A stylised analytical framework is used to show how the global carbon tax and the amount of untapped fossil fuel can be calculated from a simple rule given estimates of society's rate of time impatience and intergenerational inequality aversion, the extraction cost technology, the rate of technical progress in renewable energy and the future trend rate of economic growth. The predictions of the simple framework are tested in a calibrated numerical and more complex version of the integrated assessment model (IAM). This IAM makes use of the Oxford carbon cycle of Allen et al. (2009), which differs from DICE, FUND and PAGE in that cumulative emissions are the key driving force of changes in temperature. We highlight the importance of the speed and direction of technological change for the energy transition and how time impatience, intergenerational inequality aversion and expected trend growth affect the time paths of the optimal global carbon tax and the optimal amount of fossil fuel reserves to leave untapped. We also compare these with the adverse global warming trajectories that occur if no policy actions are taken.

Les forêts de mangrove contribuent de manière significative au flux d'énergie, au cycle des nutriments et du carbone dans l'océan côtier, étant un puits pour le CO2 atmosphérique. Les forêts de mangroves sont très productives et stockent une quantité élevée de carbone à la fois dans leurs sols et dans leur biomasse.Au cours de la décomposition de la litière, les nutriments et le carbone peuvent être recyclés ou exportés vers les écosystèmes adjacents par l'action des marées. La mangrove de Can Gio (Ho Chi Minh Ville, Vietnam), dégradée par l'épandage de défoliants pendant la guerre du Vietnam, a pu être restaurée grâce à la replantation et à la régénération naturelle. À ce jour, la forêt de mangrove de Can Gio est la plus grande forêt de mangrove contiguë au Vietnam, et est devenue la première réserve biosphère UNESCO dans ce pays. L'objectif principal de cette thèse était de caractériser le cycle du carbone dans la forêt de mangrove tropicale.Les résultats de cette thèse ont permis de:- Développer des équations allométriques permettant d’estimer la biomasse aérienne de la forêt plantée de mangroves de l’espèceRhizophora apiculatadans le sud du Vietnam;- Calculer les stocks de carbone totaux dans différents peuplements de mangrove se développant sous le climat tropical du sud du Vietnam;- Caractériser les taux de décomposition de la litière et évaluer la dynamique des nutriments et des métaux traces au cours des processus de dégradation de la litière, ainsi que l'évolution de δ13C pendant la décomposition;- Déterminer la variabilité saisonnière des flux de CO2 à différentes interfaces: sol-air, eau-air et tronc-air, et caractériser les profils de concentration en CO2 dans la canopée. / Mangrove forests significantly contribute to energy flow, nutrient and carbon cycling in the coastal ocean, being a sink for atmospheric CO2. Mangroves forests are highly productive and store high amount of carbon both in their soils and in their biomass. During leaf litter decomposition, nutrients and carbon can be recycled or exported to adjacent ecosystems by the tidal action. Can Gio mangrove, degraded by the spraying of defoliants during the Vietnam War, successfully recovered through replantation and natural regeneration after 40 years. To date, the Can Gio mangrove forest is the largest contiguous mangrove forest in Vietnam, and became the first Mangrove Biosphere Reserve in this country. The main objective of this PhD thesis was to characterize carbon cycling within the Can Gio mangrove forest, which is a tropical one.The results of this PhD thesis allowed to: - Develop allometric equations and to estimate the aboveground biomass of Rhizophora apiculata Blume planted mangroves forest in Southern Vietnam; - Calculate the total carbon stocks in different mangrove stands developing under the tropical climate of Southern Vietnam; - Characterize the leaf litter decomposition rates, and assess nutrients and trace metals dynamics during litter decay processes, as well as the evolution of δ13C during decay; - Determine the seasonal variability CO2 fluxes at different interfaces: soil-air, water-air and trunk-air, and to characterize CO2 concentrations profiles in the canopy.

