Integration of satellite interferometry and geological methods for landslide research

Resum

Properly assessing the hazard of landslides begins with mapping and characterising them from a geological perspective. Although landslides have demonstrated a worldwide impact and are the second most damaging geohazard in Spain (after floods), there is still little social awareness about them. This fact evidences the necessity of additional efforts for the study of these natural phenomena. The integration of an innovative remote sensing technique, like satellite radar interferometry, with geomorphological and geological methods has demonstrated to be an effective multi-technique approach for landslide research. The present Ph.D. Thesis is developed in this framework and improves the knowledge of landslides by applying such combination of methodologies. Two critical and interesting areas in the Province of Granada (Southern Spain) have been selected for study: the Sierra Nevada Range and the Rules Reservoir. The Sierra Nevada is a high-elevation mountain range where some landslides have been unreported or unnoticed, despite being susceptible to slope movements. In this case, Differential Synthetic Aperture Radar (DInSAR) and Landscape Analysis techniques were integrated to optimise landslide mapping and provide an updated landslide inventory map of the range. The Landscape Analysis was based on the identification of river anomalies by the double normalised channel steepness (ksnn) index, a novel derivation from the conventional normalised steepness index (ksn) that reduced the active tectonics signal of the area. The visual exploration of ksnn anomalies and the unstable ground areas obtained from DInSAR velocity maps evidenced 28 new landslides. This mapping reveals a significant increase of the area affected by landslides (33.5%) compared with the previous inventory from the Geological and Mining Institute of Spain (IGME-CSIC) (14.5%). A relevant finding of this study is the identification, for the first time, in the Sierra Nevada, of two landslide typologies: Deep-Seated Gravitational Slope Deformations (DGSDs) and rockslides. Their diffuse boundaries, homogeneous lithology (schists), and previous glacial morphologies made delimiting these landslides difficult, but the utilised techniques greatly facilitated the process. The geomorphological observations made in the field and the exploration of maps (e.g. slope, hillshade, aspect, rugosity) derived from high-resolution Digital Elevation Models (DEMs) were fundamental procedures for accurately defining the landslides’ boundaries, as well as to describe landslide-related morphologies. Due to the large size and typology of these landslides, attempts were made to provide a first insight about their hazard and potential impacts in the region. The Rules Reservoir is one of the most strategic infrastructures in the Province of Granada, with well-known slope instability problems during and after its construction. In this case, DInSAR techniques were applied in the reservoir’s slopes to derive ground velocity maps that revealed three active landslides. The thorough geomorphological investigation, based on field observations and photo-interpretation of historical aerial images, allowed to distinguish between rotational (Lorenzo-1 and Rules Viaduct landslides) and translational (El Arrecife Landslide) landslide typologies, as well as to identify surficial damages related to their activity. The DInSAR-derived times series of accumulated displacement (TSs) revealed a correlation between the acceleration of the rotational landslides’ movement and drawdowns of the reservoir water level. Due to their dimensions, rotational character and minor accelerations, a rapid slope failure and sudden collapse into the reservoir is not expected from the Lorenzo-1 and Rules Viaduct landslides. However, they pose a risk to nearby infrastructures due to the retrogressive evolution of these landslides: the Lorenzo-1 Landslide is already affecting the N-323 National Road, while the Rules Viaduct Landslide may be provoking deformation on the southern sector of the Rules Viaduct (A-44 Highway). Regarding the El Arrecife Landslide, its translational character implies a greater potential hazard and further efforts were made to characterise this landslide. The El Arrecife Landslide is located in the western slope of the Rules Reservoir and it was identified by using DInSAR data, as the poorly defined boundaries of the landslide made its recognition in the landscape challenging. This landslide was analysed by a multitechnique approach to provide a rapid characterisation and comprehensive understanding of its structure, volume and historical activity. The structural field surveys enabled the identification of many foliation orientations of the rocks (phyllites) and a kinematic analysis revealed the most probable orientation to cause a planar slope failure. The estimated location of the surface of rupture allowed determining the extremely large volume of the landslide (14.7 million m3). The short-term activity of the landslide (last 5 years) was evidenced by DInSAR, while geophysical data based on Ground Penetrating Radar (GPR) data revealed its medium-term activity (last 22 years). Both techniques evidenced a vertical movement of the landslide around 2 cm/yr. Photogrammetric techniques based on the Structure-for-Motion (SfM) method were also applied, but no rapid shallow movements were detected during the analysed period (14 years). Besides having an overall translational movement, the landslide’s foot is composed by smallersize rotational landslides. The DInSAR TSs indicated that variations in the reservoir water level do not affect the overall landslide body, but drawdowns of the water level do accelerate the movement of these rotational slides. Therefore, the most significant hazard of the El Arrecife Landslide is related to such rotational slides, that have been causing damage to the N-323 National Road for several decades and are expected to persist. Although improbable, the possibility of a rapid and sudden acceleration of the entire landslide and subsequent collapse into the reservoir cannot be underestimated, given its translational kinematics and large size. It is therefore crucial to consider the response of this landslide to possible hazardous scenarios derived from extraordinary events, such as drastic reservoir water level drawdowns, intense precipitation, or an earthquake.