HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by Vanderlinden, K. Articles by Van Meirvenne, M. Search for Related Content PubMed Articles by Vanderlinden, K. Articles by Van Meirvenne, M. GeoRef GeoRef Citation Agricola Articles by Vanderlinden, K. Articles by Van Meirvenne, M. Related Collections Pedotransfer Functions Spatial Variability Surface Hydrology

SPECIAL SECTION: ZNS'03 VADOSE ZONE RESEARCH

Soil Water-Holding Capacity Assessment in Terms of the Average Annual Water Balance in Southern Spain

^a Organic Farming and Natural Resources, CIFA Las Torres-Tomejil, IFAPA, Ctra. Sevilla-Cazalla, km 12,2, 41200 Alcalá del Río, Sevilla, Spain
^b Dep. of Agronomy, Univ. of Córdoba, P.O. Box 3048, 14080 Córdoba, Spain
^c Dep. of Soil Management and Soil Care, Ghent Univ., Coupure 653, B-9000, Gent, Belgium
^* Corresponding author (karl.vanderlinden.ext{at}juntadeandalucia.es)

Received 23 June 2004.

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Knowledge of the soil water-holding capacity, w₀, is essential to the evaluation of regional soil water balance. In this paper, we produced a map of w₀ for the region of Andalusia, southern Spain, using pedotransfer functions (PTFs) and geostatistics. The information available consisted of analytical and morphological data from 521 soil profiles in the region, and the 1:400000 soil map of Andalusia. The w₀ values were calculated using 10 published PTFs. The soil map only slightly improved the spatial interpolation of the PTF-calculated w₀'s. The PTF estimates for w₀ ranged from near 0 to 235 mm, with an average value of 110 mm and a SD of 48 mm. Since no independent field observations were available, the w₀ estimates were evaluated in terms of the average annual total runoff and actual evapotranspiration. Both components were calculated at 160 meteorological observatories using a simple bucket water balance model, driven by daily meteorological data. The spatial variability of w₀ had little effect on the calculated average annual water balance of the region. Increasing w₀ to 150 to 200 mm produced a better fit of the water balance predicted with Budyko's empirical functions. The difference could be partly explained by seasonality-related characteristics of the climate in the region. Comparison of the results with other studies suggests that the estimated w₀ values should be increased by 45%. These differences can be attributed entirely to an inconsistent definition of field capacity (FC).

Abbreviations: CRV, complement of the relative variance • FAO, Food and Agriculture Organization of the United Nations • FC, field capacity • ME, mean error • OM, organic matter • PTF, pedotransfer function • RMSE, root mean square error • SKlm, simple kriging with local varying means • w₀, soil water-holding capacity • WP, wilting point

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
REGIONAL WATER BALANCE assessment requires information on the hydraulic properties of representative soils in that region. When using simple water balance models, this information is usually expressed by the w₀, defined as the difference between water content at the FC, _–33kPa, and the wilting point (WP), _–1500kPa.

Because of the considerable spatial variability in soil hydraulic properties, observations usually exhibit high levels of uncertainty (Kutílek and Nielsen, 1994). In addition, discrepancies are known to occur between laboratory and field measurements (Ratliff et al., 1983). Several authors have resorted to the use of PTFs (Bouma, 1989) in attempts to alleviate these problems and to estimate soil hydraulic properties from easily obtainable soil information such as particle size distribution, bulk density, organic matter (OM) content, or carbon content. Many authors, such as Schaap and Leij (1998) and Wösten et al. (2001), have discussed the accuracy and reliability of PTFs. Schaap and Leij (1998) found that the performance of PTFs may strongly depend on the calibration and evaluation data sets. When developing PTFs from three independent databases, they found systematic differences in the predictions of the four parameters of the water retention equation of van Genuchten (1980). Depending on soil textural class, the differences were masked by the spread of the confidence intervals of the predicted curves. Differences in residual water content, _r, were observed at the dry end of the curves, especially for loam. When merging the three databases, the uncertainty intervals were substantially reduced, thus indicating increased reliability and generality of the obtained PTFs. Schaap et al. (1998)( 2001) developed neural network-based hierarchical PTFs for estimating, among other parameters, the van Genuchten (1980) water retention parameters. The PTFs were implemented in a computer program called Rosetta (Schaap et al., 2001), which is freely distributed through the world wide web (www.ussl.ars.usda.gov/models/rosetta/rosetta.htm, verified 27 Jan. 2005). The program offers the possibility to use five different PTFs, according to available input. Minasny et al. (1999) and Cornelis et al. (2001) evaluated a large number of PTFs for estimating the water retention curve, and proposed a classification into three groups, depending on the invoked methodology.