Estudos recentes têm demonstrado resultados surpreendentes acerca do ciclo do carbono nos ambientes aquáticos da Amazônia, com fluxos evasivos de CO2 a partir dos mesmos superando àqueles que ocorrem na descarga para o oceano. Porém, os processos que controlam estes fluxos permanecem como uma grande fonte de incerteza nas medidas dos mesmos e, conseqüentemente, nos balanços regionais de carbono. Desta forma, os objetivos deste estudo foram avaliar a variabilidade espacial e temporal (sazonal e circadiana) do fluxo de CO2 em diferentes sistemas aquáticos da Amazônia e os principais parâmetros físicos envolvidos no controle do coeficiente de troca gasosa (k600). Para isto utilizamos os dados, disponibilizados pela Rede Beija-Rio, de fluxo de CO2 (medido com câmara flutuante) e demais parâmetros, medidos em sete rios da bacia Amazônica, ao longo dos anos de 2007 a 2009. Fizemos também campanhas intensivas de medidas de fluxo e, em duas destas campanhas, realizamos medidas de fluxo utilizando simultaneamente as técnicas de câmara e de covariância dos vórtices turbulentos (eddy correlation). Os resultados mostraram que o ciclo sazonal de fluxo de CO2, com os maiores valores sendo observados no período de cheia, está diretamente relacionado ao ciclo sazonal da pCO2 que, por sua vez, é paralela à hidrógrafa. A variabilidade circadiana está relacionada às mudanças na velocidade do vento em ambientes lóticos e, em ambientes lênticos estratificados, além da velocidade de vento a mistura convectiva desempenha também um papel importante neste controle. Rios de águas pretas apresentaram os maiores fluxos, enquanto que, rios de águas claras como o Araguaia e Teles Pires, apresentaram fluxos negativos no período de seca. A produção primária em lagos e rios de águas claras desempenha um papel importante na variabilidade circadiana e sazonal da pCO2 e, consequentemente, no fluxo de CO2. As análises do k600 mostraram diferentes condições de turbulência na calha principal do rio Javaés e nas áreas mais extensas e rasas ao longo do Rio Negro, gerando modelos para a estimativa de k600 com a velocidade do vento no rio Javaés, e com a velocidade do vento e a relação entre velocidade da corrente e profundidade do canal [(w/z)0,5] no Rio Negro / Recent studies have shown surprising results on the carbon cycle in aquatic environments of the Amazon, with evasive fluxes of CO2 larger than the carbon exported by discharge into the 2ocean. However, the processes that control these fluxes remain a major source of uncertainty in upscaling and, consequently, on regional carbon balances. Thus, the objective of this study was to evaluate the spatial and temporal variability (seasonal and circadian) of CO2 flux in different aquatic systems in the Amazon basin and the main physical parameters involved in controlling the gas exchange coefficient (K600). For this we used the CO2 flux data from Rede Beija-Rio (measured with floating chamber) and other parameters measured in seven rivers of the Amazon basin, over the years 2007 to 2009. Additionally, we also conducted intensive flux measurements campaigns, in two of which we compared chamber and eddy covariance (eddy correlation) approaches. The results showed that the seasonal cycle of CO2 flux, which is higher during the high water period, is directly related to the seasonal cycle of pCO2 which, in turn, parallels the hydrograph. Circadian variations are related to changes in wind speed in lotic environments. At stratified lentic environments both wind speed the convective mixing play important roles in this control. Black-water rivers showed the highest fluxes, while clear water rivers Araguaia and Teles Pires showed negative fluxes during the dry period. Primary production in lakes and clear water rivers plays a more important role than previously observed in circadian and seasonal variability of pCO2 and hence in CO2 fluxes. Analyses of K600 showed different turbulent conditions in the main channel of the Javaés river and in longer and shallower areas along the Rio Negro, with significant models for estimation of K600 based on wind velocity in the Javaés river and on wind speed and the relationship between water current and depth [(w / z) 0.5 ] in the Negro river

