Flaring and pollution detection in the Niger Delta using Remote Sensing

Abstract

Abstract Through the Global Gas Flaring Reduction (GGFR) initiative a substantial amount of effort and international attention has been focused on the reduction of gas flaring since 2002 (Elvidge et al., 2009). Nigeria is rated as the second country in the world for gas flaring, after Russia. In an attempt to reduce and eliminate gas flaring the federal government of Nigeria has implemented a number of gas flaring reduction projects, but poor governmental regulatory policies have been mostly unsuccessful in phasing it out. This study examines the effects of pollution from gas flaring using multiple satellite based sensors (Landsat 5 TM and Landsat 7 ETM+) with a focus on vegetation health in the Niger Delta.

Over 131 flaring sites in all 9 states (Abia, Akwa Ibom, Bayelsa, Cross Rivers, Delta, Edo, Imo, Ondo and Rivers) of the Niger Delta region have been identified, out of which 11 sites in Rivers State were examined using a case study approach. Land Surface Temperature data were derived using a novel procedure drawing in visible band information to mask out clouds and identify appropriate emissivity values for different land cover types. In 2503 out of 3001 Landsat subscenes analysed, Land Surface Temperature was elevated by at least 1 ℃ within 450 m of the flare. The results from fieldwork, carried out at the Eleme Refinery II Petroleum Company and Onne Flow Station, are compared to the Landsat 5 TM and Landsat 7 ETM+ data.

Results indicate that Landsat data can detect gas flares and their associated pollution on vegetation health with acceptable accuracy for both Land Surface Temperature (range: 0.120 to 1.907 K) and Normalized Differential Vegetation Index (sd ± 0.004). Available environmental factors such as size of facility, height of stack, and time were considered. Finally, the assessment of the impact of pollution on a time series analysis (1984 to 2013) of vegetation health shows a decrease in NDVI annually within 120 m from the flare and that the spatio-temporal variability of NDVI for each site is influenced by local factors. This research demonstrated that only 5 % of the variability in δLST and only 12 % of the variability in δNDVI, with distance from the flare stack, could be accounted for by the available variables considered in this study. This suggests that other missing factors (the gas flaring volume and vegetation speciation) play a significant role in the variability in δLST and δNDVI respectively.