a b s t r a c t Keywords:Gas flare Thermal infrared remote sensing MODIS flare detection Niger Delta Gas flaring volume 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.

Pollution from oil and gas exploitation in the Niger Delta has greatly endangered the natural ecosystem, with gas flaring identified as a key agent of environmental pollution in the region. Efforts to evaluate the impacts of flaring on the surrounding environment have been hampered by limited access to official information on flare locations and volumes; hence an alternative method of acquiring such information is needed. This paper describes the development and application of the Landsat Flare Detection Method (LFDM), based on the combination of the near, shortwave and thermal infrared bands of Landsat imagery. The technique was validated using a reference dataset of flare locations interpreted from aerial photographs, achieving a user accuracy of 86.67%. The LFDM was applied to a time-series of imagery (1984 to 2012 inclusive) to obtain a long term flaring history of the region; 303 flares (251 onshore and 52 offshore) were detected over the study period. The spatiotemporal distribution of these flares corresponds with known variations in oil and gas activities in the region. There was considerable variation between states in the trajectories of gas flaring activity and the proportion of onshore versus offshore flaring, which indicates substantial spatiotemporal variations in the environmental impacts of this industry. The LFDM builds upon existing methods of flare detection, which were based on moderate resolution imagery, by offering: increased precision of flare location estimates, improved objectivity, accurate identification of onshore and offshore flares and a long flaring history. The LFDM is an efficient and cost effective method which is able to provide local to regional scale information which is complementary to that derived from other remote methods of flare detection and ground-based surveys. It could thus be used for either backward (flare history) and/or forward (monitoring) surveys, especially in monitoring the country's progress towards the recently set 30% flare reduction target by 2017.

No abstract

No abstract