Dans la mesure où les avalanches apparaissent plus brillantes dans les observations radar que le reste du manteau neigeux environnant, la télédétection SAR de ces surfaces semble pouvoir s’inscrire de manière prometteuse dans leur étude et leur surveillance. La rugosité, en modifiant le comportement d’interaction de la matière avec le rayonnement, est une bonne candidate à l’explication de cette surbrillance et l’objet de ce travail consiste en l’inversion des paramètres la décrivant à partir des observations Sentinel-1. On propose en particulier une modélisation de la rétrodiusion d’une surface recouverte par la neige à partir du modèle IEM de rétrodiusion des surfaces rugueuses et d’un modèle de rétrodiusion volumique au sein d’une couche de neige. Les paramètres du modèles sont ensuite estimés par un algorithme de Monte Carlo maximisant une fonction de vraisemblance Laplacienne entre les rétrodiusion prédites par le modèle et celles observées pour le manteau neigeux. Au préalable, un algorithme d’apprentissage machine LDA est utilisé sur une image multispectrale Sentinel-2 pour classifier les diérents couverts observé dans l’image Sentinel-1 à l’aide d’indices spectraux et de rapports de bandes et extraire les régions identifiées au manteau neigeux. L’inversion des paramètres décrivant le comportement de rétrodiusion des avalanches est ensuite réalisée en fixant les autres grandeurs communes avec le reste du manteau neigeux. L’inversion prédit bien une évolution de ces paramètres mais suggère également que d’autres eets sont en jeu pour expliquer la surbrillance des avalanches conduisant à une modification locale du comportement diélectrique de la neige.

Since its commissioning on 17 January 2019, the SEIS seismometer sent on the InSight mission has recorded numerous seismic signals, the origins of most of which remain to be determined. The possibility that some of these signals originate from slope instability processes is not zero and remains to be determined. We propose to study this hypothesis by numerical simulation by modelling the seismic signals induced by different gravity flows resulting from simulation in order to study their properties and their detectability by the SEIS seismometer. By taking into account various hypothetical internal geological models of Mars, combined with various examples of simulated slope instability modelled by the gravity flow simulation model, it is possible to calculate the synthetic signals from these sources and the geological models. Thus, we show that the geological model taken into account, and in particular the layers related to the surface layers of the Martian regolith, influence the waveforms of the synthetic signals. In the presence of a large number of sedimentary rocks, the amplitudes are strongly reduced. The sources with which the synthetics are calculated can be of two types. These sources can be by injection on the topography, i.e. the flowing material will erode the soil, adding material during the flow. This model does not respect the law of conservation of mass and seems to induce synthetics of lower amplitude than with the other type of source, the conventional sources which respect the law of conservation of mass. Following these parameters and therefore following plausible scenarios of flows on Mars, the synthetics created have a frequency content that allows them to be detected by the SEIS seismometer without being drowned in the Martian ambient 'noise'. This means that it is certainly possible to detect seismic signals from slope instabilities with the SEIS seismometer.

On Earth, the study of permafrost is an actual topic because its degradation is the direct consequence of the increase in atmospheric temperature. This degradation is reflected in particular by landslides in glacial regions. On Mars, due to the global climatic conditions of the planet, permafrost is found at all latitudes. This planet has also suffered degradation of the latter by various mechanisms such as destabilization of slopes or chaotic subsidence. Landslides, or on Mars other processes, are marked by morphologic structures that called molards. A morphological analysis based on a study in Iceland, through establishment of recognition criteria, was produced at two sites in Greenland and five sites on Mars. Measurements of the height, the width, the slope of the sides of the molards as well as the distance and the height of the slides were made. These analysis failed to discriminate molards from other structures that resemble them such as hummocks. On the other hand, the morpho-dynamic analysis of molards reveal the link that there may be between the deposition of the latter and the gravitational movement which puts them in place. In Greenland, this link is explicit because the distance traveled by molards increases when the height of fall increases. On Mars, this link is more or less according to the processes : it is clear for the processes of subsidence of chaotic lands, hazy for the landslide of Ius Chasma because of the topographic nature, and non-existent for the ejecta flow of Hale Crater. This work provides arguments in favor of the morpho-dynamic control of landslides on molards unlike hummocks, which allows them to be discriminated. The interest of the study of terrestrial molards in the topic of current climate change will be growing because it will allow to address important societal and scientific questions. For martian molards, they could also allow the chronological analysis of the climatic variations of the planet.