The use of PTFs permits the estimation of w₀ at locations with basic soil data. Since a regional analysis requires w₀ values for the entire region, one can produce quantified soil maps for this purpose, as was done by De Jong and Shields (1988), Kern (1995), Batjes (1996), and Wösten et al. (1999). When a good physical relationship exists between the qualitative property represented in the map units and w₀ (Heuvelink and Webster, 2001), the within-class residual variability will be small, the quality of the classification will be high, and the final w₀ map will be accurate. Unfortunately, common soil map classifications are usually of little predictive value since the classification criteria either do not coincide with the soil property in question, or are not related to it (Leenhardt et al., 1994).

One alternative is geostatistical interpolation (Goovaerts, 1997) that, based on the spatial correlation structure of the variable, finds an optimum (unbiased and with minimum variance) estimate of w₀, on the nodes of a regular grid. All additional information that may be available for the studied variable, such as soil associations or textural class maps, should be included in the interpolation procedure (Heuvelink and Webster, 2001). For example, Heuvelink and Bierkens (1992), Van Meirvenne et al. (1994), and Leenhardt et al. (1994) used soil maps to improve the interpolation of point observations of quantitative soil properties. When using a PTF for the computation of hydraulic properties of the soil, there are two possible ways to proceed (Heuvelink and Pebesma, 1999): (i) interpolation of the raw properties and subsequent computation of the desired properties with the PTF, or (ii) computation of the desired property at the point level and posterior interpolation. Both procedures have been used in the literature (e.g., Voltz and Goulard, 1994; Sinowski et al., 1997). However, the observed differences between both interpolation procedures generally are very small in comparison with the large uncertainty associated with PTF estimates.

Since the final aim of the produced w₀ map is its use in regional water balance calculations, and no independent w₀ observations are available, we evaluated w₀ in terms of a regional average water balance. We used a simple water balance model (Milly, 1993, 1994a, 1994b) that previously proved to be reliable for average annual water balance calculations for the eastern USA.

The aim of this paper was to produce a w₀ map for the region of Andalusia in southern Spain, using scarce and heterogeneous soil information. The resulting map should be accurate enough to be suitable for regional water balance calculations. Given the lack of reliable soil information in the region, except for a few basic soil properties (Trueba et al., 1999), we resorted to published PTFs developed for similar soils as those found in the study area or calibrated using exhaustive soil databases to calculate w₀ values. Different PTFs were considered for calculations of the water content at FC and WP (Salter and Williams, 1969; Rawls and Brakensiek, 1989; Batjes, 1996; Gonçalves et al., 1997; Wösten et al., 1999; Schaap et al., 1998, 2001).

Using the soil map of the region (Consejo Superior de Investigaciones Científicas–Instituto Andaluz de la Reforma Agraria, 1989) as a source of secondary information, the obtained w₀ point values were then interpolated using geostatistial techniques. Unfortunately, the soil map of Andalusia (Consejo Superior de Investigaciones Científicas–Instituto Andaluz de la Reforma Agraria, 1989) provides no information regarding the USDA textural classification. To evaluate the fraction of the variance of w₀ that can be explained with this map, it is necessary to classify the 319 soil profiles according to the 64 cartographic units. For each unit, the predominant and associated soil types (22 classes) and the predominant lithology (16 classes) were available. Since no independent observations of w₀ were available, the resulting map was evaluated using the regional average annual water balance. Its components, total runoff (Q) and actual evapotranspiration (E), were calculated at locations where the required meteorological information was available. Estimated w₀ values for the region were compared with w₀ calculated with empirical relationships (Budyko, 1974) and from a national hydrological reference study (Ministerio de Medio Ambiente, 1998; Monreal et al., 1999).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Soil Data and Soil Map of Andalusía
The starting point of the analysis was the database of Spanish soils (Trueba et al., 1999) and the soil map of Andalusia (Consejo Superior de Investigaciones Científicas–Instituto Andaluz de la Reforma Agraria, 1989). The soil database contained morphological and analytical soil data on approximately 2500 Spanish soil profiles, classified according to Food and Agriculture Organization of the United Nations–United Nations Educational, Scientific and Cultural Organization (1974) and the Soil Survey Staff (1999). For this study, complete, reliable information of particle size distribution and OM content from 1655 horizons was obtained from the database. These horizons belonged to 521 profiles, 319 (1015 horizons) of which were located within the region and 202 (640 horizons) in the neighboring provinces. Figure 1 shows the location of the soil profile data available in the region and the neighboring provinces. Note the lack of data within several large areas.

View larger version (37K): [in this window] [in a new window]   Fig. 1. Locations of the 521 soil profiles and the 160 meteorological observatories.

Selection of Pedotransfer Functions
We only considered published PTFs that were developed and calibrated with large databases containing soil data from a wide range of soil types. One exception was made by considering also the Gonçalves et al. (1997) PTF, developed for Portuguese soils, which are similar to soils found in our study region. Table 2 shows the principal characteristics of the selected PTFs.

