Les glaciers intègrent naturellement des informations hydrologiques et énergétiques dans des régions climatiques variées du globe et sont sensibles à des changements minimes de bilan d’énergie de surface. Parmi les paramètres mesurés in-situ traduisant les variations climatiques, le bilan de masse de surface est la variable directement reliée aux conditions atmosphériques qui contrôlent les processus d’accumulation et d’ablation. Un grand nombre de modèles, de complexités diverses, existent pour simuler l’évolution du bilan de masse. Cependant, des questions perdurent quant à l’utilisation de l’approche la plus appropriée, notamment sur de longues périodes de temps. Basés sur de longues séries de mesures réalisées sur quatre glaciers des Alpes françaises (issues du service d’observation GLACIOCLIM), ces travaux de thèse ont pour but d’étudier la sensibilité du bilan de masse aux variables atmosphériques et topographiques, afin d’identifier celles essentielles pour contraindre les modèles.En s’appuyant tout d’abord sur une approche empirique, nos résultats indiquent que pour chaque glacier, la variabilité temporelle de l’ablation s’explique principalement par la variabilité de la température tandis que la variabilité spatiale de la fonte de la neige et de la glace est largement gouvernée par l’insolation. Néanmoins, à l’échelle du glacier, considérer l’insolation en plus de la température ne conduit pas à une amélioration de la performance des modèles empiriques et un modèle type degré-jour apparaît suffisant pour simuler son bilan de masse. L’utilisation du modèle à base physique SURFEX/ISBA-Crocus a permis d’étudier la sensibilité du bilan de masse estival à chacune des variables du bilan d’énergie de surface. Le modèle s’avère très performant pour la simulation des bilans spatialisés, mais nécessite un forçage météorologique précis et distribué spatialement. En particulier, le vent, souvent complexe à représenter dans les modèles atmosphériques,apparaît comme un facteur prépondérant pour modéliser correctement les processus de fonte. Avec des forçages météorologiques bien contraints, ces deux approches (empirique et physique) se révèlent très performantes quant à la modélisation du bilan de masse estival. Toutefois, nos résultats montrent que les bilans annuels restent très conditionnés par les bilans hivernaux du fait de la rétroaction de l’albédo de surface. Une attention particulière a donc été portée sur la simulation de l’accumulation.A partir des données de l’observatoire GLACIOCLIM, notre étude n’a pas permis d’établir de relation significative entre la variabilité spatio-temporelle des bilans hivernaux et les variables morpho-topographiques.Pour cette raison, une campagne spécifique basée sur des mesures LiDAR a été mise en œuvre au Col du Midi (Massif du Mont-Blanc) afin d’affiner le réseau de mesures et d’améliorer la compréhension de la variabilité spatio-temporelle de la couverture neigeuse. Nos résultats montrent la complexité d’obtenir des mesures d’accumulation par différence de MNT, du fait des vitesses d’émergence/submergence. / Glaciers naturally record hydrological and energetic information into varied climatic regions of the world, and are sensitive to small changes in their surface energy balance. The glacier surface mass balance is directly linked to the atmospheric variables which control accumulation and ablation processes. Many models, with different complexities, enable simulations of mass balance evolution, but some questions arise about the best approach to use, especially over long time periods. Based on long measurement series on four glaciers in the French Alps (GLACIOCLIM observatory), this thesis aims at studying the mass balance sensitivity to climatic variations, to identify relevant meteorological variables for modelling. Based on an empirical approach performed on each glacier, temperature was found to be the main driver of temporal ice/snow ablation variability while solar radiations strongly influence the spatial distribution of summer mass balance. However, at glacier scale, to include solar radiation in melt models does not improvethe performances and a classical degree-day model is sufficient to simulate glacier-wide mass balance.The snowpack model SURFEX-ISBA-Crocus was then used to study summer mass balance sensitivity to all surface energy balance fluxes. Results indicate good performances of the model to simulate mass balance at each stake, but it requires accurate meteorological forcing. In particular wind, generally difficult to assessby atmospheric models, appears as a key factor for an accurate ablation modelling. Our work emphasizes that both physical and empirical approaches are very efficient for ablation modeling when forced with accurate meteorological data. Yet, annual mass balances remain very sensitive to wintermass balance due to the surface albedo feedback. For this reason, simulating accumulation processes remainsas important as ablation modelling. Using GLACIOCLIM data measurements, our study failed to find relevant relationship between winter mass balance and topographical variables such as slope, aspect, elevation, due to insufficient spatio-temporal resolution. To fill this gap of measurements, we set up a dedicated field campaign to improve our understanding of the spatio-temporal variability of the snow depth distribution on Alpineglaciers. This year-round campaign was performed using LiDAR acquisitions at Col du Midi (Mont Blancmassif). Results underline the complexity of accurate accumulation measurements from DEM differences due to submergence/emergence velocities.