Peat soils are known to be a significant emitter of atmospheric greenhouse gasses. However, the spatial and temporal variability in production and release of greenhouse gases (such as methane) in peat soils remains uncertain, particularly for low-latitude peatlands like the Florida Everglades, as the majority of studies on gas dynamics in peatlands focus on northern peatlands. The purpose of the work outlined here is focused on understanding the spatial and temporal variability in biogenic gas dynamics (i.e. production and release of methane and carbon dioxide) by implementing various experiments in the Florida Everglades at different scales of measurement, using noninvasive hydrogeophysical methods. Non-invasive methods include ground-penetrating radar (GPR), gas traps, time-lapse cameras, and hydrostatic pressure head measurements, that were constrained with direct measurements on soil cores like porosity, and gas composition using gas chromatography. By utilizing the measurements of in-situ gas volumes, we are able to estimate gas production using a mass balance approach, explore spatial and temporal variabilities of gas dynamics, and better constrain gas ebullition models. A better understanding of the spatial and temporal variability in gas production and release in peat soils from the Everglades has implications regarding the role of subtropical wetlands in the global carbon cycle, and can help providing better production and flux estimates to help global climate researchers improve their predictions and models for climate change. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2018. / FAU Electronic Theses and Dissertations Collection

Fire is a natural process that has existed on our planet for more than ~350 million years, and is a process that continues to influence our everyday lives. On Earth, a relationship exists between the process of combustion and the natural functioning of the Earth system. Here, the process of combustion has been implicated in playing an essential role for life on Earth, where natural Earth system processes have been shown to influence ignition probability, fire spread and fire behaviour, and where fire can provide a variety of feedbacks, to the Earth system over different timescales. Over medium timescales of decades to hundreds of thousands of years, the likelihood and behaviour of fires are controlled by regional climate changes and vegetation type, whilst the occurrence of fire can play a crucial role in influencing biome persistence and development. Over long timescales (hundreds of thousands to multi-million year), the components influencing the probability of fire and fire behaviour not only involve processes occurring over local and regional spatial scales, and over short and medium timescales, but also long term processes occurring globally, such as changes in atmospheric oxygen concentration and the evolution of vegetation. Across these timescales in Earth’s past, combustion has been shown to impact global ecosystems, climate and the carbon cycle by generating feedbacks that influence Earth’s biogeochemical cycles. However, it is clear that our understanding of the role that fire plays in the Earth system, although improving is still developing. This thesis provides an analysis of these Earth system - fire relationships and feedbacks across medium and long timescales in deep time, in order to understand the role that fire may have played and what the record of fire can tell us about the functioning and re-equilibrating of the Earth system during and after significant carbon-cycle perturbation events occurring in Earth’s deep past. The results presented in this thesis contribute what is believed to be the first fossil evidence that rising atmospheric oxygen and fire feedbacks may have aided in the termination of a significant carbon-cycle perturbation event, termed the ‘Toarcian oceanic anoxic event’ that occurred ~183 million years ago during the Jurassic period, and the return of the Earth system towards ‘background functioning’. This thesis also provides an analysis of the record of wildfire in the form of fossil charcoal across the initiation of an anoxic event that occurred ~93 million years ago, during the Cretaceous period. The results illustrate that CO2 - climate driven changes in wildfire activity can be observed across medium timescales even during times of significant carbon-cycle perturbations, and modelled high atmospheric oxygen concentrations. These results illustrate how hypothesized changes in the hydrological cycle, and likely moisture content of fuel, appear to be the dominant control on wildfire activity during this period. Finally, this thesis provides an analysis of charcoal abundance variations occurring across natural, orbitally forced cycles, termed the Milankovitch cycles. The results presented illustrate that natural variations in charcoal abundance are possible over intermediate timescales within the geological record. This thesis therefore illustrates a need to take into consideration and incorporate ‘natural background’ fluctuations in fire activity occurring over medium timescales, when analysing and predicting past and future climate change patterns.

