression approaches offered stronger statistical evidence of spatial heterogeneity of corn yield response to nitrogen than ordinary least squares. The spatial econometric analysis can be implemented on relatively small data sets that do not have enough observations for estimation of the semivariogram required by geostatistics. The nearest neighbor and polynomial trend analyses can be implemented with ordinary least squares routines that are available in GIS software. The main result of this study is that conclusions drawn from marginal analyses of this variable rate nitrogen trial were similar for each of the spatial regression models, although the assumptions about spatial process in each model are quite different.
Spatial distributions of soil properties at the field and watershed scale may affect yield potential, hydrologic responses, and transport of herbicides and NO3- to surface or groundwater. Our research describes field-scale distributions and spatial trends for 28 different soil parameters at two sites within a watershed in central Iowa. Two of 27 parameters measured at one site and 10 of 14 parameters measured at the second site were normally distributed. Spatial variability was investigated using semivariograms and 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. Twelve parameters at Site one, including organic C, total N, pH, and macroaggregation, and four parameters at Site two, including organic C and total N, were strongly spatially dependent. Six parameters at Site one, including biomass C and N, bulk density, and denitrification, and 9 parameters at Site two, including biomass C and N and bulk density, were moderately spatially dependent. Three parameters at Site one, including NO3 N and ergosterol, and one parameter at Site two, mineral-associated N, were weakly spatially dependent. Distributions of exchangeable Ca and Mg at Site one were not spatially dependent. Spatial distributions for some soil properties were similar for both field sites. We will be able to exploit these similarities to improve our ability to extrapolate information taken from one field to other fields within similar landscapes.
Estimation of soil hydraulic properties by pedotransfer functions (PTFs) can be used in many applications. Some of existing PTFs estimate saturated hydraulic conductivity (K(s)) of the soil, using organic matter (OM) content as one of the input variables. Several authors have shown an increase in K(s) with increasing OM content, a soil property that presumably improves soil structure. We used three popular PTFs to examine the relationship between OM content and K(s). We also used data originating from the U.S., Hungary, and the European HYPRES database, to develop additional PTFs with the Group Method of Data Handling (GMDH). It appears that existing PTFs negatively correlate K(s) with OM content for some soils. We found indications of negative relationship between OM content and K(s) with the newly developed PTFs both for directly estimated K(s), and for K(s) estimated via the effective porosity of the soil, using a generalized Kozeny-Carman approach. It is not straightforward to define the exact range of soils with the inverse relationship between OM and K(s). The range appeared to be data set dependent, but it was extensive within the valid input range of each PTF.
Soil survey activities in many countries have reached a crucial phase. Standard country wide surveys either have been completed or will be completed within the near future. This development implies that more attention can be paid to soil survey interpretations and applications. The opportunity to emphasize use of soil survey data is particularly timely because of a change in the type of questions being asked by users of soil survey data and because of the rapid advances in information technology. Questions have become more specific, and quantitative answers are increasingly required.
A soil property may be related to another and the relationships may change depending on the scale and location. Understanding these scale- and location-dependent relationships is important for prediction of one soil property based on another. The objective of this study is to use wavelet coherency analysis to examine whether the relationship between hydraulic properties and soil physical properties are scale- and location-dependent. Undisturbed cores were collected along a transect from the sandy loam soil of a farm field in northern Saskatchewan, Canada. Saturated hydraulic conductivity (K-s), sand content, and organic carbon content (OC) were measured on these cores, and their relationships as a function of scale and location were analyzed using wavelets. Results indicated that the wavelet coherency between K-s and sand content is only significantly different from that of red noises at the scales around 48 m. The cross-wavelet spectrum and wavelet coherency are predominantly in phase, suggesting a positive correlation between K-s and sand. For K-s and OC, significant coherency exists at scales from 30 to 48 and around 80 m. At the scales of 30-48 and around 80 m the relationships are predominantly out of phase, suggesting negative correlation. Therefore relationships between K-s and sand or K-s and OC are not only scale- dependent but also location-dependent. Scale and location dependence have an important implication for understanding the scaling relationships between K-s and sand and OC and for the prediction of K-s from sand and OC.
Whether processes in the natural world are dependent or independent of
the scale at which they operate is one of the major issues in hydrologic
science. In this volume, leading hydrologists present their views on
the role of scale effects in hydrologic phenomena occurring in a range
of field settings, from the land surface to deep fractured rock.