Wösten et al. (1999) developed a series of PTFs (WO) from the Hydraulic Properties of European Soils (HYPRES) database. From the 5521 soil samples that constitute this database, only 54 came from Spain and 104 from Portugal. The data were stratified into 11 classes according to their texture class and pedology (five topsoil, five subsoil, and one organic class) and for each class the van Genuchten (1980) water retention curve parameters were reported.

Gonçalves et al. (1997) developed PTFs (GO) for Portuguese soils with data from 230 horizons coming from 80 different soil profiles. The van Genuchten (1980) water retention parameters were fitted and their average values calculated, considering (i) all data, (ii) three FAO textural classes, and (iii) the Portuguese textural classification. Here we used the PTF developed for all data.

Batjes (1996) used the World Inventory of Soil Emission Potentials (WISE) database to develop a series of PTFs that enabled the estimation of the volumetric water content range () for each soil unit of the Soil Map of the World (Food and Agriculture Organization of the United Nations, 1991). The three PTFs used here (BA1, BA2, and BA3) are summarized in Table 2.

Rawls and Brakensiek (1989) and Rawls et al. (1991) provided PTFs for estimating _–33kPa and _–1500kPa from percentage sand, clay, and OM (RA1), and from USDA textural class (RA2), using 1323 soils from the USA. The RA1 estimates were only valid for sand contents between 5 and 70%, and clay contents between 5 and 60%.

Finally, we also considered the values of _–33kPa and _–1500kPa for each USDA textural class given by Salter and Williams (1969) (SW).

Estimation of w₀
For each soil horizon, h, of a soil profile with up to H horizons, the volumetric water content range, _h, from WP (_–1500kPa) to FC (_–33kPa), was calculated using each of the 10 PTFs shown in Table 2. The total w₀ of the profiles, assuming a maximum soil depth of 1 m, was then computed using

[1]

where CF_h is the volumetric coarse fraction (>2 mm) content of horizon h, and D_h is the thickness of horizon h. The maximum depth considered for any profile was 1 m. Finally, for each profile, 10 w₀ values were obtained using the PTFs from Table 2 and Eq. [1]. After comparing the results obtained with the 10 PTFs, the RO1 PTF was selected for further use (see section Exploratory Data Analysis and Comparison of PTF Performance, below).

Methodology for the w₀ Regionalization
We considered w₀ as a random function Y(x) of the form

[2]

where m(x) is the deterministic component, '(x) the spatially correlated stochastic term, and '' a Gaussian distributed white noise term. Simple kriging with local varying means (SKlm) (Goovaerts, 1997) offers the possibility to model the first two terms of Eq. [2] using the following expression:

[3]

where m(x_o) and m(x_i) are the local means at the estimation point, x_o, and in neighboring points, x_i, respectively; z(x_i) are the values at the neighboring points, and ^SKlm_i are weighing factors found by resolving the simple kriging equations system using the variogram of the residual random function,

[4]

called the residual variogram. Equation [3] takes into account secondary information and deals with the nonstationarity of the data by using the quantified soil map values as m(x). The final map was obtained by estimating average w₀ values for blocks of 1 by 1 km using block SKlm. Simple kriging with local varying means was performed using the KT3D program from GSLIB, the Geostatistical Software Library of Deutsch and Journel (1998). The reader is referred to Goovaerts (1997) and Webster and Oliver (2001) for a thorough discussion on this topic.

Another possible way to regionalize w₀ is by classification of the soil map of Andalusia (Consejo Superior de Investigaciones Científicas–Instituto Andaluz de la Reforma Agraria, 1989). If we consider the total sample variance within the region as the sum of the pooled within-class sample variance and the between-class sample variance , then the efficiency of the classification can be evaluated in terms of the complement of the relative variance, CRV (Leenhardt et al., 1994):

[5]

Each profile was classified according to the properties and the classification of the surface soil horizon as given in the database of Trueba et al. (1999).

Average Annual Soil Water Balance and the Milly Model
No validation data were available to test the performance of the PTFs and the accuracy of the w₀ estimates. Therefore, we evaluated the final w₀ map by modeling the average annual soil water balance at 160 locations where meteorological data were available (see section Climatological Data, above), and compared the obtained results with published data.

The average annual soil water balance quantifies the partitioning of the received average annual rainfall (P, l) to total runoff (sum of surface and groundwater flow) (Q, l) and actual evapotranspiration (E, l) throughout the year:

[6]

Bailey (1979), Eagleson (1981), and Brutsaert (1982) gave overviews of empirical relationships between E and P or between Q and P. Equations proposed by Budyko (1974), Lettau (1969), and Lettau and Baradas (1973) are especially useful for understanding the partitioning of Eq. [6]. Usually an index of dryness (R) is used, which is defined as

[7]

where R_n (m l² t^–2) is the average total annual net radiation, (m l t^–2) is the latent heat of vaporization of water, and ET₀ (l) is the average total annual reference crop evapotranspiration.

