Evaluation of pedotransfer functions in predicting the soil water contents at field capacity and wilting point

Transcription

1 Proceedings of The Fourth International Iran & Russia Conference 492 Evaluation of pedotransfer functions in predicting the soil water contents at field capacity and wilting point J. Givi 1 and S. O. Prasher 2 1 Soil Science Department, Faculty of Agriculture, Shahrekord University, P.O.Box 115, Shahrekord, Iran. Phone: , givi@agr.sku.ac.ir; 2 Bioresource Engineering Department, Macdonald Campus, McGill University, 21,111 Lakeshore Road, Ste-Anne de Bellevue, Quebec, Canada, H9X 3V9. Phone: (514) , shiv.prasher@mcgill.ca Abstract Thirteen pedotransfer functions (PTFs), namely Rosetta PTF, Brakensiek, Rawls, British Soil Survey Topsoil, British Soil Survey Subsoil, Mayr-Jarvis, Campbell, EPIC, Manrique, Baumer, Rawls-Brakensiek, Vereecken, and Hutson were evaluated for accuracy in predicting the soil moisture contents at field capacity (FC) and wilting point (WP), of fine-textured soils of the Zagros mountain region of Iran. PTFs were developed using the laboratory measurements made on soil moisture at FC and WP, particle size distribution, bulk density, and organic matter content. PTFs were evaluated on the basis of mean squared deviation (MSD) between the observed and predicted values. Results agreed with the concept that the PTFs developed on soils of similar properties to the ones under study generally perform better than the others. In the case of the Zagros mountain soils, the British Soil Survey and Brakensiek PTFs were found to be the best methods. Since the soils under study had a wide range of organic matter contents (0.2% to 5.5%), the better performance of these PTFs may also be explained by the fact that they happen to be the only ones that require organic matter content as input. Rosetta, a software package that involves an artificial neural network approach, was of intermediate value in estimating soil moistures of the soils in question. This was attributed to the fact that the texture and the bulk density of the Zagros soils were not in the range of those used to develop Rosetta. Keywords: Evaluation of pedotransfer functions, Field capacity, Rosetta software SOILPAR2 software, Soil water content estimation, Wilting point. Introduction Soil water contents at field capacity and wilting point are used to calculate the water depth that should be applied by irrigation (Hansen et al., 1980), and to determine water availability, which is a crucial factor in assessing the suitability of a land area for producing a given crop (Sys et al., 1991). If the area under investigation is relatively small or known to be quite homogeneous with respect to soil physical makeup and topography, determinations of soil moisture contents at FC and WP at a reasonable number of sampling sites should provide accurate estimates. However, if the area being evaluated is large enough to exhibit substantial spatial variability of soil water availability, it is virtually impossible to perform enough measurements to provide good estimates within the temporal and financial constraints of the project. In such cases, inexpensive and rapid ways to estimate the parameters are needed (Schaap et al., 2001). Many indirect methods for estimation of water content at field capacity and wilting point have been proposed in the literature. Most of these methods can be called pedotransfer functions (PTFs) (Bouma and Van Lanen, 1987), because they translate existing surrogate data into soil hydraulic data (Schaap et al., 2001). However, since PTFs are often developed

2 Proceedings of The Fourth International Iran & Russia Conference 493 empirically, their applicability may be limited to the data set used to define the method (Donatelli et al., 1996 and Wosten et al., 1999). Moreover, the available pedotransfer procedures can produce substantially different estimates. Thus, users have a difficult task in selecting the more appropriate PTF for their application (Acutis and Donatelli, 2003). Another problem is that PTFs provide estimates with a modest level of accuracy. It would therefore be useful if PTF predictions could include reliability measures (Schaap et al., 2001). Pedotransfer procedures are classified as Point Pedotransfers and Function Pedotransfers (Cornelis et al., 2001 and Acutis and Donatelli, 2003). Point pedotransfers estimate the water content of the soil at certain matric potentials. Function pedotransfers predict the parameters of a closed form analytical equation, such as the model of Brooks and Corey (Brooks and Corey, 1964) or the van Genuchten equation (Van Genuchten, 1980). Neural network analysis has also