Detection and temperature estimation of gas flares with nocturnal Landsat OLI

Abstract

Natural gas flaring is a worldwide polluting activity carried out during oil production. Satellite imagery has emerged as a low-cost, objective tool to measure and monitor gas flaring. Since 2012, hectometre-resolution infrared imagery from the Suomi NPP VIIRS sensor has been used to operationally monitor global gas flaring (Elvidge et al. 2016). Since 2013, nocturnal acquisitions of Landsat 8 OLI and TIRS imagery have been collected intermittently in a small number of locations worldwide – including part of the Williston basin in North Dakota, where oil production from shale formations results in an unusually high density of production wells. The goal of this study is to compare the use of VIIRS and nocturnal Landsat 8 Operational Land Imager (OLI) in identifying and characterizing gas flare populations associated with tight oil production, where flares are smaller, more numerous, and more closely spaced than in conventional oil production. We find that the OLI optical bands resolve individual closely-spaced (< 200 m) flares, and that most flares in this region are small and persistent on time scales of months. Most of these flares are not detected by either the VIIRS Day/Night Band or NOAA’s VIIRS Nightfire product. Of the 154 largest flares selected, 73 contained saturated pixels in at least one OLI band. We invert the Planck function for blackbody radiation to estimate temperature and emissivity for the 81 non-saturated flares. Temperature estimates are plausible (1500 K to 2200 K), but RMS misfits vary by orders of magnitude. The Nightfire product detects and provides temperature estimates for 16 of these 81 flares. OLI-derived temperatures are comparable to Nightfire estimates, but generally higher. Although Landsat’s 16-day revisit period precludes its use for flare monitoring on its own, the positional accuracy and low noise floor of OLI SWIR bands allows for repeat imaging of individual flares in settings where most flares are too small and closely-spaced to be resolved by VIIRS. Because these small flares account for most of the emission from tight oil production, nocturnal Landsat could provide an important complement to flaring estimates derived from VIIRS alone.

Avatar
Ruiwen Lee
Job Market Candidate

I am an environmental economist.