Spatial Modelling of Concentration in Topsoil Using Random and Systematic Uncertainty Components: Comparison against Established Techniques

The assessment of contaminated land often requires the collection and analysis soil samples discrete or continuous modeling contamination. involves interpolating analyte concentrations between sampled positions. contamination can be expressed either as a map showing spatial distribution across site, frequency distribution, both which compared with threshold. characterization is affected by random systematic uncertainty components sampling, chemical analysis, sampling positioning modeling. This work describes application comparison three techniques for developing models site supported quantification none, some all relevant components. 7.3 ha was characterized using 100 lead modeled “inverse distance weighting” (IDW), “ordinary kriging”(OK) new Monte Carlo simulation method (MCM). IDW only uses positions without their uncertainty. OK also “measurement error” other parameters to select variogram. MCM measurement (including effects arising from analysis) ‘GPS coordinates uncertainty’. measurements Pb concentration were log-normally distributed, therefore log-transformed prior output against model ‘probabilistic block mapping’ (PBM) that considers produced smoothed variation appears more realistic. However, underestimate contamination, while prediction most closely matches measured including impact allowed metrologically sound linear interpolation data reduced subjectivity. Both PBM made realistic estimates proportion over threshold included in modeling, respectively. identifies areas where true value contaminant could exceed value, even though single did not.