Terrace Smoothing Using Open Source Tools In Digital Terrain Models

Extreme rainfall events have been occurring more frequently due to climate change causing several distress especially in the most vulnerable Italian territory. The region of Liguria is particularly prone to hazardous events due to the morphology of the terrain. Moreover, anthropogenic modifications of the land such as terraces are among the factors that impact the risk of a natural hazard after an extreme rainfall event. A particular extreme event occurred in autumn 2014, where rainfalls affected different areas in Liguria causing three casualties, severe structural damages and more than 1600 landslides. In this context, innovative spatial analysis procedures on land topography can increase the knowledge of morphological parameters, like the well-known connectivity index. The aim of this research is to develop a method to recover the land topography before the modification caused by terraces and therefore understand their impact from comparison. The selected area for this analysis is the catchment of Rupinaro River, Italy, which has been surveyed after the 2014 event by CNR IRPI. An Airborne Light Detection and Ranging survey was carried out and Digital Terrain Models (DTM) of 50 cm resolution were produced. To this topic, our contribution is a terrace smoothing methodology, which creates a DTM of Rupinaro area that is terrace free but conserves the original elevation values. That way, a terraced model and a nonterraced model can be compared to understand the impact of anthropogenic modifications on natural hazards. An iterative approach was used for the entire procedure including the creation of smoothed DTM. A combination of QGIS and Python were used for the entire terrace smoothing procedure. The terrace smoothing steps included visualizing the terraces and walls, creating a mask to sperate the terraces and the terrace walls, extracting centerlines of the terraces and walls, converting the centerlines into points, merging these centerline points, interpolating using triangulated irregular network, and smoothing the produced DTM. After a comparison of the original data and the produced data, we successfully recreated the original morphology of the area in a state without terraces. Consequently, this artificially smoothed model can be further studied to understand the impact of terraces on natural hazards such as landslides in the catchment of Rupinaro Italy.