Relationships between rainfall characteristics and ephemeral gully erosion in a cultivated catchment in Sicily (Italy)

Recent research has shown a lack of long-term monitoring for detailed analysis of gully erosion response to climate characteristics. Measures carried out from 1995 to 2007 in a wheat-cultivated area in Raddusa (Sicily, Italy), represent one of the longest series of field data on ephemeral gully, EG, erosion. The data set collected in a surface area of almost 80 ha, permits analysis of the influence of rainfall on EG formation and development. Ephemeral gullies formed in the study area were measured on a yearly scale with a Post-Processing Differential GPS for length and with a steel tape for the width and depth of transversal sections. Ephemeral gully formation was observed for 8 years out of 12, which corresponds to a return period of 1.5 years. The measurements show strong temporal variability in EG erosion, in agreement with the rainfall characteristics. The total eroded volumes ranged between 0 and ca. 800 m3 year−1, with a mean of ca. 420 m3 year−1, corresponding to ca. 0.6 kg m−2 year−1. Ephemeral gully erosion in the study area is directly and mainly controlled by rainfall events. An antecedent rainfall index, the maximum value of 3-days rainfall (Hmax3_d), is the rain parameter which best accounts for EG erosion. This index is used here as a simple surrogate for soil water content. An Hmax3_d threshold of 51 mm was observed for EG formation. The return period of the Hmax3_d threshold is almost the same as the return period for EG formation. Although a mean of seven erosive rain events were recorded in a year, EG formation and development generally occur during a single erosive event, similarly to other semiarid environments. The most critical period is that comprised between October and January, when the soil is wetter and the vegetation cover is scarce. Empirical models for EG eroded volume estimation were obtained using the data set collected at this site. A simple power-type equation is proposed to estimate the eroded volumes using Hmax3_d as an independent variable. This equation shows an R2 equal to 0.67 and a standard error of estimation of 0.79.