Gas flaring detecting based on Normalized Hotspot Index using multi-source remote sensing data: a case study of Xinjiang

Gas flaring, a long-standing issue in the petrochemical industry as a business problem, has become the second leading cause of methane emissions. We proposed a gas flaring detection method using multi-source remote sensing data. The method combined Visible Infrared Imaging Radiometer Suite (VIIRS) Nightfire data and the Normalized Hotspot Index (NHI) algorithm. At first, the K-means algorithm is applied for spatial clustering of VIIRS Nightfire data. Then, the NHI algorithm was calculated using the Landsat-8 Operational Land Imager and Sentinel-2 Multi-spectral Instrument images. Finally, Xinjiang province in China was selected as the study area, and the gas flaring detection results were validated by the existing global gas flare datasets. The results show that the method presented in our study has high performance and provides more accurate information in detecting gas flares, with an accuracy of 72.73%. In addition, the geographic location of gas flaring sites correlates with oil and gas fields, primarily concentrated in the Tarim Basin in the southwest and Junggar and Turpan Basins in the north. Aksu accounted for the highest, representing 43.05% of the total. This method offers a reliable way to detect gas flares onshore and provides effective measurements for controlling methane emissions.