Temporal variation of global sensitivity analysis for biodegradation model using iTOUGH2

Monod kinetic equation is widely used to describe the microbial biodegradation process of organic groundwater contaminants. Due to a large number of parameters in the Monod equation, sensitivity analysis can be used to identify the importance of the parameters, which is helpful for parameter inversion and microbial biodegradation process understanding. However, most existing sensitivity analyses only focus on average sensitivity value and space variation, seldom considering sensitivity over time. In this paper, a one-dimensional sand-column experiment of toluene aerobic biodegradation was taken as an example. Based on iTOUGH2 global sensitivity analysis (GSA), we used Morris and Sobol’ methods to analyze degradation process parameters and experimental parameters changing with time. The results show that the aerobic degradation ability of microbes first increases and then decreases over time, which leads to the same trend for the degradation process parameters sensitivity. Sobol’ Index of the maximum substrate degradation rate k varies from less than 10% at early-stage to at most 62% at middle-stage and decreases to 49% at late-stage. The parameter interaction effect also varies similarly to sensitivity. We used the differences between Sobol’ Total Sensitivity Index and Sobol’ Index to describe parameter k’s parameter interaction effect, which in this case are both at around 0% for early and late stages and rises to 6% at middle-stage. Through these time-varying analyses of sensitivity and parameter interaction effect, we find that observations in the late-stage of the experiment are more sensitive to degradation process parameters and the parameter interaction effect in the late-stage is smaller, so selecting the observations in the late-stage is more beneficial for the degradation process parameters inversion. Besides, to avoid the possible increase of observations error,the one caused by improper experimental control, the experimental conditions should be more strictly controlled in the early-stage than in the middle and late stages because the experimental parameters are more sensitive in the early-stage.