S’appuyant sur de longues missions de terrain, ce travail offre le regard quantitatif et qualitatif d’ungéographe sur la dynamique nivo-glaciaire en région polaire dans un contexte de changement climatiqueactuel. Il s’inscrit dans le cadre du programme ANR Hydro-Sensor-FLOWS mis en oeuvre dansla mouvance de l’Année Polaire Internationale. En amont de ce programme pluridisciplinaire visant àcomprendre le fonctionnement hydrologique d’un petit hydrosystème polaire il fallait adjoindre uneétude plus spécifique concernant le rôle de la neige et de la glace dans l’équation hydrologique : c’estl’objet de cette thèse.Le bassin glaciaire de l’Austre Lovénbreen (10 km2, Spitsberg – 79°N) a servi de terrain d’étude dansune approche géographique à échelle locale. De nombreuses méthodes ont été testées tant sur leterrain (stations photo automatiques, carottages de la neige, DGPS et GPR) que pour l’exploitationdes données acquises.Au total, ce travail suit la dynamique nivo-glaciaire pendant trois années (2008, 2009 et 2010) à partirde 15 séries de carottages en 40 points du bassin, de mesures de densité, de profils nivologiques,de bilan de masse et de suivi continu, en 20 points, de la température de l’air. Outre l’image satellite,ce travail s’appuie également sur une collection de 3 clichés journaliers acquis en 10 points dubassin. Des méthodes ont été développées pour corriger géométriquement les photos afin d’en fairedes documents en projection verticale dont il est possible de tirer de l’information quantitative. Lesoutils de la géomatique ont permis de spatialiser l’information ponctuelle et d’établir à des pas detemps variables, par croisement des données acquises in situ, les diverses lames entrant dansl’équation hydrologique (précipitations, fonte potentielle de la neige et de la glace en relation avec leslames écoulées). Les trois années consécutives de suivi offrent des conditions très différentes quirévèlent le rôle constant et modérateur des températures et celui beaucoup plus impulsif des précipitations.Ce travail pose également le problème des mesures et de leur fiabilité dans un milieu aussi sévère etaussi impétueux que l’Arctique. / Leaning on long field campaigns, this thesis offer both qualitative and quantitative overview of polarregion nivo-glaciologic dynamics through a geographic approach, in context of climate change. Thiswork is part of the Hydro-sensor FlOWS program taking place during the International Polar Year.Upstream of this pluridisciplinary program aimed at understanding the hydrological processes of asmall polar hydrosystem, it was necessary to add a more specific study on the role of snow and icein the hydrological equation : this is the subject of this PhD.The glacial basin of Austre Lovénbreen (10 km2, Spitsberg – 79°N) served as a field of study in ageographical approach at local scale. Many methods have been tested both in the field (automaticphoto station, snow coring, DGPS and GPR) and for the data processing.Overall, this work follows the snow and ice dynamics for three years (2008, 2009 and 2010) from 15sets cores of 40 points of the basin, density measurements, nivologic profiles, mass balance and thecontinuous monitoring, on 20 points, of the air temperature. Besides the satellite images, this workalso leans on a collection of 3 daily photos acquired in 10 points of the basin. Methods were developedto geometrically correct photos in order to provide data wich could be used as a vertical projectioninto a GIS software to get quantitative informations. The geomatrical tools gave the possibility tohave accurate spatial representation at different time scale, by crossing data acquired in situ, lesdiverses lames entrant dans l’équation hydrologique (precipitation, potential melting of snow and icein relation to the lames passed).Three consecutive years of monitoring provide very different conditions that reveal temperatures’sconstant role of moderator, and the much more impulsive behavior of precipitations.This work also raises the problem of measurements and their reliability in a binding environment asthe Arctic