Self-contained and thought-provoking chapters cover both theoretical and
applied hydrology. They provide critical insights into important topics
such as general circulation models, floods, river networks, vadose zone
processes, groundwater transport, and fluid flow through fractured
media. This book is intended as an accessible introduction for graduate
students and researchers to some of the most significant questions and
challenges that will face hydrologic science in the twenty-first
century.
Modeling contaminant and water flow through soil requires accurate estimates of soil hydraulic properties in field scale. Although artificial neural networks (ANNs) based pedotransfer functions (PTFs) have been successfully adopted in modeling soil hydraulic properties at larger scales (national, continental, and intercontinental), the utility of ANNs in modeling saturated hydraulic conductivity (K-s) at a smaller (field) scale has rarely been reported. Hence, the objectives of this study are (i) to investigate the applicability of neural networks in estimating K-s at field scales, (ii) to compare the performance of the field-scale PTFs with the published neural networks program Rosetta, and (iii) to compare the performance of two different ensemble methods, namely Bagging and Boosting in estimating K-s. Datasets from two distinct sites are considered in the study. The performances of the models were evaluated when only sand, silt, and clay content (SSC) were used as inputs, and when SSC and bulk density rho(b) (SSC+ rho(b)) were used as inputs. For both datasets, the field scale models performed better than Rosetta. The comparison of field-scale ANN models employing bagging and boosting algorithms indicates that the neural network model employing the boosting algorithm results in better generalization by reducing both the bias and variance of the neural network models.
be used for input in preferential flow models, and to test the assumptions of the models. The objectives of Solute transport models that include a preferential flow component this paper were to evaluate methods to independently require many input parameters. There are well established procedures to determine micropore parameters, but procedures to determine measure or calculate macropore parameters, and to use macropore parameters are not well established. The objective of this this information to test the assumptions of the preferen- paper was to evaluate methods to independently measure macropore tial flow model MACRO. parameters. The test model used was MACRO, a transient-state, two- flow domain model. The key macropore parameters in the model are THEORY
Soil N2O emission data is typically highly skewed. In terrain where the spatial distribution of soil processes is controlled by topography, extreme N2O flux events can be highly localized and nonstationary. Wavelet analysis can be used to describe the spatial variation of these nonstationary processes. The objectives of this study were to use wavelet analysis to determine the spatial variation, scales of variability and their change over the snowmelt period for soil N2O flux. On a hummocky, agricultural landscape in the Dark Brown soil zone of Saskatchewan, N2O flux measurements were taken from a 128-point linear transect five times over the spring snowmelt season of 2004. Localized variance was determined using a continuous wavelet transform (Mexican Hat) and the local spectrum was compared with the distribution of fluxes along the transect and the relative elevation. Two spatial patterns of soil N2O emission were revealed. The first was a cyclic, landscape-element-controlled pattern with a scale of variation that ranged between 20 and 60 m. Changes in the spatial scale were due to a shift in importance between landscape elements as sources of peak N2O flux. The second pattern was composed of noncyclic, localized features that were due to extreme flux events at specific landscape positions. These extreme flux events were not temporally persistent, but represent a large, non-random contribution to mean and variance on the dates they occurred. Sample strategies to capture the full range of soil N2O emission at this site would probably require separate approaches for these two spatial patterns.
The effect of microtopography and an intertill stratigraphic sand-gravel layer on the spatial variability of A horizon thickness, density, and mass are examined using standard, spatial, and spectral statistical methods. Microtopography parameters were estimated by fitting a least squares quadratic surface to elevation measurements around each soil core. Standard correlation indicated a significant relationship between surface curvature in the direction of maximum surface gradient and A horizon thickness and mass; however, only 14% of the variability was explained. -from Authors
2 0.46), whereas soil surface curvature explained only 15% of the variations in grain yield. Wavelet analyses revealed that between slope percentage and wheat yield. significant variations occurred at scales 160 m for wheat grain yield, In semi-arid or arid regions, soil water is the major 280 m for upslope lengths, and 110 m for the wetness index. The cross limiting factor for crop production and processes that wavelet analysis indicated that significant covariance existed at scales control the soil water distribution also control crop pro- 180 m between wheat grain yield and upslope lengths and at scales duction. Water accumulation and runoff processes are 140 m between wheat yield and the wetness index, at a 99% confi- largely determined by landscape configuration (convex dence level. vs. concave). Sinai et al. (1981) showed that in arid regions where there was significant unsaturated lateral soil water movement, soil water content was highly cor-
Hydropedology is suggested as an intertwined branch of soil science and
hydrology that embraces the interdisciplinary, integrated, and
multiscale approaches to the study of interactive pedologic and
hydrologic processes and properties in the Earth's critical zone.
