Uncertainty analysis on spatial distribution prediction of BaP in a coking plant site
某焦化场地苯并(a)芘污染空间分布范围预测的不确定性分析

Uncertainty in site investigation and the determination of contamination boundary at large-scale contaminated sites is a critical issue in China. In order to test the influence of different prediction models on the determination of contamination boundary in a coking plant contaminated by benzo(a)pyrene (BaP), three spatial interpolation models, Inverse Distance Weighting model (IDW), Johnson's ordinary lognormal kriging model (OLKM), and Combination Prediction Model (CPM), were employed to compare their efficiencies and precisions in determining site contamination boundary. A recommended value 0.4 mg·kg-1 for BaP was used as the reference standard based on the Beijing Screening Levels for Soil Environmental Risk Assessment of Sites. The Results showed that the contamination areas calculated by IDW, OLKM, and CPM were 70.15%, 44.78% and 57.06%, respectively. The CPM was found to be more accurate than the other two models in predicting the actual pollution situations of the contaminated site. The area on the top-right corner of the site with less sampling points and the area in the mid-bottom part with high levels of contamination showed the largest standard errors based on the prediction standard errors map. This study gives useful references for dealing with uncertainty in determining remediation boundary of contaminated sites.