Predicting rill erosion due to extreme rainfall events on a small hillslope in Spain using the Landscape Evolution Model (LEM) 

Abstract

<p>Landscape evolution models (LEMs) simulate geomorphic changes in landscapes and, among them, modern models can provide a quantitative estimation of sediment fluxes and geomorphological processes under changing land use and climate at different spatial and temporal scales. Although much progress has been made in the development of LEMs, few works have tested their prediction capability for single extreme rainfall events on a slope with well-defined topographic and geomorphological initial parameters, leaving for uncertain if LEMs are useful and accurate enough for this purpose. In this study, CAESAR-Lisflood LEM was used to predict a geomorphological scenario following a single rainfall event that caused rill erosion on an almond orchard in a sloping area. The high resolution aerial imagery, acquired by an unmanned aerial vehicle (UAV) after the rainfall event, allowed generating a digital surface model (DSM) with a high spatial resolution (i.e. 1.5 cm) by applying the structure-from motion (SfM) image technique. To reconstruct the topography prior to the rainfall event useful for the application of the Caesar-Lisflood LEM, we generated a pre-rill landscape using an authomatic method for rill extraction and characterization. This method provided an objective rill description with a cm-accuracy and-resolution. To evaluate the accuracy of the LEM model, the characteristics of some rills produced by this model were compared with a well-known field measurement method (i.e. ISUM - Improved Stock Unearthing Method). Our results show that LEMs allow to make reliable predictions even of short-term scenarios of rill development, thus confirming the high potential of such an approach in reconstructing geomorphological evolution and lanscapes dynamics.</p>