been used to establish empirical PTFs (Pachepsky et al., 1996; Schaap and Leij, 1998; Schaap et al., 1998 and Minasny and McBratney, 2002). An advantage of neural networks over traditional PTFs is that they do not require a priori model concept. The optimal and possibly nonlinear relations that link input data to output data are obtained and implemented in an iterative calibration procedure. As a result, neural network models typically extract the maximum amount of information from the data (Schaap et al., 2001). To facilitate application of the PTFs, Schaap et al. (2001) developed Rosetta, a computer program that implements some of the models published by Schaap and Bouten (1996), Schaap and Leij (1998), Schaap et al. (1998) and Schaap and Leij (2000). Other programs developed to estimate soil hydraulic properties include SOILPAR2 (Acutis and Donatelli, 2003). Rosetta and SOILPAR2 are stand-alone programs. Rosetta uses a neural network for prediction and the bootstrap approach to perform uncertainty analysis. SOILPAR2 can compute estimates of soil hydrological parameters by up to 15 procedures, and compares the estimates with measured data using statistical indices and graphs. The objective of this paper is to evaluate the general applicability and the prediction accuracy of some pedotransfer functions in estimating water availability in the fine textured soils of Iran s Zagros mountain region. Materials and methods 1. Data collection and soil sample analysis Sixteen soil samples were collected from different horizons of five clayey or clay loamy soil profiles, located in the semiarid parts of the Zagros mountain foot slopes, Chahartagh area, Ardal township, Chaharmohal and Bakhtiari province, Iran. Particle size distribution was determined by the hydrometer method (Gee and Bauder, 1986). Organic matter content was determined by the method of Walkley and Black (Nelson and Sommers, 1982). The clod method (Blake and Hartge, 1986) was used to determine bulk density. The moisture contents at field capacity and wilting point were determined with a pressure plate apparatus (Cassel and Nielsen, 1986) at -33 and kpa, respectively. 2. Pedotrasfer functions

3 Proceedings of The Fourth International Iran & Russia Conference Point pedotransfers Eight point PTFs were used to predict the water content at specific matric potentials. These PTFs are based on a series of multiple linear regression equations that link the water content at a certain matric potential to some of the soil physical properties (Cornelis et al., 2001). Table 1 shows different point pedotransfer methods and their inputs and outputs Function pedotransfers Function pedotransfers estimate the parameters of retention functions. Four were used in this study Campbell Campbell (1985) estimated the Campbell function parameters (Campbell, 1974) from equation 1: ( θ ) b ϕ m = ϕe θs for e ϕ ϕ (1) where ψ m (J kg) is matric potential, ψ e (J kg) is the air entry water potential (potential at which the largest water filled pores just drain), θ is the moisture content (m 3 m -3 ), θ s is the moisture content at saturation (m 3 m -3 ), b is the slope of the plot of ln ψ vs ln θ and ψ is water potential (J kg). The values of ψ e and b are determined by fitting a straight line through the log-log plot of moisture release data. The slope and intercept of the best-fit line are used to find ψ e and b Rawls - Brakensiek Rawls and Brakensiek (1989) estimated the Brooks and Corey function parameters (Brooks and Corey, 1964)(Equations 2 and 3): θ ( ϕ) = θs for ϕ ϕb (2) θ ( ϕ) θ + ( θ θ )( ϕ ϕ) λ = r s r b for b ϕ ϕ (3) where ψ b is the bubbling pressure (kpa), θ r is the residual moisture content (m 3 m -3 ), ψ is the matric potential (kpa) and λ is a pore-size distribution index (dimensionless). Their regression equations were formulated for natural soils and use porosity, clay content, and sand content as input variables (Cornelis et al., 2001) Vereecken Vereecken et al. (1989) used multiple linear regression with sand and clay contents, organic carbon content, and bulk density data from undisturbed samples of 182 horizons of 40 Belgian soil series to solve for the parameters of the van Genuchten equation (Equation 4)(Van Genuchten, 1980): n m θ ( ϕ) = θr + ( θs θr ) 1 ( 1+ α ϕ ) (4)