Biomineralization refers to the production of mineralized tissues by organisms. The fine control which organisms can exert over this process produces crystals with morphologies and properties contrasting to that of non-biogenic crystals and specifically altered to suit the required functional need. A key model system of biomineralization are a unicellular marine algae, coccolithophores, which produce calcium carbonate scales known as coccoliths. These coccoliths are comprised of arrangements of single crystals of calcite interlocked to form a plate-shaped structure. Coccoliths are developed intracellularly in a specialised compartment called the coccolith vesicle, before being extruded to the cell surface. In this work, two vital components of the coccolith biomineralization process are investigated - a soluble polysaccharide thought to act as a habit modifier and an insoluble organic scaffold known as a baseplate that provides the surface for nucleation and growth of the crystals. Whilst both these elements are thought to play a key part in the biomineralization process, the role of each is not fully understood. To investigate the effect of coccolith-associated polysaccharides (CAPs) on nucleation and polymorph selection, two systems that promote different polymorphs of calcium carbonate were utilised. In both systems, the intracrystalline polysaccharide fraction extracted from one species, Gephyrocapsa oceanica, was able to promote calcite nucleation in vitro, even under conditions favouring the kinetically-privileged polymorphs of calcium carbonate: vaterite and aragonite. As this property is not observed with CAPs extracted from its 'sister species', Emiliania huxleyi, the in vivo function of CAPs may differ between the two species. Both cryo-transmission electron microscopy (cryoTEM) and scanning electron microscopy (SEM) were used to determine the mechanism of calcite growth in the presence of G. oceanica CAPs, showing its impact on the forming amorphous calcium carbonate (ACC), decreasing the size of the particles and producing irregular, angular particles. Using cryo-electron tomography (cryoET), it was possible to create a 3D representation of the structure of the baseplate from the coccolithophore Pleurochrysis carterae, revealing its two-sided organisation. Examination of several stages of the coccolith growth process demonstrated the interlocking nature of the calcite crystals that make up the coccolith and the progression of the crystal morphologies over time, and the interaction of these crystals with the baseplate rim. Additionally, the effect of inhibiting carbonic anhydrase (CA), an enzyme involved in the regulation of carbonate species, revealed that inhibition of CA can affect coccolithogenesis as well as cell proliferation.

The terrestrial CO2 exchange in the Arctic plays an important role in the global carbon (C) cycle. The Arctic ecosystems, containing a large amount of organic carbon (C), are experiencing ongoing warming in recent decades, which is affecting the C cycling and the feedback interactions between its different components. To improve our understanding of the atmosphere-ecosystem interactions, the Greenland Ecosystem Monitoring (GEM) program measures ecosystem CO2 exchange and links it to biogeochemical processes. However, this task remains challenging in northern latitudes due to an insufficient number of measurement sites, particularly covering full annual cycles, but also the frequent gaps in data affected by extreme conditions and remoteness. Combining ecosystem models and field observations we are able to study the underlying processes of Arctic CO2 exchange in changing environments. The overall aim of the research is to use data-model approaches to analyse the patterns of C exchange and their links to biological processes in Arctic ecosystems, studied in detail both from a measurement and a modelling perspective, but also from a local to a pan-arctic scale. In Paper I we found a compensatory response of photosynthesis (GPP) and ecosystem respiration (Reco), both highly sensitive to the meteorological drivers (i.e. temperatures and radiation) in Kobbefjord, West Greenland tundra. This tight relationship led to a relatively insensitive net ecosystem exchange (NEE) to the meteorology, despite the large variability in temperature and precipitations across growing seasons. This tundra ecosystem acted as a consistent sink of C (-30 g C m-2), except in 2011 (41 g C m-2), which was associated with a major pest outbreak. In Paper II we estimated this decrease of C sink strength of 118-144 g C m-2 in the anomalous year (2011), corresponding to 1210-1470 tonnes C at the Kobbefjord catchment scale. We concluded that the meteorological sensitivity of photosynthesis and respiration were similar, and hence compensatory, but we could not explain the causes. Therefore, in Paper III we used a calibrated and validated version of the Soil-Plant-Atmosphere model to explore full annual C cycles and detail the coupling between GPP and Reco. From this study we found two key results. First, similar metrological buffering to growing season reduced the full annual C sink strength by 60%. Second, plant traits control the compensatory effect observed (and estimated) between gross primary production and ecosystem respiration. Because a site-specific location is not representative of the entire Arctic, we further evaluated the pan-Arctic terrestrial C cycling using the CARDAMOM data assimilation system in Paper IV. Our estimates of C fluxes, pools and transit times are in good agreement with different sources of assimilated and independent data, both at pan-Arctic and local scale. Our benchmarking analysis with extensively used Global Vegetation Models (GVM) highlights that GVM modellers need to focus on the vegetation C dynamics, but also the respiratory losses, to improve our understanding of internal C cycle dynamics in the Arctic. Data-model approaches generate novel outputs, allowing us to explore C cycling mechanisms and controls that otherwise would not have been possible to address individually. Also, discrepancies between data and models can provide information about knowledge gaps and ecological indicators not previously detected from field observations, emphasizing the unique synergy that models and data are capable of bringing together.

