On the characterization of soil hydraulic behavior at the hillslope scale in modeling erosional and runoff events in a Mediterranean environment

Modeling key hydrologic processes and their interactions at the hillslope scale is now considered of great importance for the reliable applications of distributed models in an entire catchment. This is particularly true for models intended for simulating soil erosion processes and gaining a better understanding of the impact of anthropogenic actions on runoff and sediment yield in the hilly landscapes of arid or semi-arid environments. Examples are some recent EU research projects (e.g. the MED ALUS project) that aimed at addressing problems of intense land degradation in the Mediterranean areas. Some modelers follow the approach of dividing the catchment into a hierarchy of several sub-cachments. Each sub-catchment is then modeled by one or more statistically representative hillslope catenas. However, whichever the modeling approach, reliable catchment simulations hinge upon the good description of soil hydrologic behavior. Because of its boundary position between atmosphere and groundwater, soil plays a key role within the hydrological cycle as it contributes chiefly to the partitioning process of incident water, either rainfall or irrigation water, in runoff and infiltration. Soil hydraulic properties are typically described by the water retention, θ( h ), and hydraulic conductivity, K (θ) functions. These functions synthesize the highly nonlinear relationships between the soil-water pressure head h, volumetric water content 6, and hydraulic conductivity K. Determining the soil hydraulic properties by direct procedures is in general not easy. Employing direct methods becomes practically impossible if the soil hydraulic characterization requires also quantification of soil spatial variations in the investigated area. Parameter optimization approaches are suitable and accurate methods to determine simultaneously the soil hydraulic properties at local scale. Even if they simplify somehow the experimental tests to be carried out on undisturbed soil cores or field plots, their implementation at large scale (e.g. an entire catchment) can be still inefficient and very costly. Recently developed methods based on the so-called ‘pedo-transfer functions’ can be employed to face the problem in a cost-effective way. Basically, pedo-transfer functions are empirical regression functions that enable the soil hydraulic properties to be estimated from more easily measurable or already available soil physical data. Furthermore, distributed hydrological modeling should also account for a description of the spatial variability of soil properties. In fact, as the scale increases, for examples to embrace a catchment, the natural spatial variability of the soil characteristics becomes an important factor that can influence the assessment of the overall system’s response to specific conditions. The spatial structure of soil variability is also of crucial significance to define (i.e. to upscale) appropriately the model parameters at the grid of the numerical scale from the data obtained at the measurement scale. The major objectives of this paper is to overview some efficient methods being employed to determine the soil hydraulic properties at different scales within a hillslope and to discuss suitable approaches for assessing effective soil hydraulic parameters at the numerical grid squares. An example is presented to show the potential of a recently developed scaling technique. To demonstrate the effectiveness of using a parameter optimization method and pedo-transfer functions to gain a sound soil hydraulic characterization at hillslope scale, experimental investigations were carried out in a hillslope of the Sauro River catchment, the main tributary of the Agri River (Basilicata Region, Italy). The catchment is located in the middle Agri valley and represents an interesting area for site-specific studies on surface erosion and land degradation. The sampled area is a transect which is nearly parallel to the main slope and oriented in the N-S direction. Undisturbed soil samples were collected from the topsoil at 100 sites, spaced 50 m apart.