4 Proceedings of The Fourth International Iran & Russia Conference 495 where α (kpa -1 ), n and m (dimensionless) are regression coefficients (Cornelis et al., 2001). This equation has an inflection point that allows better performance than the Brooks and Corey model, particularly near saturation (Van Genuchten and Nielsen, 1985). Therefore, the latter is most frequently used to model the moisture retention curve Mayr and Jarvis Mayr and Jarvis (1999) estimated the parameters of the Hutson and Cass hydraulic functions (Hutson and Cass, 1987) from soil texture, bulk density and organic carbon content. Hutson and Cass (1987) introduced a modified Brooks-Corey type model (Brooks and Corey, 1964), in which water retention is described in three conditions (Equations 5,6 and 7): a) θ = θi θ i = 2 bθ s ( 1+ 2b) (5) b) θi s a 1 b θ (dry range) = θ ( ϕ ) θ (6) 2b [ ] 2 c) θ θi (wet range) θi s s 1 2 θ = θ θ ϕ a ( i s ) s θ θ (7) θ where θ i is water content at a matching point (m 3 m -3 ) and a and b are fitting parameters. All four methods require particle size distribution and bulk density as inputs. The Mayr and Jarvis (Mayr and Jarvis, 1999) and Vereecken et al. (Vereecken et al., 1989) methods also require organic carbon content as input (Acutis and Donatelli, 2003). 3. Description of the softwares used 3.1. Rosetta Rosetta is able to estimate the van Genuchten water retention parameters (Van Genuchten, 1980) and saturated hydraulic conductivity (Ks), as well as unsaturated hydraulic conductivity parameters, based on Mualem s (1976) pore size model (Schaap et al., 2001). The retention function is given by the equation 8: n ( ) ( )[ ( ) ] ( 1 1 θ h = θ n) r + θs θr 1 1+ αh (8) where θ (h) is the volumetric water content (m 3 m -3 ) at the suction h (cm, taken positive for increasing suctions). The parameters θ r (m 3 m -3 ) and θ s (m 3 m -3 ) are residual and saturated water contents, respectively; α (>0, in cm -1 ) is related to the inverse of the air entry suction, and n (>1) is a measure of the pore-size distribution (Van Genuchten, 1980). To render the PTFs of Rosetta as widely applicable as possible, a large number of records of soil hydraulic data and corresponding predictive soil properties were obtained from three databases (Schaap and Leij, 1998 and Schaap et al., 2001). Most of the samples were derived from soils in temperate to subtropical climates of North America and Europe. The textural classes of the datasets used for water retention are: sand, loamy sand, sandy loam, loam, sandy clay loam, silty loam, clay loam and silty clay loam (Schaap et al., 2001). Data from clays, silty clays, sandy clays and silts are rare SOILPAR 2

5 Proceedings of The Fourth International Iran & Russia Conference 496 SOILPAR 2 is a program for estimating soil parameters (Stockle et.al., 2003). It allows computing estimates of soil hydrological parameters using 15 procedures (see section 2.2) and comparing the estimates with measured data using both statistical indices and graphics. Twelve methods estimate one or more of the following characteristics: soil water content at predefined soil matrix tension, saturated hydraulic conductivity and bulk density. Three methods estimate the parameters of well-known soil water retention functions: Brooks-Corey, Hutson-Cass and van Genuchten, and one estimates both saturated soil hydraulic conductivity and the Campbell parameters of the soil water retention curve (Acutis and Donatelli, 2003). 4. Evaluation methods Usually, a common method to evaluate models is to plot the measured values against the predicted values and the correlation between them is used for model evaluation (Kobayashi and Salam, 2000). Ideally, this relationship should be linear with a slope of unity and intercept of zero. Although this method may be satisfactory for fitting an empirical model to observed data, it is inadequate for evaluating the performance of mechanistic models (Kobayashi and Salam, 2000). Generally, correlation-based statistics in conjunction with two other statistics, root mean squared error (RMSE), and mean deviation (MD), also called bias, are used to evaluate the performance of models. However, these statistics are not consistent with each other in their assumptions, therefore, Kobayashi and Salam (2000) have derived the following relationship among mean squared deviation (MSD), squared bias (SB), mean squared variation (MSV), squared difference between standard deviations (SDSD), and the lack of positive correlation weighted by the standard deviations (LCS) as follows: MSD = SB + MSV = SB + SDSD + LCS (9) 1 n x i y i i= 1 2 (10) and xi is the simulated value, yi is the measured value, and n is the number of observations. SB = ( x y) 2 (11) Where x and y are the average values of measured and predicted data. where MSD = n ( ) n MSV = n [ ( xi x) ( yi y) ] i = (12) ( ) 2 SDSD = SD s SD m (13) 1 2 n 2 Standard deviation of simulated values, SD s = 1 n ( xi x) (14) i= n 2 Standard deviation of measured values, SD m = 1 n ( yi y) (15) i= 1 LCS = 2 SDsSDm ( 1 r) (16) where r is the correlation coefficient.

