Accounting for Soft Information in Mapping Soil Contamination with TPH at an Oil Storage Site

Abstract

This paper presents a methodology to account for soft information in mapping soil contamination with TPH (Total Petroleum Hydrocarbons) at an oil storage site. The former function of the site consisted of storage of various fuel types, including diesel fuel, and crude oil. The soil survey consisted of 140 borings covering an area of about 17.5 ha. Two different types of data are available: an abundant set of macroscopic analysis of soil characteristics and a sensory analysis of oil content; and a limited set of 100 soil samples analysed for TPH at a laboratory. It is considered that the more abundant but more uncertain oil reaction data and the more accurate and scarce TPH analyses are measurements of the same attribute — soil contamination with oil products. The methodology proposed consists in characterising the contaminated areas using TPH data, but conditioned to the spots defined by sensory data. The methodology can be summarised by the following sequence: i) spatial estimation of two oil reaction classes defining areas of clean soil and contaminated soil based only on soft data; ii) then inside each spot (“contaminated” and “clean”) stochastic simulation of the TPH variable was performed. Although the high values of TPH are preferentially located in areas previously estimated as “contaminated”, low values of TPH can also occur in these “contaminated” areas and vice-versa. Thus, in this step the final maps are obtained by characterising the spatial distribution of TPH inside each clean and contaminated spot and reclassifying these areas based on TPH simulated values.