Glaciers in the western United States, essential freshwater reservoirs, and climate change indicators, are evaluated to understand their present conditions using Sentinel 2A satellite imagery and Random Forest machine learning. Furthermore, this study utilizes MODIS albedo and temperature data, along with Sentinel 5 Precursor, to understand trends and changes in summer glacier albedo within the region, while also investigating the underlying factors influencing glacier albedo, with Mount Rainier as case study. The results of this study reveal that glaciers have lost 236.96 km2 in their area and a corresponding volume loss of 4.9 km3 from the mid-20th century to 2020. Additionally, analysis on the albedo changes of glaciers on Mount Rainier shows a decreasing trend at a rate of 3.29 x 10-6 per day. Further analysis reveals that temperature is the dominant factor affecting the albedo of glaciers in this region with Carbon Monoxide and aerosols following suit.

In glacierised regions, suspended sediment fluxes are highly responsive to climate-driven environmental change and can provide important information regarding the relationships between glacier variations, climate and geomorphic change. As a result, understanding patterns of suspended sediment transport and their relationship with meltwater delivery is of critical importance. However, studies of glacial suspended sediment transport are often limited by interpreting patterns of suspended sediment transfer based on whole-season data, allowing precise patterns to become masked. This thesis aims to contribute to the understanding of suspended sediment transfer in glacierised basins through the investigation of patterns of suspended sediment delivery to the proglacial area of Storglaciären, a small polythermal valley glacier located in the Tarfala valley, Arctic Sweden. High temporal resolution discharge and suspended sediment concentration data were collected during two summer field campaigns at Storglaciären. Interpretations of suspended sediment transport data were made using diurnal hysteresis and sediment availability data, combined with suspended sediment shape and magnitude data classified by applying principal component and hierarchical cluster analyses. Analysis of the dominant discharge generating processes at Storglaciären was also conducted using principal component analysis, allowing patterns of discharge to be better understood. This was complemented by analysis of the structure and evolution of the glacier drainage system by linear reservoir modelling and flow recession analysis. The results suggest that patterns of discharge and suspended sediment transport at Storglaciären are complex, with distinct processes and magnitudes of transport evident at both proglacial outlet streams, Nordjåkk and Sydjåkk. These processes are intrinsically linked to meteorological variables, with both ablation-driven and precipitation-driven discharge exerting influence over patterns of suspended sediment transport in the proglacial area of Storglaciären.

Le permafrost de montagne est menacé par le réchauffement atmosphérique, une évolution qui s’accompagne de l’augmentation des phénomènes tels que les chutes de pierres, la formation de thermokarsts et l’accélération des glaciers rocheux. La déstabilisation des glaciers rocheux, qui compromet l’intégrité structurelle de ces formes, semble liée au réchauffement atmosphérique, et a suscité un intérêt grandissant au cours des dernières années. Ce phénomène, qui peut être provoqué par le réchauffement du pergélisol ou des contraintes mécaniques externes, est caractérisé par une accélération anormale des glaciers rocheux affectés, et par l’apparition des signes géomorphologiques telles que des fissures et des crevasses à sa surface. Bien que ce processus peut être transitoire, il peut déterminer une phase de crise amenant le glacier rocheux à un effondrement.Cet étude se préfixe de fournir une première évaluation des phénomènes de déstabilisation de glacier rocheux à l’échelle des Alpes françaises. Dans un premier temps, l’empreinte spatiale du pergélisol a été évaluée afin de produire une carte de répartition du pergélisol régionale, un outil nécessaire pour estimer l’état du permafrost dans les glaciers rocheux. La deuxième étape a consisté à identifier les formes déstabilisées grâce à une observation ponctuelle des images aériennes afin d’identifier les caractéristiques typiquement observables sur les glaciers rocheux déstabilisés. Il est alors possible de comprendre les conditions topoclimatiques typiques dans lesquelles se produit ce phénomène et de repérer les formes susceptibles de subir ce processus. Enfin, les efforts ont été concentrés sur le glacier rocheux du Lou, déstabilisé, qui, du fait d’un détachement de couche active, a conduit à une lave torrentielle en Août 2015. L’analyse a visé à mieux définir les circonstances de cet événement, en mettant l’accent sur les facteurs de préconditionnement, de préparation et de déclenchement et sur leur interaction avec le processus de déstabilisation.Les résultats ont fourni des informations riches sur la zone périglaciaire de la région. La modélisation de la répartition du pergélisol a mis en évidence les étendues de la zone périglaciaire dans la région qu’on peut trouver sur les pentes de débris au-dessus de 2300 - 2500 m.a.s.l. en fonction de l’exposition solaire et des