Emphasized here are potential "bridges" to address (1) knowledge gaps
between traditional pedology and soil physics/hydrology, (2) scale
differences in microscopic, mesoscopic, and macroscopic studies of
soil-water relationships, and (3) data translation from soil survey
databases into soil hydraulic properties. Such bridges signify the
potential unique contributions hydropedology can make to integrative
soil and water sciences. The needs for such bridges are illustrated
using landscape-oriented flow and transport, multiscale modeling, and
pedotransfer functions. The establishment of hydropedology would enhance
holistic understanding of
molecular-aggregate-pedon-catena-watershed-regional-global scale
connections. It is hoped that hydropedology would contribute to enhanced
understanding of a variety of environmental, ecological, agricultural,
and natural resource issues of societal importance, such as water and
soil quality, watershed processes, contaminant fate, waste disposal,
precision agriculture, and ecosystem functions. Recognizing
hydropedology would also enhance the education of the next generation of
soil scientists and vadose zone hydrologists as education in the 21st
century emphasizes interdisciplinarity.
The provenance of trace metals in soil, whether from the parent material or from pollution, is rarely known with certainty, and the metals' history must usually be pieced together from fragmentary statistical information. This is particularly true in the Swiss Jura where the concentrations of several heavy metals around La Chaux de Fonds exceed the statutory recommended thresholds for safety.
The topsoil of the 14.5-km2 region was sampled on a square grid at 250-m intervals with additional nesting with distances of 100 m, 40 m, 16 m and 6 m. The concentrations of seven potentially toxic metals, namely Cd, Co, Cr, Cu, Ni, Pb and Zn, were measured. Their coregionalization could be represented by a linear model consisting of a nugget component plus two spherical structures with ranges of 0.2 km and 1.3 km. The short-range component dominated the variograms of Cd, Cr, Cu and Pb; the long-range component dominated those of Co and Ni; the variogram of Zn combined the two in approximately equal proportions. The coregionalization matrices contain moderate correlation among the nugget and the short-range components, notably between Cu and Pb, between Cd and Zn, and between Cr, Ni and Zn. The strongest correlations are at the long range between Co, Cr and Ni, and to a somewhat smaller degree between Zn and Co. Analysis of variance showed Co and Ni to be related to geology, and to the Argovian formation in particular. The indicator variogram of this formation has also a short-range component. The analysis also showed Cr and Cu to be related to land use (in different ways). Copper and Pb are strongly correlated to one another and distinct from the five other metals. The long-range structure is almost certainly a geological effect, whereas the one of short range probably results from both the geology and human activities.
Some possible approaches to the aggregation and disaggregation of soil data and information are presented as an opener to the more detailed discussion. The concepts of hierarchy, grain, extent, scale and variability are discussed. Slight modifications to the Hoosbeek-Bryant scheme to deal with spatial and temporal scales and various types of quantitative models are suggested. Approaches to aggregation or upscaling are reviewed. The contributions of representative elementary volume (REV), variograms, fractal theory, multi-resolution analysis using wavelets, critical point phenomena, renormalisation groups and transfer functions are discussed followed by a brief presentation of some ecological approaches including extrapolation by lumping, extrapolation by increasing model extent and extrapolation by explicit integration. A clear distinction must be made between additive and non-additive variables. The scaling of the former is much less problematic than the latter. Corroboration of any approach by testing against the aggregated values seems problematic. Methods of disaggregation or downscaling including transfer functions, mass-preserving or pycnophylactic methods are also discussed. In order to make quantitative advances, nested sampling or reanalysis of data in land information systems to obtain variance information over a complete range of scales is required. Finally an appeal is made for work to begin on a quantitative scale-explicit theory of soil variation.
The method of weighted least squares is shown to be an appropriate way of fitting variogram models. The weighting scheme automatically gives most weight to early lags and down-weights those lags with a small number of pairs. Although weights are derived assuming the data are Gaussian (normal), they are shown to be still appropriate in the setting where data are a (smooth) transform of the Gaussian case. The method of (iterated) generalized least squares, which takes into account correlation between variogram estimators at different lags, offer more statistical efficiency at the price of more complexity. Weighted least squares for the robust estimator, based on square root differences, is less of a compromise.
