Cette thèse en économie de l'énergie et de l'environnement étend le modèle de Hotelling du type exploration-extraction avec contraintes géologiques d’Okullo, Reynes et Hofkes (2015), afin de prendre en compte des trajectoires en forme de cloche pour l’ajout de réserves empiriquement observées par Laherrère (2003). Le modèle LOGIMA proposé (Images à Long terme sur le Pétrole et le Gaz) explique qu’elles sont la conséquence de « sweet spots » géologiques : des zones privilégiées où la concentration d’hydrocarbures est la plus élevée. Le modèle LOGIMA a été calibré sur une base de données issues couvrant les sept principaux bassins de pétrole et de gaz non-conventionnels du pays. Les résultats indiquent que la nécessité d’apprentissage par la pratique pour découvrir l’emplacement des sweet spots conduit à une mise en œuvre d’un effort d’exploration également en forme de cloche, ce qui permet de réduire le risque des activités d’exploration. Par conséquent, la réponse en termes des volumes offerts par les producteurs à des chocs sur les prix dévient fonction de l’ensemble des ressources mondiales antérieurement découvertes. Ensuite, nous appliquons le modèle LOGIMA pour étudier l’impact causé par l’offre de pétrole et de gaz naturel « non-conventionnels » aux États-Unis, sur les efforts mondiaux d’atténuation du changement climatique. Nous y parvenons en associant les scénarios à long-terme générés par LOGIMA avec le modèle d’évaluation intégrée, IMACLIM-R. Cette étude analyse comment des différentes cibles de prix de pétrole affecteraient son offre aux États-Unis. Nous estimons cette interaction au moyen de trois cadres de politiques en matière de climat : le cadre « business as usual » (BAU), les contributions décidées à l’échelle nationale (NDC) et les scénarios de 2°C (2DS). Les résultats de l’exercice indiquent que les approvisionnements non-conventionnels sont fortement susceptibles d’affecter les marchés énergétiques mondiaux, mais leur impact sur les émissions mondiales de gaz à effet de serre serait limité, car les différents effets déclenchés dans des différents secteurs viendraient les équilibrer approximativement. / This thesis in energy and environmental economics extends the geological Hotelling-type extraction-exploration model from Okullo, Reynes and Hofkes (2015) in order to account for the bell-shaped reserve additions that were empirically observed by Laherrère (2003). The proposed model explains them as the result of geological “sweet spots”: premium areas within geological formations where the concentration of hydrocarbons is highest. The proposed theoretical formulation was programmed into the mathematical model LOGIMA – “Long-term Oil and Gas Images” – and calibrated on data covering the seven main unconventional oil and gas plays in the United States. Results indicate the need to learn the location of sweet spots through trial and error drillings leads to schedules of exploratory effort that allow the optimal “de-risking” of exploratory activities. As a result, the optimal response of producers to price shocks becomes contingent on the prevailing level of cumulative discoveries.We apply LOGIMA to investigate the impact, caused by the recent advent of large-scale supplies of unconventional oil and gas, in the United States, on the ongoing efforts to mitigate climate change. We do so by soft coupling long-term scenarios from LOGIMA with the integrated assessment model, IMACLIM-R, a recursive, computable general equilibrium model of integrated global energy, economy and environment systems. We analyze how different price targets, potentially pursued by the Organization of Petroleum Exporting Countries (OPEC), would affect supplies of unconventional oil and gas from the United States. We control this interplay under three climate policy frameworks: business as usual (BAU), nationally determined contributions (NDCs) and 2°C scenario (2DS). The results of the exercise show that, despite having a significant potential to affect global energy markets, unconventional oil and gas supplies would have a limited potential to affect global cumulative greenhouse gas emissions to 2040, as the different effects triggered in different sectors approximately balanced each other out.

High total organic carbon (TOC) and low clay content are two criteria to identify the "sweet spots" in shale gas plays. Recently, machine learning has been proved to be effective to estimate TOC and clay from well loggings. The remaining questions are what algorithm we should choose in the first place and whether we can improve the already built models. Automatic machine learning (AutoML) appears as a promising tool to solve those realistic questions by training multiple models and compares them automatically. Two wells with conventional well loggings and elemental capture spectroscopy are selected from a shale gas play to test the AutoML's ability in TOC and clay estimation. TOC and clay content are extracted from the Schlumberger's ELAN interpretation and calibrated to cores. Generalizability is proved in the blind test well and the mean absolute test errors for TOC and clay estimation are 0.23% and 3.77%. 829 data points are used to generate the final models with the train-test ratio of 75:25. The mean absolute test errors are 0.26% and 2.68% for TOC and clay, respectively, which are very low for TOC ranging from 0-6% and clay from 35-65%. The results show the AutoML's success and efficiency in the estimation. The trained models are interpreted to understand the variables effects in predictions. 235 wells are selected through data quality checking and feed into the models to create TOC and clay distribution maps. The maps provide guidance on where to drill a new well for higher shale gas production. / Master of Science / Locating "sweet spots", where the shale gas production is much higher than the average areas, is critical for a shale reservoir's successful commercial exploitation. Among the properties of shale, total organic carbon (TOC) and clay content are often selected to evaluate the gas production potential. For TOC and clay estimation, multiple machine learning models have been tested in recent studies and are proved successful. The questions are what algorithm to choose for a specific task and whether the already built models can be improved. Automatic machine learning (AutoML) has the potential to solve the problems by automatically training multiple models and comparing them to achieve the best performance. In our study, AutoML is tested to estimate TOC and clay using data from two gas wells in a shale gas field. First, one well is treated as blind test well and the other is used as trained well to examine the generalizability. The mean absolute errors for TOC and clay content are 0.23% and 3.77%, indicating reliable generalization. Final models are built using 829 data points which are split into train-test sets with the ratio of 75:25. The mean absolute test errors are 0.26% and 2.68% for TOC and clay, respectively, which are very low for TOC ranging from 0-6% and clay from 35-65%. Moreover, AutoML requires very limited human efforts and liberate researchers or engineers from tedious parameter-tuning process that is the critical part of machine learning. Trained models are interpreted to understand the mechanism behind the models. Distribution maps of TOC and clay are created by selecting 235 gas wells that pass the data quality checking, feeding them into trained models, and interpolating. The maps provide guidance on where to drill a new well for higher shale gas production.