6 Proceedings of The Fourth International Iran & Russia Conference 497 The accuracy of model performance is usually judged from the correlation coefficient, r, however, MSD is a more comprehensive evaluator of model performance. It includes LCS, which incorporates the role of r in the computation of MSD. Moreover, the values of SB and SDSD can also provide greater insight into performance of a model. Results and Discussion Mean squared deviation (MSD) and its components associated with the different methods of estimating field capacity and wilting point are given in Tables 2 and 3, respectively. The best and the worst observed and predicted values of field capacity and wilting point, on the basis of MSD computations, are given in Figures 1 and 2, respectively. The data points of the PTFs with the least MSDs are closer to the 1:1 line, whereas those with highest MSD are not. Instead of giving figures for all 13 PDFs, the regression parameters for field capacity are given in Table 2, and those for wilting point are given in Table 3. Ideally the intercept should be close to 0, however, for field capacity the intercepts were always higher than 0.13, and all significantly greater than 0 (P < 0.05) (Table 2). Similarly, the slope should be close to 1, however, in all the cases the slopes were less than 0.4, and significantly less than 1 (P <0.05). In case of wilting point, the intercepts were significantly higher than 0, and the slopes were significantly lower than 1 (Table 3). However, such statistical testing of intercept and slope is much more rigorous, and so other methods are used to evaluate model performance. A correlation coefficient is used to reflect how best the predicted data matches with the observed data. MSD also represents the deviation of predicted values from the observed ones, and it does so in a more comprehensive manner (Kobayashi and Salam, 2000). The inferences drawn on the basis of just correlation coefficient can be erroneous. For example, the positive correlation between the observed and predicted FC for Hutson is the poorest, however, the MSD value is which would make it a reasonable model in light of the MSD range of to for all models (Table 2). On the other hand, Rosetta has the highest correlation among all PTFs. However, its MSD value is 0.041, which is on the higher side of MSDs, indicating a poorer performance by the model. This is because although the observed and predicted data has a good linear relationship, the prediction is greatly biased, as indicated by a high SB value (SB = 94% of the MSD). For the Hutson model, the SB value was relatively low (SB=73% of MSD). In almost all cases, the contribution of SDSD and LCS values in the calculation of MSD appears to minimal (Table 2). It is interesting to note that the correlation for Vereecken is numerically the lowest. Also, although the negative correlation for Vereecken indicates that the model has completely failed in predicting FC, its MSD value is low (0.015), which would indicate that the model has performed reasonably well. So, one must consider both correlation coefficient and MSD values together in reaching a conclusion about model performance. In the case of wilting point, the correlation between the observed and predicted values is satisfactory for most PTFs (Table 3). The MSD for BSSSubsoil is the lowest. Although the correlation for Mayr-Jarvis is the highest, its MSD value is 29 times higher than that for BSSSubsoil. This higher value is due to a high value of SB for Mayr-Jarvis, indicating a high bias. It can also be seen from Figure 3 that the prediction is better in case of BSSSubsoil as compared to Mayr-Jarvis. Overall, the contributions of SDSD and LCS to MSD are relatively low as compared to SB (Table 3). Like the results for field capcity, although MSD is relatively