caractéristiques régionales des précipitations. L’observation des photographies aériennes a permis d’observer 46 formes en cours de déstabilisation, soit 12% des glaciers rocheux actifs des Alpes françaises. Il apparaît que la déstabilisation est plus susceptible de se produire dans certaines conditions topoclimatiques locales spécifiques, en particulier dans des pentes exposées au nord, raides et convexes situées aux marges inférieures de la zone de pergélisol. Un grand nombre de glaciers rocheux ne présentant actuellement aucune déstabilisation sont donc susceptibles d’être affectés par une déstabilisation future. L’analyse du glacier rocheux du Lou a révélé que la déstabilisation est liée à une avancée rapide du front vers un ravin torrentiel. Ce processus semble avoir accru la prédisposition des matériaux détritiques du front à être mobilisés par du ruissellement, des précipitations relativement modérées ayant suffi à déclencher l’événement.Malgré les incertitudes liées aux méthodes impliquées, les résultats suggèrent que les conditions favorables à la déstabilisation sont fréquentes, et que cette dernière peut augmenter le niveau de risque si le site est connecté à des infrastructures humaines. Des efforts supplémentaires doivent donc être entrepris, afin d’améliorer la compréhension de ces processus, notamment par la surveillance des sites ainsi que par une évaluation locale complète des cascades de processus liés à ce phénomène. / As occurring to several geosystems on our planet, mountain permafrost is threatened by climate change as prolonged warming may compromise the geotechnical properties of the frozen ground. As result, increasing occurrence of rockfall activity, thermokarst formation and rock glacier acceleration was observed in the past decades. Rock glacier destabilization, a process that compromises the structural integrity of these landforms, seems to be linked to atmospheric warming, gaining interest in the past years. The destabilization, which may be triggered by warming permafrost or mechanical stress, is characterized by an anomalous acceleration of the landform and the occurrence of specific features such as cracks and crevasses on its surface. Although the occurrence of these processes is mostly transitory, determining a textit{crisis} phase of the landform, in exceptional cases it may lead the rock glacier to structural collapse.This PhD thesis provided an assessment on the occurrence and related processes of rock glacier destabilization in the French Alps. At first, the spatial occurrence of debris permafrost was assessed in order to provide the permafrost distribution map of the French Alps, a tool that was necessary to evaluate permafrost conditions at rock glaciers sites. The second step consisted in an identification of destabilized rock glaciers in the region, which was done by multiple orthoimages interpretation aimed to identify features typically observable on destabilized rock glacier. Once identified the destabilized rock glaciers it was possible to analyse the typical topographical settings in which destabilization occurs and to to spot those landforms that are susceptible to experience this phenomenon. After these efforts at the regional scale, the focus was shifted towards local scale investigations at the Lou rock glacier, a partially destabilized landform that, due to frontal failure, in August 2015 triggered a debris flow that caused significant damages to buildings. The analysis aimed to better define the circumstances of this event, focusing on preconditioning, preparatory and triggering factors and their interaction with the destabilization process.The results provided interesting insights on the issue of destabilizing rock glaciers in the region. Permafrost distribution modeling demonstrated the large extents of the periglacial zone in the region as it can be found in debris slopes above 2300 - 2900 m.a.s.l. depending upon solar exposure and regional precipitation characteristics. Rock glacier destabilization was observed on 46 landforms, i.e. the 12% of the active rock glaciers. Destabilization was found to be more likely to occur in specific local topo-climatic conditions, consisting of north facing, steep and convex slopes at the lower margins of the permafrost zone. A large number of rock glaciers currently not showing destabilization was found to be located in these conditions and suggested to be susceptible to future destabilization. As demonstrated by the Lou rock glacier analysis, destabilization was found to be a relevant phenomena in the context of permafrost hazards. At this site, rock glacier destabilization was linked to a rapid frontal advance towards a torrential gully. This process seemed to have increased the site predisposition to frontal failure as a mild rainstorm was sufficient to trigger the event.Despite methodological uncertainties, results indicated that destabilization occurrence is widespread and it may rise the hazard level of a site connected to human infrastructures. Therefore, it is suggested that, where it has been modelled and where stakes may be at risk downslope, rock glacier destabilization deserves to be more carefully investigated. In this sense further efforts should focus towards a better understanding of the destabilization process by site monitoring as well as towards a comprehensive hazard assessment linked to this phenomenon.

