Water erosion modeling using fractal rainfall disaggregation- a study in semiarid Tunisia

Sammanfattning: In the Tunisian semiarid area, water erosion processes have led to negative economic and environmental consequences in a context of limited water resources. To characterize and predict these degradation phenomena, a comprehensive high-resolution data base on erosive rainfall, together with siltation records for 28 small reservoirs were analyzed. The studied small reservoir network displays a general life-span of about 14 years. The average soil loss is 14.5 tonnes/ha/year. The complex relationship between the erosive rainfall events and the annual soil loss rates can be explained by two important factors. The first factor is related to the soil “degradation cycle”. It determines the soil particle delivery potential of the catchment. The second factor corresponds to the “degradation front”. The latter presents a north-western/south-eastern direction. To investigate rainfall disaggregation possibilities, a regionalization of fine time-scale and daily rainfall was undertaken. The results showed that the spatial properties are typically non-isotropic. Clustering showed that two different homogeneous rainfall subgroups are closely related to the predominant convective and frontal rainfall types. The comparison illustrated important similarities between the maximum 15-min and the daily rainfall data. The above findings were an incentive to explore the multiplicative properties of a 4-year rainfall time series. The time series showed scaling behavior for time scales up to 100 min that coincides with the most active erosion process time scale. Moreover, the temporal structure of rainfall was reproduced using a disaggregation model (Olsson, 1998). The observed and generated rainfall time series displayed several similar characteristics. This allowed the reproduction of erosivity for erosive rainfall events longer than 45 minutes. The erosion modeling was performed using the USLE/GIS approach. Maps of observed and generated spatial erosion distribution were combined with the Masson’s and Wischmeier-Smith’s erosion limit intervals. According to the above, the potential of rainfall scaling-based approaches to predict water erosion levels in semiarid areas seems promising. Using this approach may help soil and water authorities in semiarid areas to better manage soil erosion problems.