Budyko (1974) used the following equations to fit data from a large number of watersheds around the world:

[8]

[9]

and observed that a large part of the data was situated between these two curves, which caused him to propose the geometric mean of both curves:

[10]

Budyko wrote these equations in terms of the annual net radiation, expressed in equivalent evaporation depth. In the work by Milly (1993)(1994a, 1994b), this value was approximated by ET₀. These equations represent the Budyko diagram, which shows the relationship between E/P and R, that characterizes the average annual water balance. The E/P is a measure for the average annual water balance, given that it quantifies the partitioning of rainfall in evapotranspiration and runoff, while R is a climatic index (Bailey, 1979, Fig. 3.1).

Milly (1993)(1994a, 1994b) analyzed the annual soil water balance using a simple model with limited water storage and an infinite infiltration capacity. The soil volume considered was bounded above by the soil surface and had a depth of 1 m, which is an approximation of the average plant root depth. Milly assumed that the horizontal dimensions of the control volume were large as compared with the horizontal water flow in the root zone, because of soil heterogeneity and local topography (approximately 100 m). The water balance of this control volume can be expressed as

[11]

where i (l t^–1) is the infiltration rate, e (l t^–1) is the actual evapotranspiration rate, and q (l t^–1) is the total discharge rate. A complete description of the invoked assumptions can be found in Milly (1994b). Using those assumptions allows Eq. [11] to be written as

[12]

where et₀ (l t^–1) is the reference crop evapotranspiration rate. Moreover, e and q are simply obtained from

[13]

and

[14]

We used this model in our study assuming a daily time-step and driven with complete meteorological data from 160 meteorological observatories within the region. From the obtained daily series of e and q, the average annual values, E and Q, were then computed and compared with Eq. [8–10] and other published data (Ministerio de Medio Ambiente, 1998).

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Exploratory Data Analysis and Comparison of PTF Performance
The descriptive statistics for the three main size fractions (clay, silt, and sand), the coarser fraction with diameters above 2 mm, and the OM content calculated for the 1015 horizons situated within the region are shown in Table 3. All properties, except for the sand content, showed a positively skewed distribution. The large skewness of the OM content was possibly due to the greater number of data values < 1% and the presence of higher values in lowland marshy areas and forests. Note the wide range of particle-size distributions in Table 3, with representation of all the USDA textural classes.

Regionalization of w₀
The efficiency of the different stratification alternatives for w₀ was evaluated in terms of the CRV (Eq. [5]), according to the textural and taxonomic classification of the 512 soil profiles. The three- and five-class FAO textural classifications and the 12-class USDA classification produced CRV values of 0.24, 0.31, and 0.52, respectively. Stratifying w₀ values using the taxonomic classifications generally produced worse results. For the principal FAO soil units, the FAO soil subunits, and the USDA soil taxonomy, we obtained CRV values of 0.24, 0.27, and 0.07, respectively. The classification efficiency according to the eight province boundaries of Andalusia produced a CRV of 0.06. This value could be interpreted as a reference for a quasirandom stratification, which should yield a CRV near zero since the division in provinces has nothing to do with soil properties. The textural classification (especially according to the USDA) produced the best results, while taxonomic classification (especially USDA soil taxonomy) performed worst. The strong reliance of the RO1 PTF on the USDA soil texture classification could explain its better performance. The USDA taxonomic classification produced similar results as those of the classification according to the different provinces, indicating its inappropriateness for classifying w₀.

The predictive capacity of the soil map of Andalusia (Consejo Superior de Investigaciones Científicas–Instituto Andaluz de la Reforma Agraria, 1989) for w₀ was found to be very low. Using w₀ estimates from 319 soil profiles, the CRV values for the cartographic unit and FAO classification were so small (0.09 and 0.08, respectively) that they were of little value for estimating w₀. Only the lithology seemed to be of some value by producing a CRV of 0.16. Nevertheless, these classifications are often used to obtain spatial estimates of the soil hydraulic properties since they constitute the only available source of information. A more detailed soil map on a larger scale is required to make a proper comparison with point data (Leenhardt et al., 1994). The lithology-based map of w₀ is shown in Fig. 2 . First average w₀ values for each lithological class were calculated, which were then assigned to the cartographic units where the corresponding lithology predominated. Large-scale features such as the Guadalquivir river basin, running NE to SW, with predominantly w₀ values of 130 to 140 mm, and mountainous areas to the N-NW and to the SE, with smaller w₀ values, are clearly distinguishable. The Pearson correlation coefficient between the calculated w₀ values at the soil profiles and the corresponding values on the map of Fig. 2 was 0.45.

View larger version (73K): [in this window] [in a new window]   Fig. 2. Quantified lithological map of the soil water-holding capacity, w0. The Pearson correlation coefficient between the calculated w0 values of the soil profiles and the corresponding values on the map is 0.45.

View larger version (22K): [in this window] [in a new window]   Fig. 3. Ordinary and residual variograms for w0, with fitted exponential models. The residual values were calculated as the difference between the 319 pedotransfer function (PTF) estimates of w0 (RO1) and their corresponding values on the map of Fig. 2.

