Secondo workshop congiunto Italia-Brasile sulla modellazione di frane
CNR IRPI Perugia, 14 maggio 2026
Contenuti: Verranno descritti e discussi i sistemi di monitoraggio in tempo reale delle frane, sviluppati in Italia dal CNR IRPI e in Brasile dall’UFRJ e dall’UFF. Questi sistemi, progettati per monitorare le frane indotte dalle precipitazioni, sono operativi e in grado di produrre bollettini giornalieri o sub-giornalieri contenenti informazioni sulla probabilità di occorrenza di frane in tutto il paese. L’obiettivo di questo workshop è discutere l’utilizzo di modelli fisici per il monitoraggio/allerta precoce e argomenti correlati. Il workshop è strutturato in brevi presentazioni, con l’intento di favorire la discussione e di esplorare possibili ulteriori collaborazioni.
- The Institute for Geo-Hydrological Research (IRPI) of the National Research Council (CNR)
- Universidade Federal Fluminense (UFF)
- Universidade Federal do Rio de Janeiro (UFRJ)
- Universidade Federal do Rio Grande do Sul (UFRGS)
- São Paulo State University (UNESP)
Contributions (Speaker, affiliation, title, short abstract)
Italian side:
I1 – Mauro ROSSI (CNR IRPI, Perugia): “Landslide early warning: lessons learned after 10-year experience in Italy”.
Description of our monitoring system for rainfall-induced landslides, which we have been running for more than ten years, its functionalities and the different customizations of the system we prepared for different users/stakeholders (Department of civil protection and the national railway company).
I2 – Stefano Luigi GARIANO (CNR IRPI, Perugia): “Data, methods, and tools for to evaluate the triggering conditions of rainfall-induced landslides”.
CNR IRPI has developed catalogues of information of rainfall-induced landslides in Italy, recently leveraging on AI. Also, methods and tools for reconstructing triggering rainfall events and calculating rainfall thresholds are developed. Thresholds can be used to predict the occurrence of landslides in different environments and to analyze variations in landslide-triggering conditions in a changing climate.
I3 – Nunzia MONTE et al. (CNR IRPI, Perugia): “A scalable GIS–database workflow for processing EGMS InSAR time series for landslide susceptibility studies”. (Poster presented at EGU 2026)
In this contribution, we present a scalable workflow for transforming EGMS data into analysis-ready inputs for dynamic landslide susceptibility studies. The methodology was developed using GRASS GIS and PostgreSQL/PostGIS, exploiting a high-performance multicore computing infrastructure. The workflow was first tested on a reduced pilot area and subsequently extended to the entire Province of Salerno (southern Italy), a region characterized by complex geomorphology and widespread slope instability.
I4 – Massimiliano ALVIOLI & Massimo MELILLO (CNR IRPI, Perugia): “Rockfall modeling in GRASS GIS with r.stone: coupling seismic and rainfall triggers with a physical model”.
The module r.stone is an implementation of the STONE model, originally developed by Guzzetti et al (2002). Here, we discuss novel, simple strategies to couple seismic and rainfall triggers with physically-based modeling with r.stone in GRASS GIS.
Brazilian side:
B1 – Gean Paulo MICHEL (UFF), Franciele ZANANDREA (UFF) & Nelson FERREIRA FERNANDEZ (UFRJ): Landslides in Brazil: Ongoing projects and possible future collaborations
Overview of the research topics developed in the research groups (GRAPHID and LAMPEGE) at the universities UFF and UFRJ. We also present a national-level factors risk assessment integrating hydroclimatic tendencies, landslides susceptibility patterns, and population distribution and expansion in Brazil.
B2 – Nestor A. BRESOLIN Jr (UFRGS): Spatio-temporal rainfall controls on landslide triggering: Can space be used in place of time?
Development of empirical rainfall thresholds based on a single extreme event, in which spatial variability replaces the need for long-term historical rainfall and landslide records. Considering the large affected area and the high spatial heterogeneity of rainfall during the 2024 southern Brazil disaster, each mapped landslide was treated as an individual triggering event. Sub-hourly landslide timing was reconstructed through field interviews with directly affected residents.
B3 – Clara MOREIRA CARDOSO (UFF, Brazil): Semi-automated landslide database development through online news and satellite images
A global news database (GDELT) to collect news available on the web in order to compile a database of the potential dates of landslides events in the city of Petrópolis. To improve the spatial accuracy of the database, the Monte Carlo method was applied to a pre- and post-event comparison of satellite images applying flexible thresholds of slope and vegetation index, resulting in probability maps of landslide scars.
B4 – Artur N. V. CERETO (UFF, Brazil): Towards the development of a machine learning based LEWS for data-scarce environments
This study, in a pilot Brazilian municipality, evaluates the feasibility of data-driven approaches for the prediction component of landslide early warning systems (LEWS). Using rainfall data as the main input, we investigate the use of Civil Defense records related to landslides as a proxy for conventional landslide inventories, aiming to support the rapid implementation of LEWS in vulnerable areas.
B5 – Danúbia Teixeira SILVA (UFRJ):A three-steps approach for modelling rainfall-triggered landslides using TRIGRS (work performed during DTS visit at CNR IRPI, January-June, 2026)
We integrated measurement data with TRIGRS numerical simulations to evaluate the model’s ability to reproduce the observed hydrological behavior and calibrated a range of geotechnical parameters, identifying the best correspondence between the simulated Factor of Safety (FS) and the observed landslide scars in the field.
B6 – Daniel METODIEV (UNESP): Correlation study between precipitation and shallow landslides to propose critical operational thresholds in Baixada Santista region/Sao Paulo state, Brazil
A methodology that correlates triggering and antecedent rainfall indices (effective half-life), soil moisture data, and landslide occurrences, to obtain rainfall thresholds for shallow landslides. Thresholds based on effective half-live rainfall indices outperform those based on simple rainfall indices. Improved representation of soil water content into the ground is useful for reliable implementation in LEWS.
Full list of participants on the Brazilian side: Gean Paulo Michel, Franciele Zanandrea, Nelson Ferreira Fernandes, Danubia Teixeira Silva, Nestor A. Bresolin Jr., Artur N. V. Cereto, Clara Moreira Cardoso, Daniel Metodiev