HERMES: A modelling tool for predicting mercury concentrations and fluxes in lakes

• Main Author: Ethier, Adrienne
• Format: Thesis
• Language: English
• Published: University of Ottawa (Canada) 2009
• Subjects: Biogeochemistry; Water Resource Management; Biology; Limnology; Canada; Dickie Lake; Antarc*; Antarctica
• Online Access: http://hdl.handle.net/10393/29777; https://doi.org/10.20381/ruor-19907

A general multimedia mass balance model was developed for Big Dam West, Kejimkujik Park, Nova Scotia to predict mercury (Hg) flux and fate in lakes. This model can be used as a screening-level tool by researchers with little to no modeling experience. The model requires no recalibration when applied to other lakes and few input variables (i.e., concentration of Hg in air and inflow water, lake and inflow water suspended particulate matter (SPM), lake temperature, mean depth, surface area, volume, precipitation rate, sedimentation and resuspension rate) need to be changed for any given location. Limits of this model termed "Hg Environmental Ratios Multimedia Ecosystem Sources" (HERMES) model were tested through reapplication and verification on Harp and Dickie Lake, along with Lake Ontario. The HERMES model predicts that small lakes with short water residence times and larger lakes with longer residence times are dominated by water inflow Hg concentration and atmospheric Hg concentration, respectively. For Lake Ontario, air concentrations of mercury appear to be most important. These results contrast with the currently held belief that the Niagara River is the main source of Hg to the lake. To improve model applicability to lakes with limited datasets, as was the case for many of the lakes used in this thesis, estimation methods were developed or adapted from the literature to estimate the most sensitive model input variables (i.e., water inflow Hg concentration, SPM, sediment resuspension rate, water inflow rate) when measured values are missing. Methyl mercury (MeHg) is the bioavailable form that accumulates through food webs, so estimation methods were developed or found to estimate the relative amount of methylated Hg in water inflow, water, and sediment as well. Error contributions to the model from estimation methods were tested through model application to thirty-five lakes in Ontario using three estimation methods (i.e., SPM, resuspension rate, water inflow Hg). The added value of SPM and resuspension rate estimates were assessed through comparisons with fixed values. A comparison between measured and predicted values for these lakes using these estimation methods revealed no significant difference for sediments. The HERMES model was used to derive water inflow Hg concentration values from measured sediment Hg. Regression of the derived water inflow Hg values against watershed and lake variables resulted in the following equation: log water inflow Hg concentration = (0.165 x log watershed area (km2)) + (0.102 x dissolved organic carbon (mg L-1)) -- (0.342 x log water inflow rate (m3 h-1)) + 0.000778 x direct runoff (mm yr-1)) + (0.0154 x mean lake depth (m)) + 0.492 (r2 = 0.68, p < 0.0001). A comparison between the water inflow Hg concentration estimation method (i.e., equation) derived in this study and average measured values for sixteen lakes located in different parts of the world (e.g., Antarctica, Russia, Canada) showed a deviation of only 15.7+/-18.0%, and was within reported ranges (n = 6). This was found to be a significant (p < 0.05) improvement over the previous estimation method for water inflow Hg concentration.