View larger version (74K): [in this window] [in a new window]   Fig. 4. Map of w0, produced using simple kriging with local varying means (SKlm) on blocks of 1 x 1 km, with values of the lithological map of Fig. 2 as local varying means.

View this table: [in this window] [in a new window]   Table 5. Descriptive statistics of the terms of the mean annual water balance, evaporation (E) and total runoff (Q) at 160 sites with meteorological data of Andalusia, computed with the model of Milly, using the corresponding w0 values from the map in Fig. 4.

View larger version (45K): [in this window] [in a new window]   Fig. 5. Budyko diagrams for different values of the soil water-holding capacity, w0. Observe how the water balance results can be adjusted to the empirical Budyko relationship by changing the average w0. The lower right-hand graph compares results for w0 taken from the map in Fig. 4 with those obtained with a single average w0 value, o = 110.5. E = actual evapotranspiration in liters; P = partitioning of the received average annual rainfall in liters.

Another interesting aspect of the w₀ is its spatial variability. Milly and Eagleson (1987) studied the influence of spatial variability in w₀ with the dynamic-statistical model of Eagleson, and concluded that its influence was most important when an appreciable amount of water was generated in the catchment. It is possible to define an equivalent soil which yields the same values of the water balance. The lower right-hand plot of Fig. 5 represents the average annual water balance of the 160 sites using (i) estimated values of w₀ at each site and (ii) the same average value _o at all the sites. For the first option, a greater variability was found, while the average relationship between E/P and R was the same for both. The figure illustrates the reduced importance of spatial variability in w₀ for describing the average annual water balance of the region. A comparison with the other plots of Fig. 5 suggests that soil water storage in the range of 150 to 200 mm yields a balance close to the Budyko empirical function.

Spatially averaged E, Q, and E/P values, with their SDs, are shown as a function of w₀ in Fig. 6 . Also included is the fitted logarithmic relationship between E and w₀. According to Milly and Dunne (1994), for every two-fold increase in w₀, E will increase about 70 mm in the interval 10 < w₀ < 600 mm. For Andalusia, the estimated increment was only 50 mm (Table 5). The line of constant evaporation according to Ministerio de Medio Ambiente (1998)(E = 448 mm) crossed the fitted curve in Fig. 6 at about w₀ = 170 mm, which indicates that when increasing w₀ by 55%, the results are close to those of Ministerio de Medio Ambiente (1998) and to the Budyko curve (Fig. 5). The location of confidence intervals within 1 SD in Fig. 6 shows how the variability in E increases and how Q and its variability decreases, as the soil water-storage capacity w₀ increases. In this way, for each w₀ value, evaporation and runoff add up to a constant value equal to the average annual rainfall, P. The mean annual runoff of Q = 136 mm, according to Ministerio de Medio Ambiente (1998), implies a water storage of w₀ = 150 mm for Andalusia. This difference is due to the higher precipitation value of Ministerio de Medio Ambiente (1998) than the value found here. Therefore, to obtain a comparable average soil water balance, the average value of w₀ in our approach should lie in the 150- to 170-mm interval. The w₀ values used by Ministerio de Medio Ambiente (1998) were predictions in which a soil water-storage capacity value was assigned to each land use class. There is neither any information on the errors nor an estimation of the accuracy of these values, which illustrates the scarcity of quantitative soil information for the whole area.

View larger version (22K): [in this window] [in a new window]   Fig. 6. Influence of the soil water-holding capacity, w0, on the components (evaporation, E, and total runoff, Q) of the average annual water balance in Andalusia, computed with the Milly model at 160 sites. The confidence intervals are within 1 SD. The dashed line represents values according to Ministerio de Medio Ambiente (1998). The E-plot shows a logarithmic fit (Milly and Dunne, 1994) between E and w0. The E/P-plot compares the results obtained with two methods.

Two possible reasons explain the underestimation of w₀. One is related to the coarse fraction, CF, in Eq. [1]. Table 3 showed that 277 of the 1015 horizons contained coarse material, with an average volumetric content of 30%, thereby reducing the average and w₀ values by approximately 8%. Although Hanson and Blevis (1979) found that water contents of coarse fragments at the WP ranged between 0.11 and 0.23 cm^–3 cm^–3, in many studies these water contents are assumed zero. For example, Wösten et al. (1999) did not take into account the coarse fraction when estimating the w₀ map of Europe, whereas Batjes (1996) did account for this when developing the world w₀ map. Batjes found the coarse fraction to be especially important for Lithosols, Redzinas, Rankers, and Regosols. The observed underestimation of w₀ by 45% could not be attributed to the coarse fraction alone.

