Detecting and quantifying methane emissions from oil and gas production: algorithm development with ground-truth calibration based on Sentinel-2 satellite imagery

Sentinel-2 satellite imagery has been shown by studies to becapable of detecting and quantifying methane emissions from oil and gasproduction. However, current methods lack performance calibration withground-truth testing. This study developed a multi-band-multi-pass-multi-comparison-date methane retrieval algorithm that enhances Sentinel-2 sensitivity to methane plumes. The method was calibratedusing data from a large-scale controlled-release test in Ehrenberg, Arizona,in fall 2021, with three algorithm parameters tuned based on the trueemission rates. Tuned parameters are the pixel-level concentration upper-bound threshold during extreme value removal, the number of comparisondates, and the pixel-level methane concentration percentage threshold whendetermining the spatial extent of a plume. We found that a low value of theupper-bound threshold during extreme value removal can result in falsenegatives. A high number of comparison dates helps enhance the algorithmsensitivity to the plumes in the target date, but values in excess of12 d are neither necessary nor computationally efficient. A high percentagethreshold when determining the spatial extent of a plume helps enhance thequantification accuracy, but it may harm the yes/no detection accuracy. Wefound that there is a trade-off between quantification accuracy anddetection accuracy. In a scenario with the highest quantification accuracy,we achieved the lowest quantification error and had zero false-positivedetections; however, the algorithm missed three true plumes, which reduced theyes/no detection accuracy. In contrast, all of the true plumes weredetected in the highest detection accuracy scenario, but the emission ratequantification had higher errors. We illustrated a two-step method thatupdates the emission rate estimates in an interim step, which improvesquantification accuracy while keeping high yes/no detection accuracy. Wealso validated the algorithm's ability to detect true positives and truenegatives in two application studies.