Slow rock slope deformations (DSGSDs and large landslides) are widespread, affect entire hillslopes and displace volumes up to billions of cubic meters. They evolve over long time by progressive failure processes, under variable climatic and hydro-mechanical coupling conditions mirrored by a complex creep behaviour. Although characterized by low displacement rates (up to few cm/yr), these slope instabilities damage sensitive structures and host nested sectors potentially undergoing rockslide differentiation and collapse. A robust characterization of the style of activity of slow rock slope deformations is required to predict their interaction with elements at risk and anticipate possible failure, yet a comprehensive methodology to this aim is still lacking. In this perspective, we developed a multi-scale methodology integrating geomorphological mapping, field data and different DInSAR techniques, using an inventory of 208 slow rock slope deformations in Lombardia (Italian Central Alps), for which we performed a geomorphological and morpho-structural mapping on aerial images and DEMs. On the regional scale, we developed an objective workflow for the inventory-scale screening of slow-moving landslides. The approach is based on a refined definition of activity that integrates the displacement rate, kinematics and degree of internal damage for each landslide. Using PS-InSAR and SqueeSAR datasets, we developed an original peak analysis of InSAR displacement rates to characterize the degree of segmentation and heterogeneity of mapped phenomena, highlight the occurrence of sectors with differential activity and derive their characteristic displacement rates. Using 2DInSAR velocity decomposition and machine learning classification, we set up an original automatic approach to characterize the kinematics of each landslides. Then, we sequentially combine PCA and K-medoid cluster analysis to identify groups of landslides characterized by consistent styles of activity, accounting for all the relevant aspects including velocity, kinematics, segmentation, and internal damage. Starting from the results of regional-scale classification, we focused on the Corna Rossa, Mt. Mater and Saline DSGSDs, that are emblematic case studies on which apply DInSAR analysis to investigate typical issues in large landslide studies (spatial segmentation, heterogenous activity, sensitivity to hydrological triggers). We applied a targeted DInSAR technique on multiple temporal baselines to unravel the spatial heterogeneities of complex DSGSDs and through a novel stacking approach on raw long temporal baseline interferograms, we outlined the permanent displacement signals and sectors with differential evolution as well as individual active structures. We then used DInSAR to investigate the possible sensitivity of slow rock slope deformations to hydrological triggers. Comparison between seasonal displacement rates, derived by interferograms with targeted temporal baselines, and time series of precipitation and snowmelt at the Mt. Mater and Saline ridge outlined complex temporally shifted seasonal displacement trends. These trends, more evident for shallower nested sectors, outline dominant controls by prolonged precipitation periods modulated by the effects of snowmelt. This suggests that DSGSDs, often considered insensitive to short-term (pluri-annual) climatic forcing, may respond to hydrological triggering, with key implication in the interpretation of their progressive failure. Our results demonstrated the effectiveness of the proposed multi-scale methodology that exploits DInSAR products and targeted processing to identify, classify and characterize the activity of slow rock slope deformation at different levels of details by including geological data in all the analysis stages. Our approach, readily applicable to different settings and datasets, provides the tools to solve key scientific issues in a geohazard-oriented study of slow rock slope deformations.

Le deformazioni lente di versante in roccia (DGPV e grandi frane) sono fenomeni diffusi che interessano interi versanti e mobilizzano volumi di roccia anche di miliardi di metri cubi. La loro evoluzione è legata a processi di rottura progressiva sotto forzanti esterne e di accoppiamento idromeccanico, rispecchiate da un complesso processo di creep. Sebbene caratterizzate da bassi tassi di spostamento (fino a pochi cm / anno), queste instabilità di versante danneggiano infrastrutture e ospitano settori potenzialmente soggetti a differenziazione e collasso catastrofico. È quindi necessaria una robusta caratterizzazione del loro stile di attività per determinare il potenziale impatto sugli elementi a rischio e anticipare un eventuale collasso. Tuttavia una metodologia di analisi finalizzata a questo scopo è ancora mancante. In questa prospettiva, abbiamo sviluppato un approccio multiscala che integra dati morfostrutturali, di terreno e tecniche DInSAR, applicandoli allo studio di un inventario di 208 deformazioni lente di versanti mappate in Lombardia. Su questo dataset abbiamo eseguito una mappatura geomorfologica e morfostrutturale di semi dettaglio tramite immagini aeree e DEM. Abbiamo quindi sviluppato un pacchetto di procedure oggettive per lo screening su scala di inventario delle deformazioni lente di versante integrando dati di velocità di spostamento, cinematica e di danneggiamento dell’ammasso roccioso per ogni frana. Utilizzando dataset PS-InSAR e SqueeSAR, abbiamo sviluppato una procedura mirata a identificare in maniera semiautomatica la velocità InSAR rappresentativa, il grado di segmentazione e l'eterogeneità interna di ogni frana mappata identificando la presenza di possibili fenomeni secondari. Utilizzando la tecnica 2DInSAR e tecniche di machine learning, abbiamo inoltre sviluppato un approccio automatico caratterizzare la cinematica di ciascuna frana. I dati così ottenuti sono stati integrati tramite analisi di PCA e K-medoid per identificare gruppi di frane caratterizzati da stili di attività simili. Partendo dai risultati della classificazione su scala regionale, ci siamo poi concentrati su 3 casi di studio emblematici, le DGPV di Corna Rossa, Mt. Mater e Saline, rappresentativi di problematiche tipiche delle grandi frane (segmentazione spaziale, attività eterogenea, sensibilità alle forzanti idrologiche). Applicando un approccio DInSAR mirato abbiamo indagato la risposta del versante a diverse baseline temporali per evidenziare le eterogeneità spaziali e, tramite un nuovo approccio di stacking su basline temporali lunghe abbiamo estrattoi segnali di spostamento permanenti ed evidenziato i settori e le strutture con evoluzione differenziale. Lo stesso approccio DInSAR è stato utilizzato per studiare la sensibilità delle deformazioni lente di versante alle forzanti idrologiche. Il confronto tra i tassi di spostamento stagionale e le serie temporali di precipitazioni e scioglimento neve per il monte. Mater e Saline hanno delineato complessi trend di spostamento stagionale. Queste tendenze, più evidenti per i settori più superficiali, evidenziano una risposta maggiore a periodi prolungati di precipitazione modulati dagli effetti dello scioglimento della neve. Ciò suggerisce che le DGPV, spesso considerate non influenzate dalla forzante climatica a breve termine (pluriennale), sono sensibili a input idrologici, con implicazioni chiave nell'interpretazione del loro fallimento progressivo. I nostri risultati hanno dimostrato l'efficacia della metodologia multi-scala proposta, che sfrutta i prodotti DInSAR e l'analisi mirata per identificare, classificare e caratterizzare l'attività delle deformazioni lente di versante includendo dati geologici in tutte le fasi dell'analisi. Il nostro approccio, è applicabile a diversi contesti e dataset e fornisce gli strumenti per indagare processi chiave in uno studio finalizzato alla definizione del rischio connesso alle deformazioni lente di versante.

