Replication Data for: Optimizing Methane Emission Source Localization in Oil and Gas Facilities Using Lagrangian Stochastic Models and Gradient-Based Detection Tools

This dataset includes synthetic methane concentration simulations, real-world case study data, and the associated outputs and scripts used to support the findings of this study. The synthetic data were generated using a Gaussian Dispersion Model (GDM) and results from a Lagrangian-back-trajectory model (TERRAFEX) under various sensor-source geometries and atmospheric conditions, simulating stationary sensor deployment scenarios in oil and gas facilities. The real-world dataset consists of concentration reconstructions and localization results from a gas distribution site in Canada, used to validate the methodology under operational conditions.
The dataset includes:
• Synthetic data:
Simulated CH₄ concentration time series, concentration maps, and source localization outputs created using forward Gaussian dispersion modeling. These datasets cover various source-sensor distances (0–125 m), height differences (0–5 m), and four atmospheric stability classes (A–D), resulting in 456 synthetic scenarios.
• Case study data:
Measurement data and localization outputs collected from a gas distribution site in Canada, including CH₄ concentration data, wind measurements, and source localization maps reconstructed using TERRAFEX and the Gradient Indicator (GI) tool.
• Outputs:
Includes reconstructed concentration fields with TERRAFEX for both synthetic and case study scenarios.
• Scripts:
All analysis and visualization scripts used in this study were written in R (version 4.2.X). The scripts include the GDM forward model, and the custom-developed GI tool used for source localization from reconstructed concentration maps.
Instrumentation and Methodology
• Synthetic Data Generation:
Synthetic data were created using forward Gaussian dispersion modeling, incorporating real wind data collected in Weyburn, Alberta during December 2015, using a Gill ultrasonic anemometer (precision: 3% RMSE for wind speed, ±3° for wind direction). The simulations used a fixed CH₄ emission rate of 41 m³/day with Pasquill-Gifford stability classes A to D. The TERRAFEX model was applied in concentration mode to reconstruct the source contribution maps from the synthetic measurement line.
• Case Study Measurements:
Data were collected at a gas distribution site in Nova Scotia using an Axetris LGD Compact-A CH₄ analyzer (precision 0.01 ppm, 2 Hz) and a Decagon DS-2 Sonic Anemometer (precision 0.01 m s⁻¹, 1°). Sensor height was varied from 2 m to 4.5 m during the measurement campaign (June 24 to July 8, 2020).

本数据集包含合成甲烷浓度模拟数据、真实场景案例研究数据，以及用于支撑本研究结论的相关输出文件与脚本代码。合成数据通过高斯扩散模型（Gaussian Dispersion Model, GDM）生成，并结合拉格朗日后向轨迹模型（TERRAFEX）的结果，在多种传感器-源几何布局与大气条件下，模拟油气设施中固定式传感器部署场景。真实世界数据集源自加拿大某燃气场站的浓度重建结果与源定位结果，用于在实际作业条件下验证研究方法。

本数据集包含以下内容：
• 合成数据：
通过正向高斯扩散建模生成的模拟甲烷（CH₄）浓度时间序列、浓度分布图，以及源定位输出结果。此类数据集覆盖0至125米的源-传感器间距、0至5米的高度差，以及A至D共4种帕斯奎尔-吉福德大气稳定度等级，共计456组合成场景。
• 案例研究数据：
采集自加拿大某燃气场站的测量数据与源定位输出结果，包含甲烷浓度数据、风速测量数据，以及通过TERRAFEX与梯度指示器（Gradient Indicator, GI）工具重建的源定位分布图。
• 输出文件：
包含针对合成数据与案例研究场景，通过TERRAFEX重建的浓度场数据。
• 脚本代码：
本研究中使用的全部分析与可视化脚本均采用R语言（版本4.2.X）编写，涵盖正向GDM模型，以及用于从重建浓度图中进行源定位的自研GI工具。

仪器与方法
• 合成数据生成：
合成数据通过正向高斯扩散建模创建，采用2015年12月在阿尔伯塔省韦本采集的真实风速数据，使用吉尔超声波风速仪（风速精度：均方根误差3%，风向精度：±3°）。模拟过程中固定甲烷排放速率为41 m³/天，采用帕斯奎尔-吉福德稳定度等级A至D。TERRAFEX模型以浓度模式运行，从合成测量线中重建源贡献分布图。
• 案例研究测量：
数据采集于新斯科舍省的某燃气场站，使用Axetris LGD Compact-A型甲烷分析仪（精度0.01 ppm，采样频率2 Hz）与Decagon DS-2型超声波风速仪（精度0.01 m·s⁻¹，风向精度1°）。本次测量作业期间（2020年6月24日至7月8日），传感器安装高度在2米至4.5米之间调整。