Probabilistic indicators for soil and groundwater contamination risk assessment

Deterministic assessments of whether, when, and where environmental safety thresholds are exceeded by pollutants are often unreliable due to uncertainty stemming from incomplete knowledge of the properties of environmental systems and limited sampling. We present a global sensitivity analysis to rank the contribution of uncertain parameters to the probability, P, of a target quantity to exceed user-defined environmental safety thresholds. To this end, we propose a new index (AMAP) which quantifies the impact of a parameter on P and can be readily employed in probabilistic risk assessment. We apply AMAP, along with existing moment-based sensitivity indices, to quantify the sensitivity of soil and aquifer contamination following herbicide glyphosate (GLP) dispersal to soil hydraulic parameters. Target quantities are GLP and its toxic metabolite aminomethyl-phosphonic acid (AMPA) concentrations in the top soil as well as their leaching below the root zone. The global sensitivity analysis encompasses six scenarios of managed water amendments and rainfall events. The biodegradation of GLP and AMPA varies slightly across scenarios, while leaching below the root zone is greatly affected by the assumed hydrologic boundary conditions. AMAP shows that, among the tested uncertain parameters, absolute permeability, air-entry suction, and porosity have the greatest impact on GLP and AMPA probability to pollute the aquifer by exceeding the aqueous concentration thresholds. Our results show that AMAP is effective to thoroughly explore time histories arising from model-based predictions of environmental pollution hazards. The proposed methodology may support informed decision making in risk assessments and help assessing ecological indicators through threshold-based analyses.