An approach to detect gas flaring sites using sentinel-2 MSI and NOAA-20 VIIRS images

Abstract

Detecting gas flaring activities during oil production on a regional scale is necessary, since it emits harmful gases and bring serious global environmental impacts. This study developed a new algorithm to detect gas flare sites (GFs) on daytime Sentinel-2 MSI images and Nighttime NOAA-20 VIIRS images. The algorithm includes three steps: Thermal Anomaly Index (TAI) on single-temporal MSI images was first used to detect potential GFs, then the daily night temperature of potential GFs was calculated based on Planck's law on VIIRS images, and finally the confirmed GFs were filtered based on the detection frequency and temperature. The algorithm was implemented in five regions with different surface covers in the top four countries (Russia, Iran, Iraq, and the USA) in terms of flared gas volumes in 2021. The algorithm achieves an average producer accuracy of 81.7 %, an average user accuracy of 80.3 %, and the root mean square error of the spatial position is 15.6 m for these five test areas. Compared with the existing global datasets of GFs, this algorithm detects the most GFs and the spatial location of the GFs is more accurate, indicating that the proposed method for GF detection has high spatial resolution and the result is reliable. This algorithm could be applied to detect the GFs globally and provide the scientific data to enact efficient measures to alleviate the environmental impacts.

Keywords

• Gas flaring
• Thermal Anomaly Index
• Sentinel-2 MSI
• Daytime observation
• NOAA-20 VIIRS