Contribution à l'Etude sur les formations tertiaires et quaternaires de la région d'Annemasse

Rapid tropical glacier retreat over the last 50 years has been well documented, and has received significant media attention. Many studies suggest these changes are due to rising global surface air temperatures, however disentangling the effects of temperature and precipitation has hampered scientific consensus. Furthermore, because of the shortness of the instrumental record, it is difficult to assess the larger significance of the climate changes associated with the decline of tropical glaciers. Here, we present a locally calibrated, independent temperature reconstruction for the past 1,500 years from Laguna Chingaza, Colombia based on distributions of branched Glycerol Dialkyl Glycerol Tetraethers (brGDGTs) in order to assess the controls on long term temperature variability in the tropical Andes, and their relationship with growth and demise of Andean glaciers. Comparison of reconstructed temperatures with the instrumental record suggests that our proxy record faithfully records decadal to century scale trends in temperature. The largest temperature decline over the last 1,500 years was a decrease of 2.5 ± 0.3 °C during the Little Ice Age (LIA), reaching lowest temperatures during the mid-17th century, and is broadly consistent with terrestrial temperature reconstructions throughout the tropics and the higher latitudes. The structure and timing of temperature changes at Laguna Chingaza are remarkably similar to recent terrestrial temperature reconstructions from elsewhere in the tropics, including sites in the tropical Pacific and equatorial Africa, suggesting that these changes are widespread in the tropics. Together, these records suggest that warming over the last few decades is unprecedented over the last 1,500 years, including the Medieval Climate Anomaly (800-1150 AD). Comparison of these temperature changes with records of Andean glacier limits suggests that temperature is the dominant driver of glacial retreat, particularly over the past few decades. Additionally, paleotemperatures inferred from LIA and recent glacial equilibrium line altitudes (ELAs) underestimate past changes in temperature when compared with brGDGTs reconstructions, suggesting that changes in precipitation complicate the use of glacier ELAs to reconstruct past temperatures. The coupling of temperature and ice extent in South America suggests that with projected future warming, the health of tropical glaciers could be in jeopardy, significantly impacting the communities and ecosystems that depend on them. / text

In the summer of 2010, fieldwork was conducted in the Robertson Valley, Canadian Rocky Mountains, Alberta to assess the magnitude and controls of microbial nitrification in proglacial till and in supraglacial, subglacial, and proglacial streams. Seasonal precipitation and glacial and proglacial runoff was sampled for hydrochemical and stable isotope analyses (δ18O and δ15N of nitrate [NO3-]). Lower Ca:Mg ratios, higher mean Σmajor ions, and an increased importance of reactions with slower dissolution kinetics in subglacial streams and proglacial seeps indicated waters here experienced longer rock-water contact time than in dilute supraglacial streams. Additionally, waters emanating from longer residence time flowpaths acquired substantial NO3- from nitrification reactions. Using δ18O-NO3- in a simple end-member mixing model, the fraction of NO3- derived from microbial nitrification was estimated to be 44 to 56% in the two subglacial streams, and greater than 80% in proglacial seeps. These results show that atmospherically-derived nitrogen (N) in this glacial valley undergoes substantial biological cycling prior to export in surface runoff. Water flowing from the east subglacial stream (RE) received a larger portion of its melt from a sediment-rich, slow drainage system and had a higher proportion of nitrified NO3- compared to the west subglacial stream (RW), where runoff was similar in composition to supraglacial runoff, indicating that the nature of subglacial flowpaths is an important factor in determining the amount of microbially-cycled nutrients that are exported from a glacier. Sixteen 34-day in situ soil incubations revealed that net mineralization and net nitrification occurred at all four sampling sites in the glacier forefield along a 1.6 km chronosequence; however, there was no significant difference among these rates with time since deglaciation or temperature. Instead, net mineralization and net nitrification rates were significantly correlated (p < 0.05, n = 16) with measured physical and chemical soil variables, including total organic carbon, total N, bulk density, pH, and clay content, suggesting that substrate availability is a larger control on N-cycling processes than time since deglaciation. High variability in inorganic soil N pools and N-cycling rates indicates that there are likely hot spots of biogeochemical activity within glacial till. / Thesis (Master, Geography) -- Queen's University, 2012-04-26 14:47:17.29

No description available.

