Soil TPH concentration estimation using vegetation indices in an oil polluted area of eastern China
Rimmington, Glyn M.
MetadataShow full item record
Linhai Zhu, Xuechun Zhao, Liming Lai, Jianjian Wang, Lianhe Jiang, Jinzhi Ding, Nanxi Liu, Yunjiang Yu, Junsheng Li, Nengwen Xiao, Yuanrun Zheng, Glyn M. Rimmington. 2013. Soil TPH concentration estimation using vegetation indices in an oil polluted area of eastern China. PLoS One, v.8 no.1 e54028
Assessing oil pollution using traditional field-based methods over large areas is difficult and expensive. Remote sensing technologies with good spatial and temporal coverage might provide an alternative for monitoring oil pollution by recording the spectral signals of plants growing in polluted soils. Total petroleum hydrocarbon concentrations of soils and the hyperspectral canopy reflectance were measured in wetlands dominated by reeds (Phragmites australis) around oil wells that have been producing oil for approximately 10 years in the Yellow River Delta, eastern China to evaluate the potential of vegetation indices and red edge parameters to estimate soil oil pollution. The detrimental effect of oil pollution on reed communities was confirmed by the evidence that the aboveground biomass decreased from 1076.5 g m−2 to 5.3 g m−2 with increasing total petroleum hydrocarbon concentrations ranging from 9.45 mg kg−1 to 652 mg kg−1. The modified chlorophyll absorption ratio index (MCARI) best estimated soil TPH concentration among 20 vegetation indices. The linear model involving MCARI had the highest coefficient of determination (R2=0.73) and accuracy of prediction (RMSE=104.2 mg kg−1). For other vegetation indices and red edge parameters, the R2 and RMSE values ranged from 0.64 to 0.71 and from 120.2 mg kg−1 to 106.8 mg kg−1 respectively. The traditional broadband normalized difference vegetation index (NDVI), one of the broadband multispectral vegetation indices (BMVIs), produced a prediction (R2=0.70 and RMSE=110.1 mg kg−1) similar to that of MCARI. These results corroborated the potential of remote sensing for assessing soil oil pollution in large areas. Traditional BMVIs are still of great value in monitoring soil oil pollution when hyperspectral data are unavailable.
Click on the DOI link or attached PDF to access the article.