A more feasible explanation for the underestimation of w₀ resides in the definition of the water content range of each horizon, _h, between WP and FC, defined as the volumetric water content at pressure heads of –1500 kPa (15000 cm) and –33 kPa (330 cm), respectively. In the dry side of the water retention curve, variations in the pressure head have only a small impact on the water content. According to Cassel and Nielsen (1986), the water content change between pressure heads of –800 to –3000 kPa is negligible, except for some fine-textured soils. At the wet end of the water retention curve, small variations in the pressure head have a great impact on the water content. Therefore, _h and w₀ depend strongly on the pressure head at which FC is defined. The existing literature on this issue is ambiguous. In the United Kingdom, a value of –5 kPa is commonly used; in the Netherlands, –10 kPa; and –33 kPa in the USA and Canada (McKeague, 1987; Batjes, 1996). Wösten et al. (1999) used –5 kPa for the w₀ map of Europe, while Batjes (1996) adopted the value of –33 kPa for the w₀ map of the world. For sandy soils in Ottawa, McKeague (1987) found that the use of a value of –5 kPa instead of –33 kPa duplicated w₀. This is illustrated in Table 6 for the RO2 PTF (Schaap et al., 2001). For each textural class, FC was calculated at both –10 and –33 kPa, and the corresponding difference in was evaluated. The increase in ranged from 43 to 142%, depending on the textural class, and excluding the data for sand. The increase for sand was very high due to the very small at –33 kPa. Since the water balance studies with which we compared our results did not provide a detailed definition of w₀, the observed differences can be attributed to different definitions of FC.

View this table: [in this window] [in a new window]   Table 6. Values of the volumetric water content range assuming field capacity (FC) to be defined at –10 and –33 kPa for the 12 USDA soil texture classes. The last column gives the difference between both, expressed as a percentage of –33kPa. Results are based on class-average values taken from Rosetta (Schaap et al., 2001).

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
A w₀ map of Andalusia was constructed from a soil database and a soil map of the region using a PTF and geostatistical interpolation. The w₀ was calculated using 10 published PTFs. The Schaap et al. (2001) PTF (RO1) was preferred for further use, yielding w₀ values between 0.4 and 235 mm, with a mean value of 110 mm and a SD of 48 mm. The soil map was not found to be very useful for the classification of the w₀ data in terms of the CRV. Only a lithology-based stratification of the data produced a CRV of as much as 0.16. Quantification of the soil map information yielded a w₀ map with a correlation coefficient of 0.45 between observations and the soil map data. Through block SKlm interpolation using the soil map as ancillary information, a final w₀ map was obtained. Since no direct observations were available, the quality of the w₀ map was evaluated by means of the average annual soil water balance, calculated at 160 sites using a simple bucket model (Milly, 1994b), with Budyko's empirical relationships and a national reference study (Ministerio de Medio Ambiente, 1998). Although the observed differences with Budyko's curves could be partly explained by the characteristics of the climate in the region, increasing w₀ toward 150 to 200 mm produced a satisfactory fit. Comparison of the obtained water balance results with those obtained by the Ministerio de Medio Ambiente (1998) study suggests that the estimated w₀ values should be multiplied on average by a factor of 1.45. We showed that this difference can be attributed entirely to an inconsistent definition of FC. Increasing the considered pressure head from –33 to –10 kPa increased w₀ by between 43 and 142%, depending on the soil textural class. Since the water balance studies with which we compared our results did not provide a detailed definition of w₀, the observed differences can be attributed to different definitions of FC. In summary, identifying accurate and precise spatially distributed w₀ values for the region is problematic since accurate quantitative soil information was very sparse or nonexistent. It is therefore of major importance that a greater cooperative effort be made among the involved institutions toward quantitative soil assessment and accurate digital soil mapping. Our experience with the data of Andalusia should be useful for avoiding or resolving similar problems in other studies trying to construct regional maps of the w₀ or related variables.

	ACKNOWLEDGMENTS

 
The authors are grateful to Dong Wang, Rien van Genuchten, and two anonymous reviewers for their useful suggestions and constructive comments.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 