O principal papel dos rios no ciclo global biogeoquímico do carbono tem sido considerado como o de exportador fluvial de carbono orgânico total (COT) e carbono inorgânico dissolvido (CID) para os oceanos. Entretanto, estudos recentes mostraram a importância dos fluxos evasivos de CO2 a partir de rios da Amazônia, sugerindo que uma parte significativa do carbono fixado pela floresta retornaria para a atmosfera por esta via. A troca gasosa entre a atmosfera e as águas supersaturadas dos rios é função do gradiente de concentração do gás através da interface ar-água. A pCO2 da água é fortemente determinada pela concentração de carbono inorgânico dissolvido e pH, que, por sua vez, são função de processos físicos, químicos e biológicos. O foco principal deste estudo foi estabelecer a importância dos fluxos evasivos de CO2 nos balanços de C, uma vez que os sistemas aquáticos são uma fonte significativa de C para a atmosfera nos ambientes tropicais úmidos. Ao mesmo tempo, avaliou-se a influência das concentrações de CID e pH na evasão de CO2. A área de estudo foi a bacia do rio Ji-Paraná, Rondônia. Vários rios desta bacia foram amostrados entre maio/99 e maio/03. Concentrações de CID, temperatura e pH foram utilizadas para calcular a pCO2 da água, baseado nas equações de equilíbrio termodinâmico. Para estimar o fluxo evasivo de CO2 utilizou-se o modelo teórico de fluxo difusivo. Os resultados mostraram que rios que drenam áreas de solos mais férteis apresentam as maiores concentrações de CID e um maior potencial de evasão. Apesar da variabilidade sazonal da pCO2, as concentrações de CO2 dissolvido indicam que as trocas de CO2 com a atmosfera são unilaterais (evasão para a atmosfera) ao longo do ano. Durante o período de cheia, mesmo com concentrações de CID menores, a diminuição do pH observado neste período foi suficiente para promover potenciais de evasão maiores; neste período a evasão é cerca de 4 vezes maior que no período de seca. Baseado no modelo de fluxo difusivo, a evasão de CO2 para a atmosfera a partir das águas superficiais dos rios da bacia do Ji-Paraná foi estimada em 128 a 318 Gg C ano-1. A descarga fluvial exporta anualmente cerca de 370 Gg C ano-1. Portanto, a evasão é da mesma ordem de magnitude da descarga fluvial, demonstrando a importância desta via no ciclo do carbono nos sistemas aquáticos da Amazônia. / The major biogeochemical role of river systems in the global carbon cycle is considered to be the fluvial export of total organic carbon (TOC) and dissolved inorganic carbon (DIC) to the ocean. However, recent studies have shown the importance of CO2 outgassing from rivers of the Amazon, suggesting that a significant part of the carbon fixed by forests returns to the atmosphere through this pathway. Gas exchange between the atmosphere and river waters supersaturated in CO2 is a function of gaseous gradients across the air-water interface. Water pCO2 is strongly influenced by concentrations of dissolved inorganic carbon (DIC) and pH, which are, in turn, a function of physical, chemical and biological processes. This study focus on the importance of CO2 outgassing in the carbon cycle of drainage basins, assuming that river systems are significant sources of atmospheric C in tropical environments, and also on the influence of DIC concentrations and the pH on CO2 evasion. The study area was the Ji-Paraná river basin, Rondônia. Several rivers of the basin were sampled between May/99 and May/03. Temperature, pH and DIC concentrations were used do calculate pCO2, based on thermodynamic equilibrium equations. A theoretical diffusive flux model was used to estimate CO2 evasion. The results show that rivers draining areas with more fertile soils present larger concentrations of DIC and of potential CO2 evasion. In spite of the pCO2 seasonality, the concentrations of dissolved CO2 suggest that this CO2 exchange with the atmosphere is unilateral (evasion to the atmosphere) throughtout the year. During high waters, even with lower DIC concentrations, decreases in pH are of a magnitude enough to promote higher potential evasions; in this period evasion is around four times higher than in the falling water. Based on a diffusion model, CO2 evasion to the atmosphere from rivers of Ji-Paraná basin was estimated to be from 128 to 318 Gg C yr-1. Since the annual export in fluvial discharge is 370 Gg C yr-1, this means that evasion is on the same order of magnitude, demonstrating the importance of this pathway in the carbon cycle of riverine systems in the Amazon.