It has become increasingly clear that climate fluctuations during the Holocene interval were unusually frequent and rapid, and that our current understanding of the temporal and spatial distribution of these oscillations is incomplete. Little paleoenvironmental research has been undertaken on the windward side of the central Coast Mountains of British Columbia, Canada. Very high annual orographic precipitation totals, moderate annual temperatures regulated by the Pacific Ocean, and extreme topographic features result in a complex suite of microclimate conditions in this largely unstudied area. Dendroclimatological investigations conducted on a steep south-facing slope near Confederation and Franklin glaciers suggest that both mountain hemlock (Tsuga mertensiana) and subalpine fir (Abies lasiocarpa) trees at the site are limited by previous year mean and maximum summer temperatures. A regional subalpine fir chronology for the central and southern Coast Mountains indicates that subalpine fir trees at the study site experience physiological stress with warm summer temperatures, despite the high annual precipitation totals experienced there. This response is likely a result of the extreme gradient and the aspect of the slope at the sampling location, underscoring the importance of site characteristics on annual radial tree growth. Local (AD 1820-2008) and regional (AD 1700-2008) tree ring width chronologies were used to reconstruct previous July mean and maximum temperatures, explaining between 13% and 36% of the variance in climate. The proxy record features cool intervals that are comparable to other paleoenvironmental research from the region, and cyclical oscillations in temperature commonly associated with the El Niño Southern Oscillation and Pacific Decadal Oscillation. Century-scale fluctuations may be connected to changes in solar irradiance. Dendroglaciological investigations were undertaken at the confluence of the Confederation and Franklin glaciers with the intention of exploring the Holocene behaviour of low-elevation maritime glaciers in this region. These glaciers are suspected to be sensitive to variations in the mean position of winter freezing level heights and warm winter temperatures, and may respond differently to changes in climate than more continental glaciers. Buried wood samples were radiocarbon-dated and cross-dated to construct three floating chronologies. Float A (r = 0.467) suggests an early Little Ice Age advance of the two glaciers, and Float B (r = 0.466) suggests an early Tiedemann advance of Confederation Glacier. Float C (r = 0.519) is dated to the Garibaldi Phase of glacier expansion, but may not have been killed by glacial activity. The temporal synchronicity of these findings with glacial events documented throughout the region suggests a spatially coherent response of maritime and continental glaciers to the dominant climate–forcing mechanisms operating in Pacific North America throughout the late Holocene. The dendroclimatological and dendroglaciological findings of this study help to fill a spatial research gap in the current understanding of Holocene climate variations in British Columbia. Because of the complex and at times topographically-controlled response of conifers to climate in the study area, this region may provide a particular challenge in terms of reconstructing Holocene climate variability.