(2021). Regional and local scale analysis of very slow rock slope deformations integrating InSAR and morpho-structural data. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2021).

Regional and local scale analysis of very slow rock slope deformations integrating InSAR and morpho-structural data

CRIPPA, CHIARA
2021

Abstract

Slow rock slope deformations (DSGSDs and large landslides) are widespread, affect entire hillslopes and displace volumes up to billions of cubic meters. They evolve over long time by progressive failure processes, under variable climatic and hydro-mechanical coupling conditions mirrored by a complex creep behaviour. Although characterized by low displacement rates (up to few cm/yr), these slope instabilities damage sensitive structures and host nested sectors potentially undergoing rockslide differentiation and collapse. A robust characterization of the style of activity of slow rock slope deformations is required to predict their interaction with elements at risk and anticipate possible failure, yet a comprehensive methodology to this aim is still lacking. In this perspective, we developed a multi-scale methodology integrating geomorphological mapping, field data and different DInSAR techniques, using an inventory of 208 slow rock slope deformations in Lombardia (Italian Central Alps), for which we performed a geomorphological and morpho-structural mapping on aerial images and DEMs. On the regional scale, we developed an objective workflow for the inventory-scale screening of slow-moving landslides. The approach is based on a refined definition of activity that integrates the displacement rate, kinematics and degree of internal damage for each landslide. Using PS-InSAR and SqueeSAR datasets, we developed an original peak analysis of InSAR displacement rates to characterize the degree of segmentation and heterogeneity of mapped phenomena, highlight the occurrence of sectors with differential activity and derive their characteristic displacement rates. Using 2DInSAR velocity decomposition and machine learning classification, we set up an original automatic approach to characterize the kinematics of each landslides. Then, we sequentially combine PCA and K-medoid cluster analysis to identify groups of landslides characterized by consistent styles of activity, accounting for all the relevant aspects including velocity, kinematics, segmentation, and internal damage. Starting from the results of regional-scale classification, we focused on the Corna Rossa, Mt. Mater and Saline DSGSDs, that are emblematic case studies on which apply DInSAR analysis to investigate typical issues in large landslide studies (spatial segmentation, heterogenous activity, sensitivity to hydrological triggers). We applied a targeted DInSAR technique on multiple temporal baselines to unravel the spatial heterogeneities of complex DSGSDs and through a novel stacking approach on raw long temporal baseline interferograms, we outlined the permanent displacement signals and sectors with differential evolution as well as individual active structures. We then used DInSAR to investigate the possible sensitivity of slow rock slope deformations to hydrological triggers. Comparison between seasonal displacement rates, derived by interferograms with targeted temporal baselines, and time series of precipitation and snowmelt at the Mt. Mater and Saline ridge outlined complex temporally shifted seasonal displacement trends. These trends, more evident for shallower nested sectors, outline dominant controls by prolonged precipitation periods modulated by the effects of snowmelt. This suggests that DSGSDs, often considered insensitive to short-term (pluri-annual) climatic forcing, may respond to hydrological triggering, with key implication in the interpretation of their progressive failure. Our results demonstrated the effectiveness of the proposed multi-scale methodology that exploits DInSAR products and targeted processing to identify, classify and characterize the activity of slow rock slope deformation at different levels of details by including geological data in all the analysis stages. Our approach, readily applicable to different settings and datasets, provides the tools to solve key scientific issues in a geohazard-oriented study of slow rock slope deformations.
AGLIARDI, FEDERICO
CROSTA, GIOVANNI
Frane molto lente; InSAR; Analisi statistica; Analisi strutturale; Approccio multiscala
Very slow landslides; InSAR; Statistical analysis; Structural analysis; Approccio multiscala
GEO/05 - GEOLOGIA APPLICATA
Italian
18-feb-2021
SCIENZE CHIMICHE, GEOLOGICHE E AMBIENTALI
33
2019/2020
open
(2021). Regional and local scale analysis of very slow rock slope deformations integrating InSAR and morpho-structural data. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2021).
File in questo prodotto:
File Dimensione Formato  
phd_unimib_763185.pdf

accesso aperto

Descrizione: Tesi di Crippa Chiara- n°763185
Tipologia di allegato: Doctoral thesis
Dimensione 19.56 MB
Formato Adobe PDF
19.56 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10281/306309
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
Social impact