Le rapport spécial du GIEC (SR1.5) publié en octobre 2018 est sans équivoque. Avec un réchauffement global d'environ 1°C en 2017 par rapport au début de l'ère industrielle et une teneur de CO2 dans l'atmosphère de 400 parties par million, l'Homme a d'ores et déjà modifié substantiellement le climat. L'évaluation de scénarios climatiques à bas niveau d'émissions limitant le réchauffement global à venir en deçà de 2°C montre que nous sommes désormais face à un défi scientifique, technique et civilisationnel sans précédent. Le GIEC stipule que chaque année perdue en matière d'atténuation rend un peu plus plausible la perspective d'une véritable "catastrophe" climatique. Dans ce contexte alarmant, les techniques de modification du rayonnement solaire sont de plus en plus étudiées comme une alternative à court terme pouvant limiter les impacts liés à la hausse de la température globale, en attendant la mise en œuvre et/ou la faisabilité de techniques d'atténuation suffisamment efficaces. La modélisation du système Terre reste à ce jour le seul moyen d'étudier dans quelles mesures ces techniques pourraient effectivement s'insérer dans la lutte contre le changement climatique. Le but principal de cette thèse est de quantifier et de réduire les incertitudes quant à la réponse des modèles aux simulations de géo-ingénierie de type modification du rayonnement solaire, en accordant une attention toute particulière aux effets collatéraux sur les cycles de l'eau et du carbone. Dans un premier temps, nous avons exploité les simulations existantes du projet GeoMIP, et avons identifié une relation statistique émergente entre le refroidissement obtenu dans les simulations de modification du rayonnement solaire, et le refroidissement induit par les éruptions volcaniques majeures dans les simulations historiques. Sur la base de plusieurs jeux d'observations, nous avons évalué la réponse des modèles aux éruptions et ainsi contraint la réponse à la modification du rayonnement solaire, réduisant son efficacité potentielle de 20% et l'incertitude associée de 40%. Par la suite, nous nous sommes intéressés à la réponse du cycle du carbone et avons montré que les changements climatiques induits par cette forme de géo-ingénierie tendent à stimuler les puits de carbone continentaux et océaniques. Nous avons cependant pointé l'incertitude qui entoure les processus responsables de cette augmentation, et également la réversibilité du cycle du carbone en cas d'arrêt de la géo-ingénierie. Malgré le renforcement des puits naturels de carbone, ce dernier résultat confirme que cette forme de géo-ingénierie ne peut être considérée comme une technique d'atténuation du fait de la non pérennité du stockage additionnel des émissions anthropiques de carbone dans les réservoirs océanique et terrestre. Enfin, nous nous sommes intéressés à d'autres sources d'incertitudes, liées au choix du protocole expérimental ou du modèle mis en oeuvre. Nous avons en particulier mis en lumière l’influence potentielle de la stratosphère et de son couplage avec la circulation troposphérique sur la réponse régionale des modèles à nos latitudes. / The IPCC Special Report (SR1.5) published in October 2018 is unequivocal. Global warming reached 1°C above preindustrial level in 2017 and atmospheric concentrations of CO2 passed 400 parts per million. Human activities have already substantially altered the Earth's climate. The assessment of low emission scenarios that limit global warming to 2°C above preindustrial levels shows that we are now facing an unprecedented scientific, technological and civilizational challenge, and stipulates that each year lost for mitigation makes the prospect of a real climate "disaster" a little more plausible. In this alarming context, solar radiation modification techniques are increasingly studied as a short-term alternative in order to limit the impacts of dangerous global warming, before the implementation and/or feasibility of sufficiently effective mitigation techniques. Earth System Models remain the only tool to investigate the extent to which these techniques could be used to counteract global warming. The main purpose of this thesis is to quantify and narrow uncertainties in model response to geoengineering simulations such as solar radiation modification, with special attention to side-effects on water and carbon cycles. First, we have used available simulations from GeoMIP, and identified an emerging statistical link between the