Detecting gas flares and estimating flaring volumes at individual flow stations using MODIS data

Gas flaring has gained global recognition as a prominent agent of pollution, leading to the establishment of the Global Gas Flaring Reduction (GGFR) initiative, which requires an objective means of monitoring flaring activity. Because auditable information on flaring activity is difficult to obtain there have recently been attempts to detect flares using satellite imagery, typically at global scales. However, to adequately assess the environmental and health impacts of flaring from local to regional scales, it is important that we have a means of acquiring information on the location of individual active flaring sites and the volume of gas combusted at these sites. In this study we developed an approach to the retrieval of such information using nighttime MODIS thermal imagery. The MODIS flare detection technique (MODET) and the MODIS flare volume estimation technique (MOVET) both exploit the absolute and contextual radiometric response of flare sites. The levels of detection accuracy and estimation error were quantified using independent observations of flare location and volume. The MODET and MOVET were applied to an archive of MODIS data spanning 2000-2014 covering the Niger Delta, Nigeria, a significant global hotspot of flaring activity. The results demonstrate the substantial spatial and temporal variability in gas flaring across the region, between states and between onshore and offshore sites. Thus, whilst the estimated total volume of gas flared in the region over the study period is large (350 Billion Cubic Metres), the heterogeneity in the flaring indicates that the impacts of such flares will be highly variable in space and time. In this context, the MODET and MOVET offer a consistent and objective means of monitoring flaring activity over an appropriate range of scales and it is now important that their robustness and transferability is tested in other oil-producing regions of the world. (C) 2014 Elsevier Inc All rights reserved.