The applicability and usefulness of Geostatistics (kriging) as a tool for optimum selection of sites for monitoring groundwater
levels has been demonstrated through a case study. The criterion used is the estimation of error variance. Groundwater level
data (pre-monsoon 1994) obtained from 32 observation wells of Upper Kongal basin, Nalgonda District, A.P. (India) has been
stochastically analyzed. The spatial distribution of water levels and its associated error variance is computed and the locations
having maximum error variance are selected as additional sites for augmenting the existing observational well network.
Measurement of soil hydraulic properties under field conditions is expensive and time consuming. Indirect methods that include estimating soil water properties from readily available soil physical data (known generally as pedotransfer functions) are preferred options. However, since environmental factors and processes operate at multiple spatial scales, the accuracy and reliability of these indirect techniques is often below expectations. The objective of this study was to determine if there is a scale-dependent relationship between soil water storage at three commonly used matric potentials (0, − 30, and − 1500 kPa) and basic soil physical properties (bulk density, sand, silt, clay, and organic carbon content). Soil physical property and water storage measurements were taken from a 384-m transect on a sandy loam soil, located at Smeaton, SK, Canada. The multiple scale relationships between these variables were studied using multifractal and joint multifractal techniques. Results indicate that at the observation scale, the spatial patterns of water storage at 0 and − 30 kPa matric potentials were significantly (r2 = 0.2 to 0.42, P < 0.01) related to clay and organic carbon content. Multifractal analysis, however, showed that the spatial pattern and variability of water storage at 0 and − 30 kPa matric potentials were related largely to sand and silt contents. Residual water (− 1500 kPa) was related only to clay content at all spatial scales studied. This study suggests that single scale analysis may not be sufficient to fully characterize spatial variability in soil and soil water properties.
Pedotransfer functions (PTFs) for estimating water-retention from particle-size and bulk density are presented for Australian soil. The water-retention data sets contain 733 samples for prediction and 109 samples for validation. We present both parametric and point estimation PTFs using different approaches: multiple linear regression (MLR), extended nonlinear regression (ENR) and artificial neural network (ANN). ENR was found to be the most adequate for parametric PTFs. Multiple linear regression cannot be used to predict van Genuchten parameters as no linear relationship was found between soil properties and the curve shape parameters. Using the prediction set, ANN performance was similar to the ENR performance for the prediction dataset, but ENR performed better on the validation set. Since ANN is still considered as a black-box approach, the ENR approach which has a more physical basis is preferred. Point estimation PTFs were estimated for water contents at −10, −33 and −1500 kPa. Multiple linear regression performed better for point estimation. An exponential increase trend was found between particles <2 μm and water contents held at −10, −33 and −1500 kPa. The point estimation ANN did not improve prediction compared to MLR. Increasing the number of functions and parameters in developing PTFs does not necessary improve the prediction. The effect of parameter uncertainty, differences in texture determination and spatial variability on the error in prediction is also discussed.
Many environmental studies on the protection of European soil and water resources make use of soil water simulation models. A major obstacle to the wider application of these models is the lack of easily accessible and representative soil hydraulic properties. In order to overcome this apparent lack of data, a project was initiated to bring together the available hydraulic data which resided within different institutions in Europe into one central database. This information was then used to derive a set of pedotransfer functions applicable to studies at a European scale. These pedotransfer functions predict the hydraulic properties from parameters collected during soil surveys and can be a good alternative for costly and time-consuming direct measurement of these properties. A total of 20 institutions from 12 European countries collaborated in establishing the database of draulic operties of uropean oils (HYPRES). This database has a flexible relational structure capable of holding a wide diversity of both soil pedological and hydraulic data. As these data were contributed by 20 different institutions it was necessary to standardise both the particle-size and the hydraulic data. A novel similarity interpolation procedure was successfully used to achieve standardization of particle-sizes according to the FAO clay, silt and sand particle-size ranges. Standardization of hydraulic data was achieved by fitting the Mualem-van Genuchten model parameters to the individual θ(h) and K(h) hydraulic properties stored in HYPRES. The HYPRES database contains information on a total of 5521 soil horizons (including replicates). Of these, 4030 horizons had sufficient data to be used in the derivation of pedotransfer functions. Information on both water retention and hydraulic conductivity was available for 1136 horizons whereas 2894 horizons had only information on water retention. Each soil horizon was allocated to one of 11 possible soil textural/pedological classes derived from the six FAO texture classes (five mineral and one organic) and the two pedological classes (topsoil and subsoil) recognised within the 1:1 000 000 scale Soil Geographical Data Base of Europe. Next, both class and continuous pedotransfer functions were developed. By using the class pedotransfer functions in combination with the 1:1 000 000 scale Soil Map of Europe, the spatial distribution of soil water availability within Europe was derived.
