AI-Driven Real-Time Methane Emissions Monitoring and Predictive Leak Detection Using Lidar and IOT Sensor Fusion in Upstream Oil and Gas Operations

Abstract

This study addresses the persistent problem of delayed, fragmented, and insufficiently integrated methane leak detection in upstream oil and gas operations, where conventional inspection-based approaches often fail to capture spatially dispersed, intermittent, and high-risk emission events. The purpose of the research was to examine whether an AI-driven real-time methane monitoring framework using LiDAR and IoT sensor fusion can improve monitoring performance, predictive leak detection, trustworthiness, and deployment readiness in upstream environments. The study adopted a quantitative, cross-sectional, case-based design focused on cloud-enabled enterprise monitoring contexts in upstream oil and gas operations, using a purposive sample of 214 valid professional respondents drawn from methane monitoring, operations, compliance, maintenance, and intelligent systems roles. Key variables included LiDAR monitoring capability, IoT sensor fusion efficiency, AI predictive analytics capability, real-time methane monitoring performance, predictive leak detection effectiveness, leak response capability, sensor fusion readiness and trustworthiness, leak detection timeline advantage, and high-risk scenario confidence. Data were collected through a structured 5-point Likert-scale questionnaire and analyzed using descriptive statistics, Cronbach’s alpha, correlation analysis, and multiple regression modeling. The findings showed strong positive evaluations across all core constructs, including AI predictive analytics capability (M = 4.23, SD = 0.58), predictive leak detection effectiveness (M = 4.20, SD = 0.60), LiDAR monitoring capability (M = 4.18, SD = 0.61), and real-time methane monitoring performance (M = 4.15, SD = 0.63). Correlation results indicated significant positive relationships, including LiDAR with real-time monitoring performance (r = .742, p < .001), AI capability with predictive leak detection effectiveness (r = .768, p < .001), and real-time monitoring performance with leak response capability (r = .733, p < .001). Regression analysis further showed that the predictors explained 67.4% of the variance in real-time monitoring performance (R² = .674) and 71.6% of the variance in predictive leak detection effectiveness (R² = .716), with AI emerging as the strongest predictor of predictive leak detection (β = .389, p < .001). The study implies that integrating AI, LiDAR, and IoT sensor fusion can provide a more proactive, trustworthy, and operationally deployable methane surveillance architecture for upstream oil and gas operations.