Assessment of Methane Emission Source Characteristics from Oil and Gas Operations Using Satellite, Weather, Environmental, and Operational Data

化石燃料能源部门贡献了当代全球约三分之一的人为甲烷排放量。然而，由于以下几个原因，对油气（O/G）作业中甲烷排放量的可靠估计是很复杂的：(1)在广泛的地理区域（例如，一个拥有100万口井的生产盆地）连续详细的现场采样数据是不可用的；(2)从大规模监测（如卫星）数据中识别和量化排放源具有挑战性，因为缺乏针对这种不适定逆问题的可靠分析技术。事实上，越来越多的证据表明，与其他基于大气测量的自上而下方法相比，涉及当地测量的自下而上方法往往会导致系统性低估。本项目旨在开发利用卫星检索大气甲烷摩尔分数（XCH4）的新技术，以量化甲烷排放率，识别排放源并评估其特征。时间平均XCH4图，连同其他环境、气象和O/G操作/生产数据（如油井活动和产量），用于推断潜在的排放源。还将分析周边地区其他主要排放源的信息（例如，垃圾填埋场、煤矿和城市住宅）。以前的研究侧重于跟踪一个或几个特定排放源的排放，这些技术不适合研究具有许多源的大型O/G盆地的排放。该项目旨在开发数据驱动的建模框架，用于量化O/G作业中的甲烷排放源，并通过以下方式评估相关特征：(1)建立机器或深度学习模型，将生产数据和油井活动与甲烷排放率相关联；(2)了解O/G活动与排放源特征之间的关系。结果，包括规范、程序和加拿大主要生产盆地的具体案例研究，将有助于评估现有作业的环境绩效，并确定进一步监测的机会。

The fossil fuel energy sector has contributed approximately 1/3 of the contemporary anthropogenic methane emissions worldwide. However, reliable estimating of methane emissions from oil/gas (O/G) operations is complicated for several reasons: (1) continual detailed field sampling data over an extensive geographical area (e.g., a producing basin with >1000 wells) is unavailable; (2) identifying and quantifying emission sources from large-scale monitoring (e.g., satellite) data is challenging because there is a lack of robust analysis techniques for this ill-posed inverse problem. In fact, there is increasing evidence that bottom-up approaches involving local measurements would often lead to systematic underestimation, as compared to other top-down approaches based on atmospheric measurements. This project aims to develop new techniques for utilizing satellite retrievals of mole fractions of atmospheric methane (XCH4) to quantify methane emission rate, identify emission sources, and assess their characteristics. Maps of time-averaged XCH4, together with other environmental, meteorological, and O/G operational/production data (e.g., well activity and production volumes), are used to infer potential emission sources. Information about other major emission sources in surrounding areas will also be analyzed (e.g., landfills, coal mines, and urban dwellings). Previous studies focused on tracking emissions from one or a few specific emission sources, and those techniques are not suitable for examining emissions from large O/G basins with many sources. This project aims to develop data-driven modelling frameworks for quantifying methane emission sources from O/G operations and assessing the associated characteristics by (1) building machine- or deep-learning models for correlating production data and well activity to methane emission rates; and (2) understanding the relationships between O/G activities and emission source characteristics. The outcomes, including codes, procedures, and specific case studies of major Canadian-producing basins, would be useful for assessing the environmental performance of existing operations and identifying further monitoring opportunities.