Soil moisture distribution and runoff response at the hillslope scale: experimental analysis in an alpine environment

This work focuses on the analysis of hydrological data collected at the hillslope scale during three summer field campaigns carried out in a small mountain catchment (1.9 km²) in the Dolomites (central-eastern Italian Alps). The thesis is partitioned in three main sections regarding i) the analysis of spatial distribution of water content over three soil depths, ii) the assessment of its temporal stability, iii) the study of runoff and water table variations. Soil moisture is one of the most important hydrological variables: it has a critical influences on several processes at different spatial scales, plays an important role in hydrological modelling and flood forecasting and represents one of the main factors in infiltration of water, surface and subsurface runoff generation. Soil moisture data were collected during summers season in 2005, 2006 and 2007 at 0-6, 0-12 and 0-20 cm depth over three hillslopes with steep relief and shallow soil depth. Volumetric water content values were collected at several points over the three hillslopes by means of an impedance and a TDR probe. Soil moisture data were used to analyse the statistical moments and their interrelationships, the relationships between data collected at various soil depths, and time stability. Results showed that the surface layer was usually wetter than deeper soil layers, particularly during dry-down. This was attributed to the effect of dew, which was observed in the field and might have contributed to the increase in the surface soil moisture. For all depths and over the three hillslopes, the spatial variability patterns were well represented by negative exponential functions between the mean and the coefficient of variation of soil moisture. Vertical water content variability was mainly attributed to the increase of soil properties heterogeneity with depth. The degree of correlation between the data collected at the three depths was relatively high, as also confirmed by the visual comparison of interpolated water content maps. Temporal stability of soil moisture patterns was investigated applying a multiple approach: i) ranking stability analysis; ii) slope-intercept analysis of linear regression; iii) autocorrelation analysis; (iv) evolution of correlation against mean soil moisture over time and relationship with piezometric increase. Results show that spatial patterns of sampling points were reasonably well preserved at the three depths. The highest degree of temporal stability was associated to wet conditions and the decline of correlation occurred at times of transition from a dry to a wetter state; such behaviour is more evident for the surface layers, more quickly affected by precipitation inputs. Less correlated patterns were found in 2006 owing to a different distribution of precipitation: this observation was confirmed by less sloping correlograms in 2005 and 2007. Temporal stability of surface measurements could be considered as good indicator of subsurface time stability; identification of these temporally stable sites within the experimental basin will assist to provide data sets for watershed hydrologic modelling of subsurface soil water content. Overland flow is recognized as an important contributor to the total stream discharge and to the determination of the size and the shape of flood peaks; nevertheless, it has hardly ever been observed through direct measurements. In this study, occurrence of overland flow was measured by special detectors installed over a small subcatchment (3.3 ha) of the main basin and stream discharge, soil moisture at 0-30 cm depth (by means of a water content reflectometer) and groundwater variations were monitored continuously. Generation of overland flow was identified in mechanisms of saturation from above and in return flows of water along preferential flow paths. At the rainfall scale, a clear interconnection among several hydrological variables was found, suggesting a strong consistency between subsurface and surface runoff, which respond with similar spatial and temporal dynamics to precipitation inputs both at the rainstorm scale and over a longer period. Water table was found to be highly correlated with stream discharge independently of the topographical localization of piezometric wells across the site. The steady state assumption for the whole hillslope was tested by applying a reformulation of Topmodel at the single rainfall scale. The linear model proved to be a fine predictor of the groundwater level with small overall errors although often the maximum rise of water table was underestimated, especially for significant variations. These deviations were attributed to the hysteresis effect of the relationship between discharge and groundwater, whose non linear behaviour prevented the linear model from accurate predictions of groundwater levels during noticeable fluctuations.