cooling obtained in solar radiation management simulations, and the cooling induced by major volcanic eruptions in the historical simulations. Using several observational datasets, we have evaluated the model response to volcanic eruptions and, thereby, constrained the response to this geoengineering technique, reducing its potential cooling efficiency by 20%, and the associated uncertainty by 40%. Subsequently, we have focused on the carbon cycle response and have shown that climatic changes induced by this form of geoengineering tends to stimulate continental and oceanic carbon sinks. However, we have pointed out the uncertainty surrounding the processes responsible for this increase, and also the reversibility of the carbon cycle in case of stopping geoengineering. Despite the enhancement of the natural carbon sinks, this last result confirms that this form of geoengineering cannot be considered as a mitigation technique because of the unsustainability of the additional storage of anthropogenic carbon emissions into ocean and terrestrial reservoirs. Finally, we have looked at other sources of uncertainty related to the choice of the implemented experiment design or model. We have in particular highlighted the potentiel influence of the stratosphere and its coupling with the tropospheric dynamics on the regional response in the northern midlatitudes.

The global continental carbon and water cycles are intimately linked through stomatal regulation during vegetation photosynthesis and biosphere-atmosphere interactions. Therefore, to have a complete understanding of both present and future climate, these cycles must be studied as an interconnected system. This thesis presents three studies that aim to better explain these interactions and provide a direction forward for improved model projections of climate. The first study shows that biosphere-atmosphere feedbacks can contribute up to 30% of climate and weather variability in certain regions that help determine the net CO2 balance of the biosphere. It demonstrates that Earth System Models are under-estimating these contributions, mainly due to the underestimation of the biosphere response to radiation and water availability. It emphasizes the importance of correctly capturing these feedbacks in models for accurate subseasonal to seasonal climate predictions. The second demonstrates that changes in soil moisture (both short-term variability and long-term trends) strongly limit the ability of the continents to act as a carbon sink, with overall effects on the same order of magnitude as the land sink itself. Photosynthesis rates tend to be reduced when soil moisture is depleted, leading to decreased carbon uptake. Additionally, respiration rates increase due to increased temperature through land-atmosphere feedbacks. These carbon losses are not compensated for during wet anomalies due to the nonlinear response of vegetation activity (both respiration and photosynthesis) to soil moisture. This suggests that the increasing trend in carbon uptake rate may not be sustained past the middle of the century and could result in accelerated atmospheric CO2 growth. The third decouples the effects of atmospheric dryness (vapor pressure deficit) and soil dryness on vegetation activity in the largest terrestrial carbon sink: the tropics. Understanding vegetation response to environmental drivers and stressors in the tropics is essential to accurately modeling these ecosystems and predicting whether they will remain carbon sinks in the future. The study finds that in regions that are water limited, vegetation is driven by precipitation and radiation while being limited by high vapor pressure deficit. Conversely, in the wettest regions that are light limited, increases in vapor pressure deficit accompany increasing rates of photosynthesis. These three studies contribute to our understanding of land-atmosphere and biosphere-atmosphere feedbacks and the coupling of the continental carbon and water cycles. They identify model process representations, such as soil moisture and vegetation water-stress, that are hindering our ability to make accurate forecasts. By improving our knowledge of these mechanisms and evaluating the ability of models to reproduce them, we pave the way forward for improved climate and weather projections. Better predictions can be used not only to protect society in the present, but also to appropriately shape climate policy to protect society in the future.