Ce travail a permis de modéliser le bilan de masse superficiel sur le bassin versant du Glacier 15 du massif Antisana (0.28 km2; 0°28'S, 78°09'W), dans les Andes tropical de l'Equateur. Le bilan de masse superficiel (MBS) entre 2000 et 2008 sur le glacier 15a de l'Antisana a été modélisé en utilisant deux approches parallèles : un modèle empirique de type degré-jour et un modèle physique de type bilan d'énergie.Dans un premier temps, la fonte a été calculée à l'aide de la température moyenne et des cumuls de précipitations alors que le calcul de la sublimation a été réalisé grâce à la vitesse du vent. Ce modèle a été appliqué au pas de temps journalier et a été calé à 4900 m d'altitude entre mars 2002 et août 2003 et validé entre janvier et novembre 2005. Un lien significatif entre la température et la fonte apparaît lorsqu'une distinction entre neige et glace est effectuée. La relation entre la fusion et la température est expliquée essentiellement par le lien significatif existant entre le bilan radiatif de courtes longueurs d'onde et la température, car le rayonnement solaire est aussi le principal moteur de la fonte. Néanmoins, cette relation disparaît dès lors que le vent devient intense. Ce point a cependant peu de conséquences sur les calculs effectués, car les périodes ventées sont marquées par des températures faibles. Le modèle a été appliqué pour évaluer dans quelle mesure cette approche est appropriée pour modéliser le bilan de masse superficiel, l'ablation, et les altitudes de la ligne de neige et de la ligne d'équilibre. Les résultats montrent qu'un modèle empirique de type degré-jour permet une modélisation précise du MBS à l'échelle d'un glacier en Equateur. Néanmoins, ce modèle ne devrait pas être appliqué à d'autres régions tropicales, en particulier là où la sublimation est importante (grâce à une saison sèche prononcée) ou lorsque les glaciers sont situés au-dessus de l'altitude moyenne de l'isotherme 0°C.Dans un deuxième temps, un modèle de bilan d'énergie superficiel (BES) distribué développé initialement pour des applications en Terre d'Adélie (Antarctique) a été appliqué sur le site du glacier 15a de l'Antisana pour quantifier l'ablation de la glace et de la neige. Ce modèle a été adapté pour les conditions tropicales puis forcé à l'aide des mesures météorologiques obtenues à proximité et sur glacier 15a. Le modèle a tout d'abord été appliqué à une altitude de référence à 4900 m pour une pente de 28° et avec un azimut NO (45°). La spatialisation du bilan d'énergie a ensuite été effectuée en discrétisant le glacier par tranches de 50 m d'altitude, puis en calculant la pente et l'azimut moyen de chaque tranche à l'aide de un modèle numérique de terrain (MNT) précis du glacier. Les données d'entrées ont été distribuées sur la langue du glacier 15a en supposant que les forçages météorologiques varient selon des gradients verticaux constants. Ces gradients ont été établis à partir de mesures de terrain. Le modèle a été calé à partir des données de bilan de masse mensuel et d'albédo journalier mesurées sur le glacier 15a entre mars 2002 et août 2003, puis validé à partir de mesures obtenues entre janvier 2005 et décembre 2008. Le modèle reproduit précisément les variations spatiales et temporelles du bilan de masse entre 4850 m et 5700 m d'altitude. Les résultats inédits de bilan d'énergie de surface distribués sur le glacier 15a ont ainsi permis d'analyser les processus physiques à l'origine des variations spatio-temporelles de l'ablation. Enfin, cette étude explique pour la première fois pourquoi les modèles de type degré-jour donnent systématiquement des résultats de bonne qualité dans cette région, venant contredire les résultats d'études antérieures. Cette thèse constitue ainsi une étape importante dans la compréhension de la sensibilité des bilans de masse aux variations de température dans cette région. / Antisana Glacier 15 basin surface mass balance (SMB) (0.28 km2;0°28'S,78° 09'W, 5750 m), located in the tropical Andes of Ecuador, between 2000 and 2008 has been modeled using two models: an empirical positive degree-day (PDD) and other physical surface energy balance approach..As a first stage, melting was calculated using daily temperature and the cumulative precipitation while sublimation was computed as wind-speed function. The PDD was applied at daily-time step and calibrated at 4900 m a.s.l. The period time to calibrate was March 2002 - August 2003 and validated was January - November 2005. A significant link between melting rates and temperature was revealed when model discriminates snow and ice surface. This relationship is explained by the link between the short-wave radiation and air temperature. However, this relationship disappears under intense wind-speed