Integrated techniques for slope erosion modelling and badland monitoring in key sites of Central Italy

The growing interest in studying badland dynamics reflects the need to increase knowledge of geomorphologic processes and dynamics in subhumid badland areas, particularly because of their importance in generating extremes of water and sediment production. Field studies of soil erosion are expensive, time-consuming and data needs to be collected over many years. Though providing detailed understanding of the erosion processes, field studies have limitations because of the complexity of interactions and the difficulty of generalising from the results. Cost-efficient methods of estimating erosion over whole catchments are required as ways of predicting erosion after disturbance or following various erosion management strategies. Thus, the indirect estimation and the prevision of erosion rates is still one of the main research topics of the scientific community in the field of geomorphology and is far from solved. This Ph.D. research project is aimed at defining an integrated methodology of denudation intensity estimation and prevision, for areas greatly affected by badlands, and it is based on both quantitative geomorphic analysis and multivariate statistical investigations, in order to deepen the relationships between the main denudation effects and the potential causal factors favoring geomorphologic instability in badlands areas. The research have allowed to propose a statically based method for water erosion hazard assessment, conceived as a spatially distributed prevision of calanchi badlands, and associated erosion rate, occurrence. Direct measures of erosion intensity in badlands were used to validate the water erosion estimates and previsions. As the research project is focused mainly on methodological objectives, well-known study areas of Tyrrhenian side of central Italy, included within the Tevere and Ombrone River Basins, have been selected, in order to compare the results gradually achieved with the earlier available data. Erosion rate estimations were performed refining some empirical equations (“Tu Denudation index”), that estimate the suspended sediment yield (SSY) as a function of morphometric parameters related to drainage network and relief (Ciccacci et al., 1981, 1986). Tu denudation index was confirmed to be a good estimator of the suspended sediment yield (SSY) for catchments characterized by the prevalence of sedimentary and weakly coherent outcropping lithologies. The improvement of the regression relations contributed to better estimate sedimentary output for catchment widely affected by badland areas. In these basins, in fact, SSY is strongly correlated to the areal ratio affected by badlands to the total catchment area. Thus, using the not-projected drainage density parameter (D3d), instead of the traditional drainage density parameter (D), even if not improving the SSY estimation for large basins, was considered to better reflect the conditions predisposing erosions than D for smaller catchments, where large calanchi badlands and related high slope gradients are present. The dominant role of drainage density in estimating erosion rate for badland areas was confirmed by the attempt of zoning the estimated erosion rates using the Tu Grid Analysis. This attempt proved to be very efficient in estimating the erosion rate due to runoff within badland areas, as confirmed by the comparison between the estimated and the measured erosion rates. This result seems to increase the prospective of using Tu grid analysis when prolonged denudation monitoring is not possible. Moreover, even where punctual erosion rates are measured by pin monitoring, the estimated erosion rate map represents a validated continuous representation of water erosion rate for larger areas. Geomorphological susceptibility evaluation was performed applying a multivariate statistical method based on conditional analysis (Bayesian interpretation of probability) integrated by a proposal of a new method for most influential causal factors selection. The procedure provided satisfactory results for the unbiased prediction of landslide and water erosion susceptibility for the Upper Orcia Valley. The method is conceptually simple but, at the same time, effective in evaluating the conditional probability of hazardous events given a certain combination of causal factors: the proposed factor selection procedure has proved to be a useful tool for the unbiased detection of the factors really discriminant for instability landforms in the study area, and can be very helpful when analyzing new areas. Moreover the use of vector datasets allow to create vector easy-to-read susceptibility maps, in which the fragmentation generally characterizing raster outputs is avoided. These characteristics make this susceptibility method easy to be understood and each resulting map easy to be read, thus suitable for policy makers in planning land management strategies. The association of the estimated erosion rate for calanchi badland areas to the surveyed landforms allowed to use the susceptibility method to evaluate the water erosion hazard, since the temporal information about the erosion processes was related to the spatial data. This procedure is proposed when direct erosion rate measures are not available. Different techniques of direct monitoring of erosion rates and processes have been performed, with the aim of identifying the main geomorphic processes acting in the study areas and quantifying their intensity. In particular, direct measurements (erosion pin monitoring, geomorphologic survey, DGPS survey and digital photogrammetric analysis) were used to validate the results obtained after indirect erosion rate estimations and susceptibility and hazard assessment models application. Even though, interesting remarks have been concluded on the applicability of various methods of erosion monitoring. The size of the study area, the time available, and the quality of the data required are perhaps the most critical issues to be considered when looking for the most appropriate technique. As well-known, the traditional erosion pin method generally allows to carry out very accurate punctual measures, whose error is measurable in few millimeters. So, it can be used to quantify very detailed temporal variations (monthly or after-event ground level changes). On the other hand, DGPS survey can be proper when a single hillslope of less than few hectares is being monitored, as the time and effort required would be acceptable. For larger areas or wider time interval, high resolution photogrammetric analysis could be more appropriate. However, all these methods are affected by many error sources, that limit their use to very specific time and spatial ranges. Finally, some new contributions to the knowledge of the physical factors influencing the initiation and the development of different water erosion landforms in the studied badland areas have been achieved. Comparison of pluviometric data and measured ground level variations for Bargiano site (Tevere River Basin) has highlighted that clay removal by water erosion is generally due to intense rainfall event preceded by quite long dry periods, while accumulation (due to gully banks collapsing) is favoured by intense rainfall after a certain number of rainy days (frequent in spring). Moreover, in inter-rill position, where almost the lonely water erosion acts, intense events are significantly more effective than long events. Considering the distribution of calanchi and biancane landforms of Upper Orcia Valley (Ombrone River Basin) among the different classes of the main topographic and physiographic factors, it is a matter of fact that calanchi badlands develop on steeper slopes and where higher values of amplitude of relief occur, due to the morphoevolutionary processes. Moreover, observations on present embryonic biancane of Lucciolabella site confirm the leading role played by reticular systems of joints in the dissection of original, gently-dipping surfaces. Actually, a resolute difference on dispersivity level of the biancana parent material samples of La Piaggia subcatchment was not found with respect to calanchi badlands samples of the same subcatchment. On the other hand, a significant influence of clay properties was observed on the different erosion rates measured during decadal monitoring investigations by means of erosion pins in the study areas. Calanchi badlands show lower erosion rates due to surface runoff. The major facility of biancane clays to be entrained at very low stream powers is reflected in their major dispersivity, while, in badlands, the morphoevolution and sediment removal is predominantly caused by widespread mudsliding from the rill and gully heads, as also confirmed by the mean positive variations of ground level recorded at some calanchi monitoring stations. This observation can be also related to the higher sand content in calanchi badlands, which may favour the infiltration processes to the detriment of runoff. Finally, as already observed by several authors, the agricultural exploitation of these lands lead to a decrease of exchangeable cations concentration (and, thus, clay dispersivity), even if the permanent inhibition of chemical dispersion due to increase of soil stability hypothesized by Phillips (1998) cannot be completely agreed. Decadal monitoring and observation in the study areas and in other sites of central Italy have outlined that badlands initiation is even enhanced by agricultural manipulation: grazing and farming are among the most important triggers for accelerated water erosion, and tillage erosion has been recognized as an increasing factor of water erosion. The performed investigations have allowed to carry out some new remarks about both the applied and proposed methodologies and the studied areas and related processes. In particular, results from this research have contributed to improve some methods useful to deepen the knowledge of processes and denudation intensity acting in badland areas of Mediterranean drainage basins.