(RS) e Santa Catarina (SC), Brasil meridional. Para tanto foram usados os dados diários e das normais climatológicas (1961-90) das estações do Instituto Nacional de Meteorologia (INMET). Foi elaborada uma caracterização climática da área de estudo, que forneceu o apoio para a contextualização do fenômeno. Os dados climatológicos básicos (temperaturas médias mensais compensadas, temperaturas mínimas absolutas e totais pluviométricos mensais) foram obtidos diretamente das normais climatológicas publicadas pelo INMET. Os dados referentes aos dias de neve foram compilados diretamente das observações do INMET para planilha eletrônica onde se totalizaram os dias de neve e foram calculadas médias e percentuais convencionais. Em termos sazonais, o período de ocorrência estende-se de maio até setembro, quando a massa de ar polar marítima (mP) tem maior atuação no sul do Brasil. O mês de julho concentra a maioria dos eventos, com mais de 40% dos dias de neve na maioria das estações analisadas. A trajetória do anticiclone móvel polar tem papel fundamental no entendimento do fenômeno, associando-se a maioria dos eventos com passagens interiores (i.e., continentais) do referido sistema atmosférico. As áreas mais altas do Brasil meridional, a partir da cota de 600 m, são as que apresentam as maiores médias anuais de dias com neve, com valores que chegam até a média de 1,8 dias/ano em Cambará do Sul e 2,7 dias/ano em São Joaquim (SC). Esse município possui as mais altas médias de ocorrência da neve. A área de maior recorrência do fenômeno foi chamada de “Planalto da Neve”, subdividida em dois setores: I, acima de 900 m e II, entre 600 m e 900 m. O referido Planalto possui 95.242 km² distribuídos no RS e SC, localizando-se no seu interior 90 cidades gaúchas e 101 catarinenses. A análise da variação inter-anual (com dados de São Joaquim, SC) indicou que, de toda a série estudada, apenas os anos de 1961, 1963, 1971, 1973, 1982 e 1986, não apresentam registro de neve. As nevadas mais relevantes ocorreram nos anos de 1965, 1975, 1988 e 1990. Os intervalos entre períodos de ausência de neve e de nevadas mais importantes indica um ciclo decadal na dinâmica do fenômeno. Apesar dos picos de precipitação em 1988 e 1990, não existe uma tendência de aumento da precipitação de neve naquela cidade. / This work analyzes the seasonal and spatial distribution of snow in Rio Grande do Sul (RS) and Santa Catarina (SC), southern Brazil. This analysis is based on the daily data and the Climatological Normals (1961-1990), coming from the local weather stations, maintained by the Instituto Nacional de Meteorologia (INMET), to obtain the basic climatological data (compensated monthly medium temperatures, minimum absolute temperatures and monthly pluviometric totals). To contextualize the study of this phenomenon, the characterization of the climate was prepared for the region. The total days of snow, averages and conventional percentages were compiled directly from INMET observations. In seasonal terms, the occurrence period extends from May to September, when the marine polar air mass (mP) has greater influence in the south of Brazil. The month of July concentrated most of the events, with more than 40% of the days of snow in most of the analyzed stations. The polar anticyclone has a fundamental role in the understanding this phenomenon, associating most of the events with continental passages of this atmospheric system. The highlands of southern Brazil, starting at 600 m, present the largest annual averages of days with snow, values averaging from 1.8 day/year in Cambará do Sul (RS) to 2.7 day/year in São Joaquim (SC), the highest averages of snow occurrence. The area of occurrence of the phenomenon is called “Planalto da Neve”, and is subdivided in two sectors: 1) above 900 m and, 2) between 600 m and 900 m. This plateau possesses 95.242 km² and is spread out between RS and SC, comprising 90 municipalities of RS and 101 in SC. The analysis of the interannual variation considering the data from São Joaquim, SC ) indicated that, from the studied series, 1961, 1963, 1971, 1973, 1982 and 1986, did not register snow. The most relevant snowfalls happened in the years of 1965, 1975, 1988 and 1990. The intervals between periods with absence of snow and of more important snowfalls indicates a decadal cycle in the dynamics of the phenomenon. Regardless of the peaks in precipitation registered for 1988 and 1990, there is no trend in the increase of snow precipitation for São Joaquim.