events, however this events show low air temperature, therefore they are not affect the final results. The PDD model was applied for modeling SMB, ablation, snow-line position and equilibrium line altitude. The results show this model type fits to SMB measures over Antisana glacier. However, it should not apply to other tropical regions, particularly where there are high sublimation rates (pronounced dry seasons) or where glaciers are located under the isotherm 0 ° C.A second stage, a distributed surface energy balance (DBES) model over glacier 15a of the Antisana was applied to quantify ice and snow ablation. For this task was used a model adapted to tropical conditions from Adélie Land-Antarctica BES model. Initially, the model was applied at 4900 m.a.s.l. , with a slope of 28º and NW (45 °) azimuth. For computing DBES, the glacier was divided in 18 altitude strips ( 50m/level). The model generated slops and azimuths from a digital elevation model (DEM). The meteorological vertical gradients were measured in the study zone. This values was used into de model as constants during both periods calibration and validation. The DBES model reproduced the spatial variation and temporal of SMB between 4850 m.a.s.l. and 5700 m.a.s.l.. The unpublished results of SEB distributed on the glacier 15a allowed to analyze the physical processes that to generate the spatio-temporal variation of the ablation.The DBES results show that the model applied in a former research was based on wrong assumptions and inaccurate. The results of DBES and PDD models show that the accumulated precipitation measured at the bottom of the glacier are greatly underestimated. It implies the need for new protocols for improving the precipitation measurement. Both the former research and SEB confirm the crucial role of shortwave radiation and albedo values over glacier melt. It puts also in evidence that snow metamorphism in the study zone occurs faster than other latitudes, associated with accelerated degradation of surface albedo. Finally, this research explains the reason why the PDD model provides good quality results in this region, which contradicted the results of previous study. This research contributes a better understanding about sensibility of SMB with temperature variations in the inner tropics.

Le changement climatique global participe à la redistribution des ressources en eau à l’échelle de la planète. Les impacts sont particulièrement notables en région de montagne où les gradients de précipitations et de températures sont très importants. Dans les régions tropicales de hautes altitudes, au sein desquelles la variabilité spatiale des précipitations est forte, l’estimation de cette variable sur une large gamme d’échelles d’intérêt pour les climatologues, météorologues et hydrologues est un défi. La modélisation glacio-hydrologique a pour but de comprendre les processus observés actuellement pour avoir la capacité de donner des réponses sur les évolutions possibles des écoulements qui seront causés par un changement climatique et une évolution des activités anthropiques. L’enjeu d’une telle modélisation, en zone de montagne tropicale, est de reproduire non seulement l’évolution saisonnière du débit, mais aussi l’évolution annuelle et pluri-annuelle des zones englacées. Cependant, ce type d’étude reste limité par la régionalisation des variables météorologiques. Durant l’année hydrologique 2012/2013, trois types de produits de précipitations (mesures de terrain, données satellitaires, sorties de modèle climatique régional (MCR)) sont tout d’abord comparés sur le bassin versant du Rio Santa au Pérou, d’un superficie de 10 400 km², dont 3,3 % est englacée, puis l’impact de ces différentes variables de forçage sur les sorties d’un modèle glacio-hydrologiques semi-distribué est évalué.Le produit satellite est celui de TRMM3B42; les sorties du MCR sont celles de WRF, à trois résolutions imbriquées : 27, 9 et 3 km ; les données journalières de précipitations sont interpolées en utilisant un algorithme de krigeage avec dérive externe (KED), en appliquant l’altitude ou bien les cumuls annuels des sorties de WRF comme dérive externe. Les sorties de WRF surestiment fortement les cumuls annuels, comparées aux produits TRMM ou KED. La physique du modèle permet cependant d’avoir un produit qui restitue correctement les cycles temporels saisonnier et journalier des précipitations. Le produit satellitaire TRMM montre des erreurs dans la saisonnalité des précipitations au niveau des pixels englacés, et une surestimation des cumuls en saison sèche, et en fait un produit qui n’est pas représentatif des champs de précipitations en zone de montagne. Les produits de krigeage sont de bons interpolateurs des précipitations à une résolution de 3 km. L’utilisation