• Bailey, H.P. 1979. Semi-arid climates: Their definition and distribution. p. 73–97. In A.E. Hall et al. (ed.) Agriculture in semi-arid environments. Springer-Verlag, Berlin.
• Batjes, N.H. 1996. Development of a world data set of soil water retention properties using pedotransfer rules. Geoderma 71:31–52.
• Bouma, J. 1989. Using soil survey data for quantitative land evaluation. Adv. Soil Sci. 9:177–213.
• Brutsaert, W. 1982. Evaporation into the atmosphere. Theory, history, and applications. D. Reidel, Dordrecht, the Netherlands.
• Budyko, M.I. 1974. Climate and life. Academic Press, New York.
• Cassel, D.K., and D.R. Nielsen. 1986. Field capacity and available water capacity. p. 901–926. In A. Klute (ed.) Methods of soil analysis: Part 1—Physical and mineralogical methods. Agron. Monogr. 9, 2nd ed. ASA and SSSA, Madison.
• Commission of the European Communities. 1985. Soil map of the European Communities (1:1000000). CEC-Directorate General VI (Agriculture and Environment) EU Commission, Luxemburg.
• Cornelis, W.M., J. Ronsyn, M. Van Meirvenne, and R. Hartmann. 2001. Evaluation of pedotransfer functions for predicting the soil moisture retention curve. Soil Sci. Soc. Am. J. 65:638–648.[Abstract/Free Full Text]
• Consejo Superior de Investigaciones Científicas–Instituto Andaluz de la Reforma Agraria. 1989. Mapa de suelos de Andalucía (E. 1:400000). (In Spanish.) Consejería de Medio Ambiente, Seville, Spain.
• De Jong, R., and J.A. Shields. 1988. Available water holding capacity maps of Alberta, Saskatchewan, and Manitoba. Can. J. Soil Sci. 68:157–163.
• Deutsch, C.V., and A.G. Journel. 1998. GSLIB. Geostatistical software library and user's guide, 2nd ed. Oxford University Press, New York.
• Eagleson, P.S. 1981. Dynamic hydro-thermal balances at macroscale. p. 289–357. In P.S. Eagleson (ed.) Land surface processes in atmospheric general circulation models. Papers presented at the World Climate Research Programme Study Conference, Greenbelt, Maryland. Cambridge Univ. Press, Cambridge.
• Food and Agriculture Organization of the United Nations. 1991. The digitised soil map of the world. World Soil Resources Rep. 67 (Release 1.0). FAO, Rome.
• Food and Agriculture Organization of the United Nations–United Nations Educational, Scientific and Cultural Organization. 1974. Soil map of the world (1:5000000). FAO, Rome.
• Gobin, A., P. Camping, and J. Feyen. 2001. Soil-Landscape Modelling to Quantify Spatial Variability of Soil Variables. Proc. Pedometrics Conf. Working Group on Pedometrics of the Int. Union of Soil Sci., 4th, Ghent. 19–21 Sept. 2001. Ghent Univ., Ghent.
• Gonçalves, M.C., L.S. Pereira, and F.J. Leij. 1997. Pedo-transfer functions for estimating unsaturated hydraulic properties of Portuguese soils. Eur. J. Soil Sci. 48:387–400.[CrossRef]
• Goovaerts, P. 1997. Geostatistics for natural resources evaluation. Oxford Univ. Press, New York.
• Hanson, C.T., and R.L. Blevis. 1979. Soil water in coarse fragments. Soil Sci. Soc. Am. J. 43:819–820.[Abstract/Free Full Text]
• Heuvelink, G.B.M., and M. Bierkens. 1992. Combining soil maps with interpolations from point observations to predict quantitative soil properties. Geoderma 55:1–15.
• Heuvelink, G.B.M., and E.J. Pebesma. 1999. Spatial aggregation and soil process modeling. Geoderma 89:47–65.
• Heuvelink, G.B.M., and R. Webster. 2001. Modelling soil variation: Past, present and future. Geoderma 100:269–301.[CrossRef][ISI]
• Kern, J.S. 1995. Geographic patterns of soil water-holding capacity in the contiguous United States. Soil Sci. Soc. Am. J. 59:1126–1133.[Abstract/Free Full Text]
• Kutílek, M., and D.R. Nielsen. 1994. Soil hydrology. Catena Verlag, Cremlingen-Destedt, Germany.
• Leenhardt, D., M. Voltz, M. Bornand, and R. Webster. 1994. Evaluating soil maps for prediction of soil water properties. Eur. J. Soil Sci. 45:293–301.
• Lettau, H. 1969. Evapotranspiration climatonomy. 1. A new approach to numerical prediction of monthly evapotranspiration, runoff and soil moisture storage. Monthly Weather Rev. 97:691–699.
• Lettau, H.H., and M.W. Baradas. 1973. Evaporation climatonomy II: Refinement of parameterization, exemplified by application to the Mabacan river watershed. Monthly Weather Rev. 101:636–649.
• McKeague, J.A. 1987. Estimating air porosity and available water capacity from soil morphology. Soil Sci. Soc. Am. J. 51:148–152.[Abstract/Free Full Text]
• Milly, P.C.D. 1993. An analytic solution of the stochastic storage problem applicable to soil water. Water Resour. Res. 29:3755–3758.[CrossRef]
• Milly, P.C.D. 1994a. Climate, interseasonal storage of soil water, and the annual water balance. Adv. Water Resour. 17:19–24.