Avalanches change the amount of energy exchange between air and snow and can modify the surface roughness of snowpack depositions. So, surface roughness evaluation is very important to better understand and therefore better predict the evolution of the snowpack. The aim of this study is to evaluate the variation of surface roughness by studying the backscattering coefficients from time series Synthetic Aperture Radar (SAR) observations from Sentinel-1 during avalanche events in Tarentaise valley (French Alps). This analysis is carried out by using a pysical backscattering model (namely IEM). First, the backscattering coefficient of snowpack covers is derived from the Sentinel-2 optic images. Then, the posterior probability function of roughness parameters has been concluded using a MCMC inversion method. we present that snowpack had higher surface roughness after avalanches. Also we demonstrate that with Integral Equation Model (IEM), we are able to retrieve the correlation length of real terrain surface after the avalanche.

Sediment transport is a fundamental geomorphological process which shapes the landscape and governs the temporal evolution of landforms. It turns out that the SAR interferometry technique is an effective tool to evaluate the surface deformation, and thus, could be employed to study the sediment transport in rivers. In this work, we investigate the sediment transport using this technique on the Remparts river located in the Réunion island, which is an ideal natural laboratory for the study of sediment transport and related issues due to its tropical volcanic nature. We show that the topography of the Remparts valley is satisfactory for the SAR imaging. By measuring the displacement from the InSAR interferogram in different time periods, we demonstrate that the sediment transport in the Remparts river is significantly driven by extreme climate events.

Avalanches pose significant risks to people and infrastructure. Avalanches also change the surface condition of the snowpack. Assessing this impact is therefore important in order to better understand and therefore better predict the evolution of the snowpack. The aim of this study is to evaluate the snowpack backscattering coefficients from time series Synthetic Aperture Radar (SAR) observations from Sentinel-1 during avalanche events in Tarentaise valley (French Alps). The analysis is carried out by using a micro- wave backscattering model (namely IEM). First, a segmentation of the SAR data is derived from the Sentinel-2 optical observations. Then, we empirical derived snow properties from inversion using a classical Monte Carlo method. We show that snowpack presents a higher surface roughness after avalanches. We also demonstrate that this surface roughness is recovered once snowfall occurs.

During the 20th century, aerial campaigns carried out by the IGN photographed the glaciers surrounding the Chamonix-Mont-Blanc Valley. These images allowed us, using the technique of photogrammetry, to reconstruct the 3D geometry of the Bossons glacier during the years 1939, 1949, 1967, 1979 and 1988. The goal of this study is to see the relationship between the glacier's geometry and climatic variations. We then highlight the very reactive nature of this glacier, indeed we notice a retreat of the Bossons front in 1949 and then a progression towards the valley until 1988. This retreat was caused by the lack of winter precipitation and a pronounced summer ablation during this period. The advance of the glacier is mainly explained by the general climate that prevailed from 1939 to 1988 with a relatively stable temperature that decreased slightly over time. However, this study is based on models that need to be refined in order to best confirm this evolution of the glacier.