7 Proceedings of The Fourth International Iran & Russia Conference 498 low for Vereecken, the correlation between the observed and predicted WP is the poorest. Therefore, both correlation coefficient and MSD values should be used together in evaluating model performances. In the MSD based analysis, the best PTFs for estimating FC and WP are British Soil Survey Topsoil, British Soil Survey Subsoil, and Brakenssiek (Tables 2 and 3). This corroborates the results obtained by Donatelli et al. (1996). Many researchers emphasize that pedotransfer functions should be applied to soils whose characteristics are similar to those of the soils from which the functions were derived (Hutson and Cass, 1987; Schaap and Leij, 1998; Mayr and Jarvis, 1999; Cornelis et al., 2001 and Nemes et al., 2002). In the development of the Brakensiek and British Soil Survey PTFs, LEACHM model (Hutson and Wagenet, 1992) was used (Acutis and Donatelli, 2003). The Hutson and Cass hydraulic functions (Hutson and Cass, 1987) are included in this model (Mayr and Jarvis, 1999). In the development of Hutson and Cass hydraulic functions, mainly clay and clay loam soils were used. The bulk density of the samples was usually close to 1.6 g / cm3. In our study, soil textures are clay loam and clay, and the bulk densities range from 1.6 to 1.7 g/cm3. Thus, the British Soil Survey and Brakensiek PTFs simulations of moisture contents at FC and WP were found to be better in our study. Moreover, given the range of organic matter contents of the soils in our study was fairly wide (0.2 to 5.5 percent), and the fact that these PTFs also require organic matter content as input (Table 1), better predictions with these models are justified. Rawls et al. (2003) concluded that organic carbon and bulk density improve estimates of soil water retention derived from soil texture. Rawls and Brakensiek (1982), Rawls et al. (1983), De Jong (1983), Jamison and Kroth (1958), Petersen et al. (1968), Riley (1979), Ambroise et al. (1992) and Kern (1995) all found that inclusion of organic carbon content as an input to PTFs was useful in improving estimates of soil water at 33 and 1500 kpa. Thus, inclusion of organic matter in PTFs could be expected to improve the performance. References Acutis M., Donatelli M. (2003). SOILPAR 2.00: software to estimate soil hydrological parameters and functions. Eur. J. Agron. 18: Ambroise B., Reutenauer D., Viville D. (1992). Estimating soil water retention properties from mineral and organic fractions of coarse-textured soils in the Vosges mountains of France. In: Van Genuchten M.Th., Leij F.J., Lund L.J. (eds.), Indirect Methods for Estimating the Hydraulic Properties of Unsaturated Soils. University of California, Riverside, CA, pp Blake G.R., Hartge K.H. (1986). Bulk density. In: Klute, A. (ed.), Methods of soil analysis. Part 1, 2nd ed., Agron. Monogr.9. ASA and SSSA, Madison, WI., pp Boucneau G., Van Meirvenne M., Hofman G. (1998). Comparing pedotransfer functions to estimate soil bulk density in northern Belgium. Pedologie-Themata. 5: Bouma J., Van Lanen J.A.J (1987). Transfer functions and threshold values: from soil characteristics to land qualities. In: Beek K.J., Burrough P.A., McCormack D.A. (eds.), Quantified land evaluation procedures. Proceedings of the international workshop on quantified land evaluation procedures, 27 April-2 May 1986, at ITC, the Netherlands, Publ. no. 6, Brooks R.H., Corey A.T. (1964). Hydraulic properties of porous media. Colorado State University, hydrological paper No. 3, p 27. Campbell G.S. (1974). A simple method for determining unsaturated conductivity from moisture

