Impacts of acid deposition at Plastic Lake: forecasting chemical recovery using a Bayesian calibration and uncertainty propagation approach

Abstract

Given the importance of models in the development of environmental polices it is necessary to assess the uncertainty introduced by model parameterisation and its impact on predictions. In the current study, an uncertainty framework designed to perform automated calibrations and developed for use with the Model of Acidification of Groundwater in Catchments (MAGIC) was applied to Plastic Lake, a long-term study site in Southern Ontario, Canada. The primary objectives were to investigate the chemical response of soil and surface water at Plastic Lake to proposed acid (sulfur and nitrogen) emissions and assess the use of the framework at a regional level. Despite the relatively high amount of uncertainty associated with many of the model parameters, calibration resulted in relatively narrow parameter convergence. The importance of time-series stream data was clearly evident, with uncertainty decreasing with more observation years. The forecast improvements in stream Acid Neutralizing Capacity at Plastic Lake from–40 μeq/L in 1988 to 14 μeq/L in 2060 had 5 and 95% confidence bounds of–3 and 29 μeq/L, respectively. Despite the limited availability of soil chemical data in Ontario, the approach applied at Plastic Lake is viable on a regional basis given the abundance of water chemistry data.