Description of the multimedia and multipathway fate and exposure models IMPACT 2002 and USEtox
The IMPACT 2002 model (Pennington et al. 2005) was originally developed for Western Europe, but the Canadian nonspatial version of the model was used for this study (Manneh et al. 2010). The fate assessment includes degradation and intermedia transfer through the environmental compartments air, water, sediment, vegetation (stem, leaf, and root), and soil (surface, root, and vadose). Exposure is assessed via inhalation and ingestion (drinking water consumption and intake via agricultural products, animal products, and fishes). USEtox is a scientific consensus multimedia and multipathway fate and exposure model including both continental and global scales. Fate and exposure is assessed in a similar manner as IMPACT 2002. Further details on the algorithms used in IMPACT 2002 and USEtox can be found in Pennington et al. (2005) and Rosenbaum et al. (2008).
Development of the winter and summer Canadian models based on IMPACT 2002
Two nonspatial versions specific to the Canadian context were created for the winter and summer seasons. To predict the iF, the IMPACT 2002 model makes use of 11 physicochemical properties and 105 landscape and exposure parameters. However, it was found that these fate and exposure parameters do not share the same importance when it comes to the calculated iF (Manneh et al. 2010). Only parameters to which the iF was the most sensitive and that presented the highest contribution to its uncertainty were varied along with the summer and winter conditions. For emissions into air, Manneh et al. (2010) found that the rainfall rate, the half-lives in air and water, the soil area, and the octanol–water partition coefficient (KOW) were the most important model parameters. Whereas, for emissions into water, the Henry's constant, the half-lives in air and water, and the KOW were identified as the most important.
Seasonal temperature and precipitation data were provided from Environment Canada and Natural Resources of Canada (Ressources Naturelles du Canada 2007; Environnement Canada 2010). These data indicated an average Canadian temperature for the summer and winter seasons of +12° C and −18° C, respectively. The rainfall rate was 210 mm/y during the summer and 134 mm/y during the winter.
Physicochemical properties at 25° C were obtained from Mackay et al. (2006). The KOW and Henry's constant for the summer and winter conditions were predicted by performing an integration of the van't Hoff equation (Schwarzenbach et al. 2002), as shown by Equations 1 and 2, respectively.
Hsummer/winter is the predicted Henry's constant for the summer and winter seasons (Pa.m3/mol). H298.15 is the Henry's constant at the temperature of 298.15 K (Pa.m3/mol). is a constant (K), defined as , where ΔHaw is the standard enthalpy change of partition between the air and water phases (J/mol). R is the universal gas constant (J/mol.K). Tsummer/winter is the summer or winter temperature (K).
For KOW, the chemical-related standard enthalpy change of partition between the water and octanol phases (ΔHOW) was taken from literature data for some of the chemicals considered (refer to Table 1). For other substances for which only data on the excess enthalpy in water were available, the excess enthalpy in octanol was assumed to be equal to 10 kJ/mol and the standard enthalpy change was calculated as the difference between the 2 excess enthalpy values. This is considered to be a reasonable approximation because, for most organic compounds, the absolute value of the excess enthalpy in octanol does not exceed (+) or (−) 10 kJ/mol (Schwarzenbach et al. 2002). When no data was available for the excess enthalpies in octanol and water, the first was assumed to be 10 kJ/mol whereas the second was assumed to be equal to 0. This was another justified assumption because for many compounds, the excess enthalpy in water was found to have a small absolute value. For instance, its value was −3.6 kJ/mol for carbon tetrachloride and −1.6 for 1,2-dichloroethane (Okochi et al. 2004).
As for the Henry's constant (H), it was also predicted using the integration of the van't Hoff equation (Eqn. 2). The ratio of the derivative of the natural logarithm of the Henry's constant to the derivative of the inverse of temperature was obtained mainly from Sander (1999).
For the degradation rates in air and water, IMPACT 2002 assumes a first-order kinetic reaction. Rate constants for a temperature of 25° C were calculated using the available half-life values at this temperature (t0.5a for air and t0.5w for water, respectively) from Mackay et al. (2006). Usually, one can use the Arrhenius equation to determine the dependence of the rate constant on temperature. However, as data was not available for the chemicals considered, rate constants for summer and winter were calculated assuming that the rate constant doubles with every 10° C increase in temperature (Fogler 2006). This assumption was verified by sensitivity analyses, as mentioned later on in the article. Half-life values for the 2 seasons were then deduced.
Table 1 presents the physicochemical properties for the default temperature of 25° C as well as for the summer and winter seasons for the Canadian nonspatial version of IMPACT 2002.
Development of the summer and winter USEtox models
Two models for the summer and winter seasons were created based on the USEtox consensus multimedia model and included environmental data at the global scale. The average world global temperature was found to be 15.6° C for summer (National Oceanic and Atmospheric Administration 2008) and 12.1° C for winter (National Oceanic and Atmospheric Administration 2010). As for precipitation data, maps taken from the Global Precipitation and Climatology Center indicated that, even though continental precipitation data could change seasonally, the average global precipitation did not vary significantly (Global Precipitation and Climatology Center 2011). Therefore, the precipitation data was assumed to be the same as the annual one, i.e., 1000 mm/y. In addition to the KOW, Henry's constant, and half-lives in air and water, the vapor pressure was also calculated for these specific summer and winter temperatures. The latter was determined using the available USEtox data at 25° C and the heat of vaporization (National Institute of Standards and Technology 2008). The list of chemicals considered and their physicochemical properties for the global summer and winter temperatures can be found in Supplemental Data.
The USEtox consensus model assumes a global annual temperature of 12° C but uses physicochemical properties for a temperature of 25° C. Therefore, a 25° C model was created, assuming a temperature of 25° C with physicochemical properties calculated at that temperature, to ensure consistency.
Sensitivity analyses were carried out on the IMPACT 2002 summer and winter models to test a few key assumptions. First, the excess enthalpy in octanol was assumed to be −10 kJ/mol for the chemicals for which only data on the excess enthalpy in water were available, instead of +10 kJ/mol. This is a justified approach because for most organic chemicals, the value of the excess enthalpy in octanol is (+) or (−) 10 kJ/mol (Schwarzenbach et al. 2002). Second, the excess enthalpy in water was assumed to be +25 kJ/mol for the chemicals for which no data was available, instead of zero. This choice is justified by the fact that for some organic chemicals, such as lindane, this value was found to be equal to 24 (Schwarzenbach et al. 2002). Third, the degradation rate constants were assumed to increase by a factor of 1.25 and 4 for every 10° C increase in temperature, instead of doubling every 10° C. These factors were chosen to be higher than 1 because the rate constant increases with temperature. They were selected to see if the difference between the summer and winter rate constants would change significantly compared to the assumed factor of 2. A fifth sensitivity scenario was added to verify if results would be altered when considering seasonal agricultural and animal production. In this case, iFs were calculated using the same summer and winter models, but including seasonal rather than annual production data. Finally, the last sensitivity analysis scenario was included to verify if half-life values in air calculated on the basis of summer and winter hydroxyl radicals concentrations would alter the results obtained. Seasonal concentrations of these radicals were taken from Bousquet et al. (2005).