Saturated hydraulic conductivity is one of the key parameters for water transport models in the unsaturated zone. Several attempts have been made to estimate this parameter from available soil data as e.g. particle size distribution, bulk density, and organic matter content. All of these estimation methods (pedo-transfer functions) exhibit large differences between the predicted and the measured saturated hydraulic conductivities, because they calculate the saturated hydraulic conductivity as a deterministic physical parameter. In this study the saturated hydraulic conductivity is defined as a lognormally distributed random variable, the (geometric) mean and standard deviation of which depend on the texture. The confidence limits of the geometric mean and standard deviation of the measured saturated hydraulic conductivity within the FAO textural classification scheme are presented. Due to the lognormal distribution the ratio of the predicted and the measured saturated hydraulic conductivity (error ratio) is statistically analyzed within the FAO textural classes using the database of the Lower Saxony Soil Information System. For some pedo-transfer functions the geometric mean error ratio is near one, but the geometric standard deviation of the error ratio is generally large and within the investigated textural classes has nearly the same value as the geometric standard deviation of the measured saturated hydraulic conductivity. This indicates that the pedo-transfer function approach is useful to predict saturated hydraulic conductivity, if mean and standard deviation are considered.
Soil variability within fields results from complex geological and pedological processes, therefore soil variables are expected to be correlated in a scale dependent way. An accurate soil characterization, taking into account soil and spatial variability, will allow the farmer to follow crop management practices fitted with the real soil situation. The study of the scale-dependent correlation structure of some variables was investigated by means of Factorial Kriging Analysis (FKA) developed by Matheron. This analysis consists: modeling of the coregionalization, analysing the correlation structure between the variables, estimating and mapping the regionalized factors. The present study was conducted in a 4-ha field located in central Italy. Soil samples were collected at 0–10 and 10–30 cm depth and were then analysed for pH, (cation exchange capacity) CEC, total N, Olsen P, exchangeable K, and Na. The analytical results were submitted to different kinds of analysis: classical statistical and geostatistical analysis which showed that plant macro- and micro-nutrients coming from fertilization had led to short-range variation, especially in Na, N, and CEC. The effects of these were superimposed on long-range processes, producing systematic patterns in soil fertility.
Understanding the correlation between soil hydraulic parameters and soil physical properties is a prerequisite for the prediction of soil hydraulic properties from soil physical properties. The objective of this study was to examine the scale- and location-dependent correlation between two water retention parameters (alpha and n) in the van Genuchten (1980) function and soil physical properties (sand content, bulk density [Bd], and organic carbon content) using wavelet techniques. Soil samples were collected from a transect from Fuxin, China. Soil water retention curves were measured, and the van Genuchten parameters were obtained through curve fitting. Wavelet coherency analysis was used to elucidate the location- and scale-dependent relationships between these parameters and soil physical properties. Results showed that the wavelet coherence between alpha and sand content was significantly different from red noise at small scales (8-20 m) and from a distance of 30 to 470 m. Their wavelet phase spectrum was predominantly out of phase, indicating negative correlation between these two variables. The strong negative correlation between alpha and Bd existed mainly at medium scales (30-80 m). However, parameter n had a strong positive correlation only with Bd at scales between 20 and 80 m. Neither of the two retention parameters had significant wavelet coherency with organic carbon content. These results suggested that location-dependent scale analyses are necessary to improve the performance for soil water retention characteristic predictions.
Methods and theory for unconfined infiltration measurements. Pages 123Á141 in G. C. Topp et al. eds. Advances in measurements of soil physical properties: Bringing theory into practice
- D W Anderson
- J G Ellis
- S K Saskatoon
- M D Ankeny
Anderson, D. W. and Ellis, J. G. 1976. The soils of the
provincial forest reserves in Prince Albert map area. Extension
Publication 261, Saskatoon, SK.
Ankeny, M. D. 1992. Methods and theory for unconfined
infiltration measurements. Pages 123Á141 in G. C. Topp et al.
eds. Advances in measurements of soil physical properties:
Bringing theory into practice. SSSA Spec. Publ. 30. SSSA,
Madison, WI.