The high rates of erosion that have been evident for several decades on the slopes of volcanic islands in tropical environments have caused landslides in the past, which contribute to the gradual filling of river valleys that are strongly characterized by their steep and funnel-like shape, with the risk of flooding the towns located at their mouths. The erosion processes leading to these phenomena appear to be strongly linked to the tropical climate and volcanic activity seem to be evolving worse with climate change. A textbook case exists with the Rivière des Remparts located south-east of Reunion Island, where transfers of detrital materials from slope scree and landslides have been highlighted. The study by Garcin et al. in 2005 quantified a sedimentary accumulation in the river via the study of SPOT satellite images coupled with numerical models of the terrain mainly resulting from the great Mahavel landslide of 1965; this quantity was estimated at 8 million m3 between 1997 and 2002, according to a topographic differential of 20m obtained from DTMs according to a transport of material over a distance of 5km, attesting to the critical level of the bed of the Rivière des Remparts. In this new study, the approach of estimating the material transfer by the river during the wet season 2016-2017 was enriched by the use of Radar satellite images which provide a sedimentary volume transported on three locations of the river estimated between 2000 and 45 000 cubic meters by calculation from InSAR processing. Optical imagery supported these results with the detection of change in the river and also highlighted the effects of climatic phenomena on the slopes. The aim of this project is to establish the feasibility of such a methodology by studying all its aspects.

During its one-year orbital mission, the Dawn probe mapped the surface of Vesta, brightest object in the Asteroid Belt. Features, such as craters, avalanches, cliffs and ejectas, with unexpected and unexplained brightness were noticed over various phase angles. In order to explain these variations in brilliance, here in the crater Cornelia and the Matronalia Rupes cliff, the Hapke model, requently used in photometry, is tested, throughout a Monte Carlo and Bayesian approach. The several parameters of the model allow to characterize photometrically and physically the local surface of Vesta (in terms of simple scattering albedo, porosity, roughness, direction and shape of the scattering lobe). Due to lack of data near the opposition effect (low phase angles), a more suitable Hapke model is used, without the opposition effect functions. Tests performed on synthetic data validate this choice. Results for the seven bright features show that the porosity parameter cannot be constrained by the model, or by the data. All Cornelia avalanches have a tendency to scatter forward in a rather anisotropic way whereas it is undetermined for the Matronalia Rupes areas, due to insufficient data. Almost all of seven features (except one) share a quite similar photometric roughness where it could be expected to be much more various due to their diverse morphology.

The largest moon of Saturne, Titan, presents unique characteristics in the Solar System: first, its huge size (larger than Mercury), a rich and dense atmosphere mostly composed of methane and dinitrogen are factors to consider, but an uncommon aspect is the presence of hydrocarbons in a liquid state. Lakes, rivers and other surprisingly familiar Earth-like landscapes (dunes, mountains) have been observed by the Cassini mission probe sensors (namely the VIMS color camera, and the RADAR). These geological structures bear witness to a little-known activity of the geological layers of Titan. In order to understand the mechanisms of these landscapes formation, and their bounds to the local climate, a model of the geological evolution has been developed. It was calibrated for Titan, in particular, for the landing site of the Huygens probe, surrounded by mountains eroded by rivers. With this model, we simulated morphological patterns similar to those observed (the comparison was made with accurate criteria). We derived a methane rainfall rate from this model, which is consistent with the General Circulation Models (GCM) at our disposal. Also, we could predict a relatively young geological formation of the surface of Titan from our model.

Landslides play a major role in the erosion and transport of material on the surface of our planet and constitute a major natural hazard for populations and infrastructures. The objective of the study is to produce numerical surface models on Reunion Island and to assess their quality. We are setting up a processing chain that is as automatic as possible to generate digital surface models (DSM) from IGN® aerial images and by integrating tools from the MicMac software. Then we estimate the quality of our digital terrain model by means of a roughness estimator, the HRMS, and compare our results with a reference model. We can thus produce DEM from 150 images in 9 hours and know the quality of the rendered product. This method is applicable to future surveys on areas with steep slopes (by UAV for example). The study can be continued on all aerial missions since 1978, carry out DEM differences between missions to quantify the volumes displaced by landslides and put them into context with the microseismic, geochemical and climatic measurements carried out by the IPGP in Reunion Island.