(RS) e Santa Catarina (SC), Brasil meridional. Para tanto foram usados os dados diários e das normais climatológicas (1961-90) das estações do Instituto Nacional de Meteorologia (INMET). Foi elaborada uma caracterização climática da área de estudo, que forneceu o apoio para a contextualização do fenômeno. Os dados climatológicos básicos (temperaturas médias mensais compensadas, temperaturas mínimas absolutas e totais pluviométricos mensais) foram obtidos diretamente das normais climatológicas publicadas pelo INMET. Os dados referentes aos dias de neve foram compilados diretamente das observações do INMET para planilha eletrônica onde se totalizaram os dias de neve e foram calculadas médias e percentuais convencionais. Em termos sazonais, o período de ocorrência estende-se de maio até setembro, quando a massa de ar polar marítima (mP) tem maior atuação no sul do Brasil. O mês de julho concentra a maioria dos eventos, com mais de 40% dos dias de neve na maioria das estações analisadas. A trajetória do anticiclone móvel polar tem papel fundamental no entendimento do fenômeno, associando-se a maioria dos eventos com passagens interiores (i.e., continentais) do referido sistema atmosférico. As áreas mais altas do Brasil meridional, a partir da cota de 600 m, são as que apresentam as maiores médias anuais de dias com neve, com valores que chegam até a média de 1,8 dias/ano em Cambará do Sul e 2,7 dias/ano em São Joaquim (SC). Esse município possui as mais altas médias de ocorrência da neve. A área de maior recorrência do fenômeno foi chamada de “Planalto da Neve”, subdividida em dois setores: I, acima de 900 m e II, entre 600 m e 900 m. O referido Planalto possui 95.242 km² distribuídos no RS e SC, localizando-se no seu interior 90 cidades gaúchas e 101 catarinenses. A análise da variação inter-anual (com dados de São Joaquim, SC) indicou que, de toda a série estudada, apenas os anos de 1961, 1963, 1971, 1973, 1982 e 1986, não apresentam registro de neve. As nevadas mais relevantes ocorreram nos anos de 1965, 1975, 1988 e 1990. Os intervalos entre períodos de ausência de neve e de nevadas mais importantes indica um ciclo decadal na dinâmica do fenômeno. Apesar dos picos de precipitação em 1988 e 1990, não existe uma tendência de aumento da precipitação de neve naquela cidade. / This work analyzes the seasonal and spatial distribution of snow in Rio Grande do Sul (RS) and Santa Catarina (SC), southern Brazil. This analysis is based on the daily data and the Climatological Normals (1961-1990), coming from the local weather stations, maintained by the Instituto Nacional de Meteorologia (INMET), to obtain the basic climatological data (compensated monthly medium temperatures, minimum absolute temperatures and monthly pluviometric totals). To contextualize the study of this phenomenon, the characterization of the climate was prepared for the region. The total days of snow, averages and conventional percentages were compiled directly from INMET observations. In seasonal terms, the occurrence period extends from May to September, when the marine polar air mass (mP) has greater influence in the south of Brazil. The month of July concentrated most of the events, with more than 40% of the days of snow in most of the analyzed stations. The polar anticyclone has a fundamental role in the understanding this phenomenon, associating most of the events with continental passages of this atmospheric system. The highlands of southern Brazil, starting at 600 m, present the largest annual averages of days with snow, values averaging from 1.8 day/year in Cambará do Sul (RS) to 2.7 day/year in São Joaquim (SC), the highest averages of snow occurrence. The area of occurrence of the phenomenon is called “Planalto da Neve”, and is subdivided in two sectors: 1) above 900 m and, 2) between 600 m and 900 m. This plateau possesses 95.242 km² and is spread out between RS and SC, comprising 90 municipalities of RS and 101 in SC. The analysis of the interannual variation considering the data from São Joaquim, SC ) indicated that, from the studied series, 1961, 1963, 1971, 1973, 1982 and 1986, did not register snow. The most relevant snowfalls happened in the years of 1965, 1975, 1988 and 1990. The intervals between periods with absence of snow and of more important snowfalls indicates a decadal cycle in the dynamics of the phenomenon. Regardless of the peaks in precipitation registered for 1988 and 1990, there is no trend in the increase of snow precipitation for São Joaquim.