Physical scales and spatial predictability of transport processes in the environment
- P Baveye
- C W Boast
Baveye, P. and Boast, C. W. 1999. Physical scales and spatial
predictability of transport processes in the environment. Pages
261Á280 in D. L. Corwin, ed. Assessment of non-point source
pollution in the vadose zone. Geophysical Monogr. 108.
The Canadian system of soil classification
- A Castrignanò
- L Giugliarini
- R Risaliti
- N Martinelli
Canadian Society of Soil Science. 1998. The Canadian system
of soil classification. NRC Research Press, Ottawa, ON. Castrignanò, A., Giugliarini, L., Risaliti, R. and Martinelli, N.
Spatial scaling analysis of soil physical properties: a review of spectral and wavelet methods
- R H Shumway
- D S Stoffer
Shumway, R. H. and Stoffer, D. S. 2000. Time series analysis
and its applications. Springer, New York, NY.
Si, B. C. 2008. Spatial scaling analysis of soil physical
properties: a review of spectral and wavelet methods. Vadose
Zone J. 7: 547Á562.
Geostatistical theory and application to variability of some agronomical properties
- S R Viera
- J L Hatfield
- D R Nielsen
- J W Biggar
Viera, S. R., Hatfield, J. L., Nielsen, D. R. and Biggar, J. W.
1982. Geostatistical theory and application to variability of
some agronomical properties. Hilgardia 51: 1Á75.
Burrough, P. A. 1983. Problems of superimposed effects in the statistical study of the spatial variation of soil
- D R Brillinger
Brillinger, D. R. 2001. Time series: Data analysis and theory.
Soc. Ind. Appl. Math. Philadelphia, PA. p. 540.
Burrough, P. A. 1983. Problems of superimposed effects in the
statistical study of the spatial variation of soil. Agric. Water
Manag. 6: 123Á143.
25: 281Á302. Goovaerts, P. 1994. Study of spatial relationship between two sets of variables using multivariate geostatistics
- G W Gee
- J W Bauder
- Asa
- W I Madison
- P Goovaerts
Gee, G. W. and Bauder, J. W. 1986. Particle size analyses. In A.
Klute, ed. Method of soil analyses. Part 1: Physical and
mineralogical methods. ASA, Madison, WI.
Goovaerts, P. 1993. Spatial orthogonality of the principal
components computed from coregionalized variables. Math.
Geol. 25: 281Á302.
Goovaerts, P. 1994. Study of spatial relationship between two
sets of variables using multivariate geostatistics. Geoderma 62:
93Á107.
Goovaerts, P. 1997. Geostatistics for natural resources evaluation. Oxford University Press, New York, NY.
Goovaerts, P. 1998. Geostatistical tools for characterizing the
spatial variability of microbiological and physio-chemical soil
properties. Biol. Fertil. Soils 27: 315Á334.
Goovaerts, P. and Chiang, C. N. 1993. Temporal persistence of
spatial patterns for mineralizable nitrogen and selected soil
properties. Soil Sci. Soc. Am. J. 57: 372Á381.
Goulard, M. and Voltz, M. 1992. Linear coregionalization
model: Tools for estimation and choice of cross variogram
matrix. Math. Geol. 24: 269Á286.
Method of soil analyses. Part 1: Physical and mineralogical methods
- A Klute
- C Dirksen
- Asa
- W I Madison
- L H Koopmans
Klute, A. and Dirksen, C. 1986. Hydraulic conductivity and
diffusivity: laboratory methods. Pages 687Á734 in A. Klute, ed.
Method of soil analyses. Part 1: Physical and mineralogical
methods. ASA, Madison, WI.
Koopmans, L. H. 1974. The spectral analysis of time series.
Academic Press, New York, NY.
Kravchenko, A. N., Boast, C. W. and Bullock, D. G. 1999.
Multifractal analyses of soil properties. Agron. J. 91: 1033Á
1041.
Prediction of soil water retention using soil physical data and terrain attributes
- W D Reynolds
- D E Elrick
- E G Youngs
- H W G Booltink
- J Bouma
- Sssa
- W I Madison
- N Romano
- M Palladino
Reynolds W. D., Elrick, D. E., Youngs, E. G., Booltink, H. W.
G. and Bouma, J. 2002. Saturated and field saturated water
flow parameters. In J. H. Dane and G.C. Topp, eds. Methods
of soil analysis. Part 4. Physical methods. SSSA, Madison, WI.
Romano, N. and Palladino, M. 2002. Prediction of soil water
retention using soil physical data and terrain attributes.
J. Hydrol. 265: 56Á75.