Assessing different survey and gridding techniques for Digital Elevation Models generation and the related influence on stony debris flows modelling. A case study from Cancia basin (Venetian Dolomites, North-Eastern Italian Alps)

In the Dolomites area (North-Eastern Italian Alps), debris flows can be regarded as one of the most hazardous geomorphological processes. In the last few years, these natural phenomena amplified their occurrence rate due to the rise of extreme rainfall events, and the increasing availability of debris material yielded by the retreat of the glaciers and the permafrost areas to higher elevations. In order to cope with debris flow hazard, it is common to couple structural and non-structural measurements, such as the zoning of risk prone areas by means of routing models. Since the motion of gravity-driven flows is extremely sensitive to surface morphology, topographic data in the form of Digital Elevation Models (DEMs) represent the most important input in debris flow routing models. As a matter of fact, a DEM can be regarded as a mathematical representation of the bare earth in digital form, and it is commonly used to represent the surface morphology in three dimensions. The “quality” of DEMs depends on the accuracy, density, and spatial distribution of the topographic data (i.e., on the employed survey technique); on the characteristics of the surveyed surface; and on the applied gridding methodology. Therefore, the choice of both the survey technique and the gridding methodology might represent a critical concern for the reliability of routing modeling outcomes. In order to advance in the knowledge regarding the influence of geomatic techniques on the numerical modeling of stony debris flows routing, in the present research we initially assessed the performances of common digital terrain modelling algorithms (i.e., linear triangulation, natural neighbor, nearest neighbor, inverse distance to a power, ANUDEM, completely regularize spline function, thin-plate spline function, thin-plate spline plus tension function, multi-quadratic function, inverse multi-quadratic function, point ordinary kriging, and block ordinary kriging) and survey techniques (i.e., full-waveform Light Detection And Ranging, LiDAR; and Global Navigation Satellite System, GNSS) in characterizing the complex topography of a debris flow channel located in the Venetian Dolomites. After that, their inherent influence on the results of a Geographic Information System (GIS)-based cell model for simulating stony debris flows routing is investigated through a combination of statistical and visual techniques, by considering both high- and low-magnitude flow conditions. On one and, the research points out that the linear triangulation, the natural neighbor algorithm, and the thin-plate spline plus tension and completely regularized spline basis functions could represent the best choice for applications relying on the proper representation of the surface shape (e.g., hydraulic and hydrological modeling). In fact, these gridding algorithms proved to ensure the best trade-off between interpolation accuracy and shape reliability. However, the research also shows that the choice of the gridding methodology actually does not represent a determining factor in debris flows routing modeling. On the other hand, the analysis carried out on the capability of the two tested survey techniques in characterizing the topography of the studied debris flow channel, highlights a high degree of interoperability, since both of them could be used to generate bias-free and accurate high-resolution DEMs of morphologically complex areas. However, the pairwise comparison of the GNSS- and LiDAR-derived DEMs reveals that, although the two investigated survey techniques provide a comparable (i.e., not statistically different) topographic characterization of the channel bathymetry, meaningful vertical discrepancies could be detected in correspondence of morphologically complex channel features (e.g., channel banks and longitudinal/transversal slope discontinuities). Furthermore, the detected discrepancies proved to be able to affect the cell routing model behavior, thus leading to the conclusion that the choice of the survey technique could represent a critical concern for the reliability of routing modeling outcomes.