TWINFALL at M4E 2026 and AGU26: from rock fracture modelling to slope monitoring
TWINFALL research will be presented at two international meetings in the coming weeks. The contributions cover numerical modelling of rock slope failure, environmental and seismic monitoring, and simulation of how rock slopes respond to seismic waves.
Mathematics for Planet Earth, Milan, 9–11 November 2026
The second Mathematics for Planet Earth workshop (M4E) takes place at Politecnico di Milano. Denise Grappein will present “Numerical modeling of rock slope instability using a phase-field Material Point Method”, written with Carlo De Falco, Luca Formaggia, Alessio Fumagalli and Anna Scotti of the MOX Laboratory on November 10.
The work simulates how slope failure is triggered, focusing on the fracture propagation that leads to rockfalls. The phase-field method describes fracture through a continuous damage variable, so cracks do not need to be tracked explicitly. The Material Point Method suits the complex geometry and heterogeneous materials of natural slopes. It also handles large deformations, so the same framework can be used to simulate the runout that follows failure. The approach will be illustrated on real case studies in the municipality of Lecco.
AGU26, San Francisco, 7–11 December 2026
Two TWINFALL contributions will be presented at the AGU26 Annual Meeting, held at the Moscone Center, San Francisco.
“Multi-parametric environmental monitoring and ambient seismic noise analysis for unstable rock slope characterization: Insights from the TWINFALL project”, by Lorenzo Panzeri, Yixin Guo, Laura Longoni, Monica Papini and Diego Arosio. This study asks whether changes in the seismic response of the towers point to lasting structural change or to reversible seasonal effects. Long-term meteorological, hydrological and geotechnical records from ARPA are analysed to identify periodic patterns, seasonal trends and short-lived anomalies, particularly around thermal cycles and intense rainfall.The seismic work starts with single-day surveys from 2017, 2018 and 2019, which are compared to look for early signs of change in resonance frequencies, spectral peaks and polarisation. A new network of five stations, one on each tower and one on stable rock, now records ambient noise continuously. This makes it possible to follow these indicators over time and to compare the behaviour of the four towers. Combining the seismic results with environmental records and with extensometer, crack meter, inclinometer and rock temperature data is intended to support early-warning strategies tailored to the site. It will be presented on December 8, at 17:35.
“Physics-based numerical simulation of rock slope seismic response using SPEED: Integrating complex site geology and anthropogenic source characterization”, by Yixin Guo, Alessio Fumagalli, Ilario Mazzieri, Lorenzo Panzeri, Laura Longoni, Monica Papini and Diego Arosio. This work simulates how seismic waves travel through the slope and are amplified, using SPEED with detailed 3D models of the site’s topography, lithology and major vertical fractures. Two of the main uncertainties in such simulations are the seismic source and the subsurface properties, and field data are used to address both. Seismometers at the base of the slope record the continuous noise from the adjacent road and railway to characterise its spectrum and spatial distribution. The five stations at the top provide the recordings used to validate simplified benchmark simulations and calibrate the dynamic material properties (Vp, Vs, Qp and Qs). These calibrated values then feed the full 3D model, whose results are compared with field data. The aim is to evaluate how topographic amplification, structural discontinuities and directional effects influence slope stability, and to provide a tool for testing predictive scenarios. The work will be presented during the poster session scheduled on December 8, at 8:30.
Together, these contributions reflect the aim of TWINFALL: digital twins of unstable slopes that combine field data, monitoring systems and numerical models.