8 Proceedings of The Fourth International Iran & Russia Conference 499 retention data. Soil Sci. 117: Campbell G.S. (ed.) (1985). Soil physics with basics. Elsevier, Amsterdam. Cassel D.K., Nielsen D.R. (1986). Field capacity and available water capacity. In: Klute A. (ed.), Methods of soil analysis. Part 1, 2nd ed., Agron. Monogr. 9. ASA and SSSA, Madison, WI., pp Cornelis W.M., Ronsyn J., Meirvenne M.V., Hartmann R. (2001). Evaluation of pedotransfer functions for predicting the soil moisture retention curve. Soil Sci. Soc. Am. J. 65: De Jong R. (1983). Soil water desorption curves estimated from limited data. Can. J. Soil Sci. 63: Donatelli M., Acutis M., Laruccia N. (1996). Pedotransfer functions: Evaluation of methods to estimate soil water content at field capacity and wilting point. www. Isci.it / mdon / research / bottom_modeling_cs.htm, pp Gee G.W., Bauder J.W. (1986). Particle size analysis. In: Klute A. (ed.), Methods of soil analysis. Part 1, 2nd ed., Agron. Monogr. 9. ASA and SSSA, Madison, WI., pp Hansen V.E., Israelsen O.W., Stringham G.E. (Eds.) (1980). Irrigation principles and practices. 4th ed., John Wiley & sons, New York. Hutson J.L., Cass A. (1987). A retentivity function for use in soil water simulation models. J. Soil Sci. 38: Hutson J.L., Wagenet R.J. (1992). LEACHM, Leaching Estimation And Chemistry Model. Series No. 92.3, Dept. of Soil, Crop and Atmospheric Sci. Research, Cornel University, New York, p Jamison V.C., Kroth E.M. (1958). Available moisture storage capacity in relation to texture composition and organic matter content of several Missouri soils. Soil Sci. Soc. Am. Proc. 22: Kern J.S. (1995). Evaluation of soil water retention models based on basic soil physical properties. Soil Sci. Soc. Am. J. 59: Kobayashi K., Salam M.U. (2000). Comparing simulated and measured values using mean squared deviation and its components. Agron. J. 92: Mayr T., Jarvis N.J. (1999). Pedotransfer functions to estimate soil water retention parameters for a modified Brooks-Corey type model. Geoderma 91: 1-9. Minasny B., McBratney A.B. (2002). The Neuro-m method for fitting neural network parametric pedotransfer functions. Soil Sci. Soc. Am. J. 66: Mualem Y. (1976). A new model predicting the hydraulic conductivity of unsaturated porous media. Water Resour. Res. 12: Nelson D.W., Sommers L.E. (1982). Total carbon, organic carbon, and organic matter. In: Page A.L. (ed.). Methods of soil analysis. Part 2, 2nd ed., Agron. Monogr. 9. ASA and SSSA, Madison, WI. pp Nemes A., Schaap M., Wosten H. (2002). Validation of international scale soil hydraulic pedotransfer functions for national scale applications. 17th World Congress of Soil Science, August 2002, Thailand. Pachepsky Y.A., Timlin D., Varallyay G. (1996). Artificial neural networks to estimate soil water retention from easily measurable data. Soil Sci. Soc. Am. J. 60: Petersen G.W., Cunningham R.L., Matelski R.P. (1968). Moisture characteristics of Pensylvania soils: II. Soil factors affecting moisture retention within a textural class-silt loam. Soil Sci. Soc. Am. Proc. 32: Rawls W.J., Brakensiek D.L. (1982). Estimating soil water retention from soil properties. J. Irrig.