(RS) e Santa Catarina (SC), Brasil meridional. Para tanto foram usados os dados diários e das normais climatológicas (1961-90) das estações do Instituto Nacional de Meteorologia (INMET). Foi elaborada uma caracterização climática da área de estudo, que forneceu o apoio para a contextualização do fenômeno. Os dados climatológicos básicos (temperaturas médias mensais compensadas, temperaturas mínimas absolutas e totais pluviométricos mensais) foram obtidos diretamente das normais climatológicas publicadas pelo INMET. Os dados referentes aos dias de neve foram compilados diretamente das observações do INMET para planilha eletrônica onde se totalizaram os dias de neve e foram calculadas médias e percentuais convencionais. Em termos sazonais, o período de ocorrência estende-se de maio até setembro, quando a massa de ar polar marítima (mP) tem maior atuação no sul do Brasil. O mês de julho concentra a maioria dos eventos, com mais de 40% dos dias de neve na maioria das estações analisadas. A trajetória do anticiclone móvel polar tem papel fundamental no entendimento do fenômeno, associando-se a maioria dos eventos com passagens interiores (i.e., continentais) do referido sistema atmosférico. As áreas mais altas do Brasil meridional, a partir da cota de 600 m, são as que apresentam as maiores médias anuais de dias com neve, com valores que chegam até a média de 1,8 dias/ano em Cambará do Sul e 2,7 dias/ano em São Joaquim (SC). Esse município possui as mais altas médias de ocorrência da neve. A área de maior recorrência do fenômeno foi chamada de “Planalto da Neve”, subdividida em dois setores: I, acima de 900 m e II, entre 600 m e 900 m. O referido Planalto possui 95.242 km² distribuídos no RS e SC, localizando-se no seu interior 90 cidades gaúchas e 101 catarinenses. A análise da variação inter-anual (com dados de São Joaquim, SC) indicou que, de toda a série estudada, apenas os anos de 1961, 1963, 1971, 1973, 1982 e 1986, não apresentam registro de neve. As nevadas mais relevantes ocorreram nos anos de 1965, 1975, 1988 e 1990. Os intervalos entre períodos de ausência de neve e de nevadas mais importantes indica um ciclo decadal na dinâmica do fenômeno. Apesar dos picos de precipitação em 1988 e 1990, não existe uma tendência de aumento da precipitação de neve naquela cidade. / This work analyzes the seasonal and spatial distribution of snow in Rio Grande do Sul (RS) and Santa Catarina (SC), southern Brazil. This analysis is based on the daily data and the Climatological Normals (1961-1990), coming from the local weather stations, maintained by the Instituto Nacional de Meteorologia (INMET), to obtain the basic climatological data (compensated monthly medium temperatures, minimum absolute temperatures and monthly pluviometric totals). To contextualize the study of this phenomenon, the characterization of the climate was prepared for the region. The total days of snow, averages and conventional percentages were compiled directly from INMET observations. In seasonal terms, the occurrence period extends from May to September, when the marine polar air mass (mP) has greater influence in the south of Brazil. The month of July concentrated most of the events, with more than 40% of the days of snow in most of the analyzed stations. The polar anticyclone has a fundamental role in the understanding this phenomenon, associating most of the events with continental passages of this atmospheric system. The highlands of southern Brazil, starting at 600 m, present the largest annual averages of days with snow, values averaging from 1.8 day/year in Cambará do Sul (RS) to 2.7 day/year in São Joaquim (SC), the highest averages of snow occurrence. The area of occurrence of the phenomenon is called “Planalto da Neve”, and is subdivided in two sectors: 1) above 900 m and, 2) between 600 m and 900 m. This plateau possesses 95.242 km² and is spread out between RS and SC, comprising 90 municipalities of RS and 101 in SC. The analysis of the interannual variation considering the data from São Joaquim, SC ) indicated that, from the studied series, 1961, 1963, 1971, 1973, 1982 and 1986, did not register snow. The most relevant snowfalls happened in the years of 1965, 1975, 1988 and 1990. The intervals between periods with absence of snow and of more important snowfalls indicates a decadal cycle in the dynamics of the phenomenon. Regardless of the peaks in precipitation registered for 1988 and 1990, there is no trend in the increase of snow precipitation for São Joaquim.