Modeling the multimedia fate dynamics of γ-hexachlorocyclohexane in a large Chinese lake

Long-term annual dynamics from 1984 to 2020 (simulation #1) and seasonal variation from May 2010 to February 2011 (simulation #2) of γ-hexachlorocyclohexane (γ-HCH) in various environmental media in Lake Chaohu, China were simulated with an already developed fugacity-based level IV Quantitative Water Air Sediment Interaction (QWASI) model (Kong et al., 2012). The model was modified, as a fish sub-phase was included. Also the emission flux was added to study the impact of the lindane prohibition in simulation #1. Sensitivity analysis was conducted for both static and dynamic parameters, while in uncertainty analysis, both basic Monte Carlo and Bayesian Markov Chain Monte Carlo (MCMC) methods were undertaken for simulation #2 and the results were compared. Simulated data were consistent with the observations in simulation #1. Seasonal patterns in various media were also successfully modeled in simulation #2 and factors leading to this seasonality were discussed. Atmospheric advection input was the main source. In simulation #2, approximately 36 kilogram (kg) of γ-HCH in Lake Chaohu was removed per year. In addition, 31 kg of γ-HCH was added to Lake Chaohu by air–water interface fluxes, and 13 kg of γ-HCH was added to the sediment by water–sediment interface. Sensitivity of static and dynamic parameters was discussed. Uncertainty analysis by the basic Monte Carlo method for simulation #1 showed that the dispersion for each media was less than two orders of magnitude. Higher dispersions in fish and two sub-phases of the sediment were attributed to a larger variation in the relevant parameters. The MCMC method for simulation #2 eliminated 77% of the model true uncertainty in water ascertained by basic Monte Carlo method and significant elimination in other phases can be speculated. It is suggested that rather than calibrating the model, the main function of the MCMC for fugacity model should be to avoid overestimating uncertainty in model prediction.