A prototype, high-resolution multi-temporal landslide inventory for the Sillaro River basin, Northern Apennines

In mountain environments, landslides are dominant geomorphic processes of sediment transfer and as such, they play a fundamental role in landscape evolution and sediment management at the watershed scale. While monitoring of landslide dynamics at the scale of the single slope failure provides precious site-specific information, an appraisal of landslide-driven sediment dynamics at more representative spatial scales is rarely pursued. In this context, the compilation of multi-temporal, high-resolution landslide inventory represents a challenging but critical task.

In the Emilia-Romagna region (Italy), landslides cover about the 24% of the hilly and mountain areas within the Northern Apennines. Here, the most common types of landslides are earth slides and earthflows that mainly involve the terrain of clay Ligurian Units and usually are the re-activations of preexisting mass movements. Since the mid ’80, the Geological Survey of Emilia-Romagna Region (RER) has started compiling and updating a region-wide landslide inventory, which includes all movement types, as well as both active (n = 44,377) and dormant (n = 36,608) landforms. The inventory update is customarily performed in selected areas, mainly where landslides have created damages or pose risk to infrastructures, or where ad hoc land planning is needed. In this context, a systematic multi-temporal approach that could provide robust information on landslide occurrence and recurrence is missing.

To address this gap, in this contribution we propose a multi-temporal inventory prototype, which includes a set of attributes aimed at characterizing landslide sediment transfer across decades. The prototype is developed in the mountain portion of the Sillaro River basin (139 km²). The basin is chiefly underlain by argillites of the Ligurian domain, where earth slide and earthflow activity is pervasive.

The compilation of the multi-temporal landslide inventory is conducted through visual inspection of 10 sequential aerial photo sets (1954, 1969, 1976, 1985-88, 1996, 2000, 2006, 2008, 2011, and 2014), as well as Google Earth satellite imagery (2016 and 2018). In particular, each landslide polygon encloses the total disturbed area, which includes initiation, transport and deposition zones. Polygon planimetric changes are then recorded across sequential photosets. In this way, it is possible to record recurring landslide movements.

Landslide planimetric geometry includes length, width, and area. Landslide attributes include movement type, photo year of occurrence, morphologic position at initiation (source), and sediment delivery target (sink). Subsequently, for each landslide we subdivide total disturbed area into initiation-transport and deposition polygons. For recurring landslides, we note whether the movement involved: (i) the whole landslide body or only part of it; (ii) headscarp migration; and (iii) advance of the landslide terminus. Finally, we note whether the landslide deposition zone displayed headward incision by means of gully development and/or revegetation.

This work, as part of the projects BEDFLOW and BEFLOW PLUS, is partially funded by Fondazione Cassa di Risparmio in Bologna.