Uncertainty assessment of heavy metal soil contamination mapping using spatiotemporal sequential indicator simulation with multi-temporal sampling points.

Abstract

Mapping the space-time distribution of heavy metals in soils plays a key role in contaminated site classification under conditions of in situ uncertainty, whereas uncertainty assessment is based on the quantification of the specific uncertainties in terms of exceedance probabilities. Geostatistical space-time kriging (STK) is increasingly used to estimate pollutant concentrations in soils. Sequential indicator simulation (SIS) technique is popular in uncertainty assessment of heavy metal contamination of soils. However, these techniques cannot handle multi-temporal data. In this work, spatiotemporal sequential indicator simulation (STSIS) based on an additive space-time semivariogram model (STSIS_A) and on a non-separable space-time semivariogram model (STSIS_NS) was used to assimilate multi-temporal data in the mapping and uncertainty assessment of heavy metal distributions in contaminated soils. Cu concentrations in soils sampled during the period 2010-2014 in the Qingshan district (Wuhan City, Hubei Province, China) were used as the experimental data set. Based on a number of STSIS realizations, we assessed different kinds of mapping uncertainty, including single-location uncertainty during 1 year and during multiple years, multi-location uncertainty during 1 year, and during multiple years. The comparison of the STSIS technique vs. SIS and STK techniques showed that STSIS performs better than both STK and SIS.