9 Proceedings of The Fourth International Iran & Russia Conference 500 Drain. Div., Proc. ASCE. 198 IR2, Rawls W.J., Brakensiek D.L. (1989). Estimation of soil water retention and hydraulic properties. In: Morel S. (ed.), Unsaturated flow in hydrologic modeling: Theory and practice. Kluwer Academic Publishers, pp Rawls W. J., Brakensiek D.L., Soni B. (1983). Agricultural management effects on soil water processes. Part I: Soil water retention and Green Camnt parameters. Trans. ASAE 26: Rawls W.J., Pachepsky Y.A., Ritchie J.C., Sobecki T.M., Bloodworth H. (2003). Effect of soil organic carbon on soil water retention. Geoderma (in press). Riley H.C.F. (1979). Relationship between soil moisture holding properties and soil texture, organic matter, and bulk density. Agric. Res. Exp. 30: Schaap M.G., Bouten W. (1996). Modeling water retention curves of sandy soils using neural networks. Water Resour. Res. 32: Schaap M.G., Leij F.J. (1998). Database related accuracy and uncertainty of pedotransfer functions. Soil Sci. 163: Schaap M.G., Leij F.J. (2000). Improved prediction of unsaturated hydraulic conductivity with the Mualem-Van Genuchten model. Soil Sci. Soc. Am. J. 64: Schaap M.G., Leij F.J., Van Genuchten M.Th. (2001). Rosetta: a computer program for estimating soil hydraulic parameters with hierarchical pedotransfer functions. J. Hydrol. 251: Stockle C.O., Donatelli M., Nelson R. (2003). CropSyst, a cropping systems simulation model. Eur. J. Agron. 18: Sys C., Van Ranst E., Debaveye J. (Eds.) (1991). Land evaluation. Part I, General Administration for Development Cooperation, Brussels, Belgium..Van Genuchten M.Th. (1980). Predicting the hydraulic conductivity of unsaturated soil. Soil Sci. Soc. Am. J. 44: Van Genuchten M.Th., Nielsen D.R. (1985). On describing and predicting the hydraulic properties of unsaturated soils. Ann. Geophys. 3: Vereecken H., Maes J., Feyen J., Darius P. (1989). Estimating the soil moisture retention characteristics from texture, bulk density and carbon content. Soil Sci.148: Wosten J.H.M., Lilly A., Nemes A., Le Bas C. (1999). Development and use of a database of hydraulic properties of Europian soils. Geoderma 90: Table 1. Point pedotransfers and their inputs and outputs (Acutis and Donatelli,2003) Methods Inputs Outputs Baumer PSD 1,OC 2 BD 3, FC 4, WP 5 Brakensiek PSD, OC, BD SWC 6 BSSS 7 PSD, OC, BD SWC BSST 8 PSD, OC, BD SWC EPIC PSD, BD FC, WP Hutson PSD, BD SWC Manrique PSD, BD FC, WP Rawls PSD BD, FC, WP 1 Particle size distribution 2 Organic carbon 3 Bulk density 4 Field capacity 5 Wilting Point 6 Soil water content at several tensions 7 British Soil Survey Subsoil 8 British Soil Survey Topsoil

10 Proceedings of The Fourth International Iran & Russia Conference 501 Table 2: Evaluation of the PTFs for estimating field capacity PTFs r SDs SDm SB SDSD LCS MSD Intercept Slope BSST * * Rawls * * Brakensiek * * BSSS * * Baumer * * EPIC * * Vereecken * * Hutson * * Campbell * * Ra-Brak * * Rosetta * * Manrique * * Mayr-Jarvis * * 1 British Soil Survey Topsoil 2 British Soil Survey Subsoil 3 Rawls-Brakensiek SDs: Standard deviation of simulated values (m3/m3), SDm: Standard deviation of measured values (m3/m3), SB: squared bias (m3/m3), SDSD: squared difference between standard deviations (m3/m3), LCS: lack of positive correlation weighted by the standard deviations (m3/m3), MSD: mean squared deviation (m3/m3). * Intercept significantly different from 0, and slope significantly different from 1, P < Table 3: Evaluation of the PTFs for estimating wilting point PTFs PTFs R SDs SDm SB SDSD LCS MSD Intercept Slope BSSS * * BSST * * Brakensiek * * EPIC * * Hutson * * Rawls * * Campbell * * Ra-Brak * Baumer * Vereecken * * Manrique * Rosetta * * Mayr-Jarvis * * 1 British Soil Survey Subsoil 2 British Soil Survey Topsoil 3 Rawls-Brakensiek SDs: Standard deviation of simulated values (m3/m3), SDm: Standard deviation of measured values (m3/m3), SB: squared bias (m3/m3), SDSD: squared difference between standard deviations (m3/m3), LCS: lack of positive correlation weighted by the standard deviations (m3/m3), MSD: mean squared deviation (m3/m3). * Intercept significantly different from 0, and slope significantly different from 1, P < 0.05.

11 Proceedings of The Fourth International Iran & Russia Conference 502