Application of the EGEM model to predict ephemeral gully erosion in Sicily, Italy

Recent research has shown the importance of ephemeral gully (EG) erosion in the context of
global erosion and, at the same time, the lack of adequate models to estimate it. At present, the
ephemeral gully erosion model (EGEM) is the only conceptual model specifically developed for
ephemeral gully erosion estimation. The main aim of the paper is to evaluate the performance of
EGEM model adapted to local conditions compared with that of simple empirical equations. A data
set of 92 ephemeral gullies formed between 1995 and 2000 in a wheat-cultivated study area in Sicily,
Italy, was used for the purpose. In order to improve EGEM performance, two important adaptations
were made. Both the rain distribution type and the causative rainfall of the hydrological component
of EGEM were modified to reproduce local conditions. The EGEM assumption of a constant depth
for the entire length of the ephemeral gully was also modified. EGEM performance was tested with
two different rain distribution types, three different causative rainfalls and two different EG depths.
The adaptation of the EGEM hydrological component improves the volume prediction with respect
to the standard version of EGEM; the improvement is mainly due to the rain distribution type,
whereas causative rainfall seems to be less important. The use of adjusted EG depths is, however,
essential to minimise the errors between measured and predicted volumes. The capability of EGEM
to predict the ephemeral gully cross-section is weak and non-significant relationships were found
between measured and estimated width. A regression analysis between ephemeral gully-eroded
volume and different morphological, topographical and hydrological parameters confirms the great
importance of EG length to explain eroded volume (R2=0.64 for the simple correlation between volume and length). Considering that the adaptation of the hydrological component and the
measurement of EG depths are necessary for the EG volumes estimated by EGEM to be reliable, it
seems more simple, as suggested by other authors, to use empirical relations between eroded volume
and length performed in different environments, until more precise physically based models are
developed.