USE OF THE JONASSON’S MODEL TO ESTIMATE THE VAN GENUCHTEN PARAMETERS FROM TEXTURAL DATA OF SOME SOILS IN SOUTHERN ITALY

001_Comegna(470)_01 14-06-2007 16:57 Pagina 1 J. of Ag. Eng. - Riv. di Ing. Agr. (2007), 1, 1-6 USE OF THE JONASSON’S MODEL TO ESTIMATE THE VAN GENUCHTEN PARAMETERS FROM TEXTURAL DATA OF SOME SOILS IN SOUTHERN ITALY Alessandro Comegna, Angelo Sommella, Gerardo Severino 1. Introduction In defining the hydraulic properties of unsaturated porous media, the field and laboratories methodologies that follow the instantaneous profile method [10,24] have proved to be capable of producing correct hydraulic characterization. However, they require complex, costly equipment and great accuracy in measuring water content θ and potential h [3,7]. An alternative method for estimating soil hydraulic properties is necessary, especially if large catchments are simulated, allowing for only basic soil data like soil texture and bulk density. This method is referred to as pedotranfer function technique (PTF) [2]. Several reviews on PTF development and use have been published [15,22,26]. Large databases, such as UNSODA [12] and HYPRES [26] are suitable for PTF development. PTFs, developed from regional databases, provide good results in regions having similar hydraulic and pedological characteristics. Some examples are provided by i) water retention PTFs developed in Belgium [23], which were more accurate than 13 others developed for the database in Northern Germany [19]; and ii) water retention PTFs developed in Italy [16], applicable to the Agri Basin in Basilicata. It is to be demonstrated, however, whether these observations can be generalised to other situations. We also need to understand which soils or other landscape characteristics shall prove to be similar in different regions for mutual reliability of the PTFs developed to be fully asserted. A new approach in developing PTFs consists in using geophysical and/or topographic information as a direct input in the PTFs. Ground-penetrating radar, electric-conductivity meters, etc. all provide spatial coverage that shows a potential to be included in PTFs [17]. Terrain attributes were used to recalibrate ___________ Paper received 02.01.2006; accepted 22.04.2006 Eng. ALESSANDRO COMEGNA, phD; Prof. ANGELO SOMMELLA, Full Professor; Eng. GERARDO SEVERINO, Assistant Professor, Department of Agricultural Engineering and Agronomy, University of Naples, ITALY. a PTF, and soil water retention exhibited strong dependence on terrain attributes in the study of Pachesky et al. [14]. Another frontier is the upscaling of PTF estimates: Scale dependence in soil hydraulic properties was recognized [5]. Currently these dependences are ignored and may limit PTF reliability, [16]. According to Tietje and Tapkenhinrichs [19], PTFs can be subdivided in three different groups: i) the point regression method which predicts the water content at certain matric potential by means of regression analysis [8,15]; ii) the functional parameter regression method which estimates the parameters of a closed form equation establishing the relationship between h and θ using regression techniques [15,24] and iii) the physical model method which uses the transformation between particle size (PSD) and pore size. This can be linear [8], non linear [1] or fractal [20,4]. Physical models are usually preferred but they are complex; it is sometimes difficult to find appropriate parameters, and calculations are not easily done and require suitable hardware and software, like in Arya and Paris’ model. Jonasson has recently provided a simple method to estimate the parameters in the van Genuchten soil water retention equation from PSD data similar to Arya and Paris’ method [1]), albeit expressed in a more direct analytical way. Given the relevance of this procedure in soil physics and hydrology, we evaluate Jonasson’s predictive model potential which was tested in this study using a data set of 15 soil samples, by comparing water retention curves obtained with the above-mentioned model with those from experimentation. 2. Materials and methods 2.1 Model Jonasson’s prediction method consists of three steps: i) transforming the grain size distribution PSD into the θ(h) curve; ii) setting out parameters in the van Genuchten equation, and iii) combining steps i) and ii). Arya and Paris’ model [1], if applied to each frac- 001_Comegna(470)_01 26-06-2007 12:52 Pagina 2 2 tion of PSD curve, gives a discrete rather than continuous description of water retention curve θ(h). To modify Arya and Paris’ method into a continuous analytical equation, Jonasson [10] suggests that Arya and Paris’ equation be rewritten as follows: (1) (7) The values for αAP and β that give the best results for the prediction of h from the grain size distribution of a given soil, can consequently be evaluated based upon known water retention curves and corresponding grain size distribution curves according to: (8) where (2) which specifies a continuous function of the head hSe as a function of dp, which is the grain diameter at a cumulative percentage P of the grain-size distribution, e is the void ratio, Wf is the weight fraction of soil (or a weighing factor) in a representative grain-size interval and αAP is Arya and Paris’ α factor. Hydraulic parameters are set based upon the θ(h) curve, in a similar way as the graphical procedure proposed by van Genuchten. In Jonasson’s method, pressure heads at two different effective saturations (Se=25% and 75%) are used to describe the “head sorting”, and thus to evaluate the n parameter. It is further assumed that Therefore, Jonasson obtained the relationship: (3) where (4) in which h25 is the pressure head at 25% effective saturation and h75 is the pressure head at 75% effective saturation. If the n value is known, one can determine the value of αvG, by rearranging the van Genuchten equation as follows: Finally, the h75 parameter, and subsequently αVG is calculated from d75 using: (9) where h75 is in m and d75 is in mm. 3. Applications The model illustrated above was used in assessing 15 soils of different texture from southern Italy (Fig. 1). Undisturbed soil cores were taken from the surface layer (Ap horizon) by driving a steel cylinder (76 mm x 76 mm) perpendicularly into the soil while carefully excavating soil from around the sampler. Then the core was removed; all cores were plugged at the top and bottom and stored at 4°C constant temperature before making laboratory measurements. Laboratory measurements were performed on each soil core to determine: (i) particle-size distribution; (ii) bulk density; (iii) particle density; (iv) water content at saturation, and (v) the water retention curve. Sand particle size distribution was determined using sieve analysis while silt and clay were determined using the hydrometer method [6]. Bulk and particle density was calculated on an oven-dry basis. Prior to determining the water retention curves, the soil cores were gradually saturated from below using the de- (5) The value of αvG may be obtained from any correspo (...truncated)