• Milly, P.C.D. 1994b. Climate, soil water storage, and the average annual water balance. Water Resour. Res. 30:2143–2156.[CrossRef]
• Milly, P.C.D., and K.A. Dunne. 1994. Sensitivity of the global water cycle to the water-holding capacity of soils. J. Clim. 7:506–526.[CrossRef]
• Milly, P.C.D., and P.S. Eagleson. 1987. Effects of spatial variability on annual average water balance. Water Resour. Res. 23:2135–2143.
• Minasny, B., A.B. McBratney, and K.L. Bristow. 1999. Comparison of different approaches to the development of pedotransfer functions for water-retention curves. Geoderma 93:225–253.[CrossRef][ISI]
• Ministerio de Medio Ambiente. 1998. El libro blanco del agua en España. (In Spanish.) MMA, Madrid, Spain.
• Monreal, T.E., F. Cabezas Calvo-Rubio, and F. Estrada Lorenzo. 1999. La evaluación de los recursos hídricos en el libro blanco del agua en España. (In Spanish.) Ingeniería del Agua 6:125–138.
• Odeh, I.O.A., and A.B. McBratney. 2000. Using AVHRR images for spatial prediction of clay content in the lower Namoi Valley of eastern Australia. Geoderma 97:237–254.
• Pannatier, Y. 1996. VARIOWIN: Software for spatial data analysis in 2D. Springer Verlag, New York.
• Ratliff, L.F., J.T. Ritchie, and D.K. Cassel. 1983. Field-measured limits of soil water availability as related to laboratory-measured properties. Soil Sci. Soc. Am. J. 47:770–775.[Abstract/Free Full Text]
• Rawls, W.J., and D.L. Brakensiek. 1989. Estimation of soil water retention and hydraulic properties. In H.J. Morel-Seytoux (ed.) Unsaturated Flow in Hydrologic Modelling. Theory and Practice. Proc. of the NATO Advanced Research Workshop on Unsaturated Flow in Hydrologic Modeling, Arles, France. Kluwer, Dordrecht, the Netherlands.
• Rawls, W.J., T.J. Gish, and D.L. Brakensiek. 1991. Estimating soil water retention from soil physical properties and characteristics. Adv. Soil Sci. 16:213–235.
• Salter, P.J., and J.B. Williams. 1969. The influence of texture on the moisture characteristics of soils. V. Relationships between particle-size composition and moisture contents at the upper and lower limits of available water. J. Soil Sci. 20:126–131.
• Schaap, M.G., and F.J. Leij. 1998. Database-related accuracy and uncertainty of pedotransfer functions. Soil Sci. 163:765–779.[CrossRef]
• Schaap, M.G., F.J. Leij, and M.Th. van Genuchten. 1998. Neural network analysis for hierarchical prediction of soil hydraulic properties. Soil Sci. Soc. Am. J. 62:847–855.[Abstract/Free Full Text]
• Schaap, M.G., F.J. Leij, and M.Th. van Genuchten. 2001. Rosetta: A computer program for estimating soil hydraulic parameters with hierarchical pedotransfer functions. J. Hydrol. (Amsterdam) 251:163–176.
• Sinowski, W., A.C. Scheinost, and K. Auerswald. 1997. Regionalization of soil water retention curves in highly variable soilscapes. II. Comparison of regionalization procedures using a pedotransfer function. Geoderma 78:145–159.
• Soil Survey Staff. 1999. Soil taxonomy: A basic system of soil classification for making and interpreting soil surveys. 2nd ed. Agric. Handb. no. 436. USDA–NRCS, U.S. Gov. Print. Office, Washington, DC.
• Trueba, C., R. Millán, T. Schmid, C. Roquero, and M. Magister. 1999. Base de datos de propiedades edafológicas de los suelos españoles. (vol. 882 and 883) (In Spanish.) Informes Técnicos CIEMAT, Madrid.
• Vanderlinden, K. 2002. Análisis de procesos hidrológicos a diferentes escalas espacio-temporales. (In Spanish.) Ph.D. diss. Univ. of Córdoba, Córdoba, Spain.
• Vanderlinden, K., J.V. Giráldez, and M. Van Meirvenne. 2004. Assessing reference evapotranspiration by the Hargreaves method in Southern Spain. J. Irrig. Drain. Eng. 130:184–191.[CrossRef]
• van Genuchten, M.Th. 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44:892–898.[Abstract/Free Full Text]
• Van Meirvenne, M., K. Scheldeman, G. Baert, and G. Hofman. 1994. Quantification of soil textural fractions of Bas-Zaire using soil map polygons and/or point observations. Geoderma 62:69–82.
• Voltz, M., and M. Goulard. 1994. Spatial interpolation of soil moisture retention curves. Geoderma 62:109–123.
• Webster, R., and M.A. Oliver. 2001. Geostatistics for environmental scientists. John Wiley & Sons, Chichester, UK.
• Wösten, J.H.M., A. Lilly, A. Nemes, and C. Le Bas. 1999. Development and use of a database of hydraulic properties of European soils. Geoderma 90:169–185.[CrossRef][ISI]
• Wösten, J.H.M., Ya.A. Pachepsky, and W.J. Rawls. 2001. Pedotransfer functions: Bridging the gap between available basic soil data and missing hydraulic characteristics. J. Hydrol. (Amsterdam) 251:123–150.

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (2) Citing Articles via Google Scholar Google Scholar