Three-dimensional empirical Bayesian kriging for soil PAHs interpolation considering the vertical soil lithology

Abstract

Contaminated sites usually showed the characteristics of vertical stratification heterogeneity due to the local pollutant accumulation and complex hydrogeological conditions. However, the existing three-dimensional interpolation methods are deficient in addressing the vertical stratification heterogeneity. In order to overcome this shortcoming, we proposed a novel three-dimensional empirical Bayesian Kriging (3-D EBK) interpolation method considering soil lithology. Then, we applied the method to examine the influence of soil lithology on pollutants transfer in the vertical direction and the 3D interpolation accuracy. A typical PAHs contaminated site was used to verify the three-dimensional spatial distribution pattern of pollutants. The results showed that the geographic detectors model divided soli lithology into three and six types according to the concentration of Benzo(a)pyrene (BaP) and Dibenzo (A, H) anthracene (DBA), respectively. PAHs showed significant vertical heterogeneity in different soil lithologies, which influenced the prediction accuracy of PAHs 3D interpolation in soil. Compared with the existing 3-D EBK model, the 3-D EBK model considering soil lithology proposed in the present study could improve the interpolation accuracy of BaP by 9.46% and DBA by 8.33%. The root mean square error (RMSE) of BaP and DBA interpolation results decreased with increasing depth. The minimum RMSE of BaP and DBA was 0.35 mg·kg−1 and 0.08 mg·kg−1, respectively. The maximum root mean square error reduction rate (RMSERR) of BaP and DBA was 20.46% and 50.00%, respectively. Compared with DBA, BaP had a strong downward migration trend. The 3-D EBK model considering soil lithology overcame the deficiency of the existing 3-D EBK model in capturing the spatial anisotropic characteristics and improved the prediction accuracy of PAHs distribution. The study provides an innovative method for precisely depicting the three-dimensional spatial pattern of pollutants at contaminated sites.