Identification of large geomorphological structures based on 2D discrete wavelet transform

The areas located at the front of the chains are characterized by severe deformations related to the pushing and underthrust of allochthonous nappes induced by tectonic actions. This deformation may create large geomorphological structure like giant landslides. The problem of identification and geological interpretation these large geomorphological structures constitutes an important issue, since it can be difficult to identify and to map these structure. . A solution to this problem can be the analysis of the Digital Elevation Model (DEM) of the area, based on a geomorphic numerical approach. This can provide evidences of large anomalies of the surface topography, and of peculiar landforms. In particular, here an approach based on 2D discrete wavelet transform (Antoine et al., 2003; Bruun & Nilsen, 2003, Booth et al., 2009; Doglioni & Simeone, 2011) is attempted. An analysis of the peculiar topography of the region of the front of the chain at the level of river Biferno and river Fortore valley is developed, see Figure 1. This analysis is performed looking for possible anomalies of the DEM as well as at the position of the front of the Apennine. A medium resolution DEM, e.g. 30 m grid-based ASTER GDEM, is here used. The anomalies and discontinuities of the surface topography are expected to entail the tectonics of the chain and in the foredeep area they may be consistent with large gravitational collapse and landslide (Guerricchio et al., 2010; Galeandro et al., 2012). The 2D wavelet decomposition of the DEM, and in particular the analysis of the detail coefficients of the wavelet transform can provide evidences of anomalies or singularities, i.e. discontinuities of the land surface. The hierarchical representation of the DEM, can provide evidences of anomalies or singularities of the land surface, which are not directly evident from the DEM itself. In particular, 2D wavelet transform preserves the average values of the elevation at different scales, and this is particularly suitable for grid-based DEM. The grid-based DEM can be assumed as a matrix, where a discrete wavelet transform (Daubechies, 1992) is performed columnwise and linewise, which basically represent North-South and East-West directions. The significant outcomes of this analysis are low-frequency approximation coefficients and high-frequency detail coefficients. The detail coefficients in particular are analyzed, whereas sudden and wide variations of these coefficients are related to the variations and discontinuities of the DEM. Detailed coefficients are therefore mapped, thus allowing to visualize and quantify potential anomalies of the land surface. In this kind of approach, the choice of a proper wavelet is a tricky point, since it conditions the analysis and then their outcomes. Therefore multiple levels as well as multiple wavelet analyses are guessed. Here the introduced approach is applied to two interesting cases study of south Italy, located at the front of the Apennine. For both the analysed scenarios, the 2D discrete wavelet transform allows to delimitate two large anomalies, the earlier in low Biferno valley and the latter in Fortore valley, see Figure 2. These barely corresponds to the assumed landslide bodies. In particular, landslide bodies correspond to areas, which do not show discontinuities of the topography. On the other hands, the 2D discrete wavelet analysis evidenced strong discontinuities of the topography where the scarps of the landslides are assumed or located, and also where the front of the Apennine is expected. The latter result shows that landslides and the buried front of the Apennine are very close, thus somehow supporting the assumption made on the interaction between the tectonic thrust and the landslides. Moreover, looking at other well-known landslides on the Adriatic coast, they are also close to the front of the Apennine, delineated by the detail coefficients of the 2D discrete wavelet transform.