Effectiveness of Best Management Practices (BMPs) for erosion control at the watershed scale

In Europe, as well as in many other parts of the world, soil erosion is the main cause of land degradation. The consequences of this process are a loss of soil fertility, soil biodiversity, and depletion of a natural resource with a relevant economic impact. The EU Member States are called to identify the areas having a high erosion risk and to adopt mitigation measures to improve water quality and decrease land degradation. To do this, it is necessary to develop methodologies for quantifying the soil erosion and supporting watershed management. The general aim of the present Ph.D. thesis is to address the problem of soil erosion and its management in a Mediterranean watershed trough field measurements, modelling activities, and scenario development to reduce soil erosion based on European and regional Policies. In this context, the specific aims are to: (1) apply the Soil and Water Assessment Tool (SWAT) model in simulating total streamflow and sediment load; (2) compare the results generated by the SWAT model with those obtained through the Annualized Agricultural Non-Point Source (AnnAGNPS) model and from two European scale erosion models (RUSLE2015 and PESERA); (3) quantify the soil erosion for the current management and identify the sediment source areas where specific Best Management Practices (BMPs) to reduce soil erosion are needed; (4) detect the effectiveness of check dams as a management practice to reduce channel erosion in a Mediterranean stream. The methodological approach and the economic analysis carried out in order to estimate the feasibility of the BMPs were tested in the Carapelle watershed (SE Italy), where continuous measurements of streamflow and sediment concentration were collected over a 5‐year period, on a half‐hour timescale, processed on a daily timescale. After the sensitivity analysis, the SWAT model was calibrated and validated both for hydrology and sediment load at daily timescale either manually and by means of the automated procedure (SWAT-CUP). To further improve sediment simulation performance of SWAT and identify better the sediment source areas two different approach were used: (i) a seasonal calibration scheme, in which data recorded in the dry and wet seasons were used to calibrate sediments separately; and (ii) a finer watershed discretization, in order to sub-divide better the steep slope areas from the plan areas. Moreover, SWAT and AnnAGNPS models were calibrated and validated, using the same input dataset, at monthly timescale for a 4-years period (2007-2008; 2010-2011) and their results were compared. Performances were evaluated using the Nash and Sutcliffe efficiency coefficient (NSE), the percent bias (PBIAS) and the coefficient of determination (R2). Statistics showed generally a satisfactory efficiency. At watershed scale, the average water yield (186 mm) corresponds to 27% of the total rainfall (686 mm), and average annual sediment load was estimated to be between 6 and 6.8 t ha−1 y−1. A gradient of sediment yield was found among the subbasins that is characterized by a large difference among the upper (7 to 13 t ha−1 y−1), central, and lower parts (<1 t ha−1 y−1) of the study area. The results show that in the Carapelle the soil erosion is irreversible as soil losses are higher than the soil formation rate and that winter wheat and olive land use are the major source areas, in terms of sediment. The average gross erosion estimated by the RUSLE2015 model (12.5 t ha−1 y−1) resulted comparable with the average specific sediment yield estimated by SWAT (8.8 t ha−1 y−1) and AnnAGNPS (5.6 t ha−1 y−1), while it was found that the average soil erosion estimated by PESERA is lower than the other estimates (1.2 t ha−1 y−1). To reduce economic and environmental impacts, specific measures are needed, especially in the upper part of the basin. Indeed, European policies (PAC 2014-2020) aims at reducing soil erosion through the application of BMPs. Four different scenarios were developed and simulated in order to compare the effect in terms of reduction of soil erosion. A threshold of sediment yield greater than 10 t ha-1y-1 was selected to discretize the target areas in which apply the BMPs. Results resulting from the analysis of the four BMPs scenarios simulated showed that the combination between “contour lines ploughing” and “reforestation” was the most effective with a reduction of the total sediment load of 37% (from 6.0 t ha-1 to 3.7 t ha-1). No tillage gave 30% of reduction (from 6.0 t ha-1 to 4.2 t ha-1) and contour line 22% (from 6.0 t ha-1 to 4.6 t ha-1). Finally, reforestation showed a reduction of 14% (from 6.0 t ha-1 to 5.0 t ha-1). The analysis of the farmer return-production cost ratio (FR/PC) highlighted that winter wheat production managed with conventional tillage (Baseline) as well as with “contour farming” (BMP1) are economically sustainable in non-steep slope areas (FR/PC=1.12 and 1.11, respectively) while in steep slope areas the Baseline scenario is not economically advantageous (FR/PC = 0.93). In non-steep slope areas “no tillage” (BMP2) is the most economically advantageous (FR/PC=1.67), while it is the second for steep slope areas (FR/PC=1.41) where the scenarios “reforestation” (BMP3) resulted to be the best (FR/C=1.49). Finally, to analyze the effectiveness of check dams in reducing channel erosion, field measurements were carried out in fifty-three transects of the Cammarota stream, located in the upper Carapelle watershed. The Leaf Area Index (LAI) was measured using a Plant Canopy Analyzer and subsequently retrieved from two Landsat 8 satellite images. The performance of three equations for deriving LAI values from the corrected Normalized Difference Vegetation Index (NDVI) was evaluated. The observed LAI values of riparian vegetation were found lower in correspondence of damaged or destroyed check dams while they were higher in the reaches with greater geomorphological stability. Results obtained from the analysis of the Landsat 8 satellite images highlighted that The Lambert-Beer equation gave the best statistical performance for predicting LAI within the reaches with better geomorphological stability. The results of this research can give useful indications for water resource managers to select the most appropriate model for a specific area based on the study aims and input data availability. The approach of this work can help modelers in calibrating and validating hydrological models in areas prone to soil erosion, especially in basins with temporary river systems. In addition, the results are very useful for selecting and prioritizing conservation interventions (BMPs) aimed at reducing soil erosion in sediment source areas. Among the conservation measures, check dams show their effectiveness in reducing channel erosion, increasing the geomorphological stability and, consequently, improving the conditions of the riparian ecosystem. N.B. The language of the thesis is English, as authorized by the PhD Committee and has been drawn up according to the specifications, reported in the Art.12 of the regulations of the PhD program in "Biodiversity, Agriculture and the Environment", approved by the PhD Committee on 06/02/2014. The thesis is composed of a general introduction, a general conclusion and four chapters reporting four scientific papers produced by the candidate, three published and one submitted in ISI/SCOPUS indexed scientific journals.