Methane (CH4) is the major component of natural gas, a potent greenhouse gas, and a precursor for the formation of tropospheric ozone. Enhanced drilling and recovery techniques have tremendously increased the natural gas supply in the U.S. At the same time, fugitive releases occurring during gas extraction, distribution and use are common and sometimes very large, especially considering that the gas transmission and distribution system is very extensive and much if very old and prone to leaks. These releases, in part, are responsible for the increased levels of CH4 found locally as well as the rapid rise CH4 levels globally. Thus, better methods to detect leaks and CH4 releases have taken on both local and global significance. In this study, we examine spatial and temporal trends of ambient CH4 levels and identify leaks and releases from natural gas pipelines in Detroit, Michigan, an area with vulnerable communities, extensive construction, considerable truck traffic, poorly maintained infrastructure, and a variety of commercial and industrial emission sources. We also focus on improved algorithms for detecting and quantifying CH4 peaks and identifying locations of pipeline leaks. We collected field data over the 2019 – 2021 period using the Michigan Pollution Assessment Laboratory (MPAL), a mobile platform that collected rapid (1-sec) and precise (1 ppb resolution) CH4 measurements using two cavity ringdown instruments with inlets at heights above the road surface of 0.1 m (just above the road) and 3 – 3.5 m (at rooftop). We contrast the nature of peaks observed using the two inlets, and refine the peak detection algorithm to account for changes in background levels and spatial and temporal fluctuations. We identify hundreds of leaks in southwest Detroit, many of which persisted for months. The largest peak was 24 ppm above background, which was typically about 2.1 ppm. Differences between 0.1 and 3 – 3.5 m height inlets were not always large, indicating a fairly high degree of mixing. Using a subset of the data, we evaluated the effects of changing several parameters in the peak detection algorithm and settled on the use of the 5th percentile level in a 450-s window looking both forward and backward from a peak to estimate background levels, based the peak edge detection on a threshold of 5% above background, and merged peaks occurring within 5 s. This algorithm detected and characterized peaks reliably. Overall, our results show numerous pipeline leaks that represent an important and controllable source of CH4, and several ways to improve the performance of mobile monitoring.
.jpg)