de l’altitude comme dérive, entraîne une augmentation des précipitations jusqu’aux plus hauts sommets de la Cordillère Blanche ; l’utilisation des cumuls de WRF bénéficie de la prise en compte de la physique atmosphérique du modèle pour correctement représenter les précipitations orographiques. Malgré une forte surestimation des volumes précipités, la modélisation climatique fournit, pour des zones de topographie complexe peu échantillonnées, des informations essentielles sur la temporalité et la distribution spatiale des précipitations. Cependant, les mesures in situ restent essentielles pour estimer les précipitations en termes de quantités et développer des méthodes d’interpolation ou de correction des sorties de modèle atmosphérique.La modélisation glacio-hydrologique est réalisée avec l’outil DHSVM-GDM, incluant une représentation des zones englacées et de la dynamique glaciaire. La force de cet outil est de pouvoir évaluer l’ensemble des composantes du bilan en eau, à différents pas de temps. Cependant, il reste difficile de représenter correctement à la fois les zones glaciaires et les zones non englacés sur un bassin versant tel que celui du Rio Santa. L’utilisation de différentes variables de forçage montre qu’une étude approfondie sur la variable de précipitations est nécessaire en amont de tout modélisation glacio-hydrologique pour simuler les zones de hautes altitudes, impactant les résultats de la modélisation en terme de perte en volume glaciaire. / Global climate change participates in the redistribution of water resources at the global scale. Impacts are particularly important in mountainous regions where precipitation and temperature gradients are important. In high altitude tropical regions, where precipitation spatial variability is significant, the estimation of precipitation over the broad range of scales of interest for climatologists, meteorologists and hydrologists is challenging. Glacio-hydrological modeling aims at understanding current processes in order to have the ability to give answer on possible evolutions of stream flow that will be caused by climatic changes and the development of anthropogenic activities. The issue of such modeling, in a tropical mountain area, is to reproduce not only the evolution of seasonal flow, but the annual and multi-annual evolution of glaciated areas. However, this type of study is limited by the regionalization of meteorological variables. During the hydrological year 2012/2013, three types of precipitation products (in situ data, satellite data, outputs from a regional climate model (RCM)) are first compared over the Rio Santa watershed in Peru, with an area of 10400 km² which 3.3 % is glaciated, then the impacts of those forcing variables on the outputs of a semi-distributed glacio-hydrological model are evaluated.The satellite product is TRMM 3B42 ; RCM outputs are obtained from WRF at three nested resolutions: 27, 9 and 3 km; daily in situ data are interpolated using a kriging with external drift (KED) algorithm, with the altitude or WRF annual amounts as the external drift. WRF outputs largely overestimate the annual totals, compared to TRMM or KED. However, the physics of the model allows to accurately reproduce the seasonal and daily time cycles of precipitation. TRMM performs poorly over ice covered surfaces and overestimate monthly value during the dry season, making it non representative of precipitation patterns in mountainous areas. Kriging products are good interpolators of precipitation at 3 km resolution. Using altitude as the external drift results in an increase of precipitation to the highest peaks of the Cordillera Blanca; using annual totals of WRF benefits from the atmospheric physic of the model to correctly represent orographic precipitation. Despite a strong overestimation of precipitation quantities, climate modeling provides, for sub-sampled complex topographic area, essential information on the temporal and spatial distribution of precipitation. However, in situ measurements remain essential to estimate precipitation in terms of quantities, and develop interpolation or correction methods of atmospheric model outputs.Glacio-hydrological modeling is performed with the DHSVM-GDM model, including the simulation of glaciated areas and of the glacier dynamic. The strength of such model is to assess the overall element of the water balance, at different time steps. However, it remains difficult to properly represent both the glaciated and non-glaciated areas on a watershed such as the Rio Santa. Using various forcing variables show that a comprehensive study on precipitation is needed before any glacio-hydrological modeling to simulate high altitudes area, impacting the modeling results in terms of ice volume loss.