Objective 3: Develop an integrated GIS-based MCE model, unifying the USLE and DRASTIC models to determine the potential of an area to contaminate surface and groundwater
The USLE, although developed decades ago and conceived during the dust bowl of the 1930s (Renard et al. 1994), is one of the most widely used soil erosion models in use today. There is no shortage of literature on the USLE, and due to the ease of comparability of its applications, the Ontario Ministry of Agriculture, Food and Rural Affairs (OMAFRA) has adopted the USLE as a tool in their Nutrient Management Program (Stone & Hilborn 2000). It is a highly adaptable model, as shown by Machito Mihara et al. (2005), where it was successfully used to estimate soil nutrient losses. The parameterisation of the USLE was accomplished following the model guidelines published by Stone and Hilborn (2000). Although there have been other parameterisations of the ULSE, Stone and Hilborn's (2000) method is specific to the farming practices and soil characteristics found in Southern Ontario. For example, the R factor, as discussed in Objective 1, is the precipitation factor and is parameterised based on rainfall intensities found in Southern Ontario; likewise the values assigned to the C and P factors are tailored specifically to crop types and management practices common in the study site. The full parameterisation of the USLE can be found in Table 1. Additionally the LS parameterization in the USLE was calculated using Equation 4 in conjunction with values found in Table 1.
LS = [0.065 +
0.0456(slope) + 0.006541(slope)2] (slopelength
÷ 22.1)NN
(4)
Where slope it the percent grade of the slope, and slope length is the
length of the slope in feet. NN can be found in Table 1 and is based on
the percent grade of the slope.Table 1: Parameterisation values used in the USLE as outlined by Stone and Hilborn (2000)
These USLE parameters all play a role in modelling the factors which influence the soil erodability. The USLE accounts for field characteristics which relate to the precipitation (R), soil (K, L, S), and land management practices(C, P). In turn, these parameters link directly to the mechanisms responsible for the three distinct stages of water erosion, detachment, transport and deposition (Brady & Weil, 2010). The factors are given values which range accordingly based on the factors inclination to promote or hinder soil erosion in a specific area (Table 1). Thus, in the USLE a lower factor value signifies a reduction in erosion compared to a higher factor value. For example, as precipitation (R) becomes more intense so will the amount and size of raindrops hitting the surface, resulting in an increase in the detachment of soil (Franzluebbers, 2002). Thus, the different counties within the study area are weighted accordingly based on occurrence, volume, intensity and seasonal distribution of rainfall (Stone & Hilborn 2000, Erdogan et al. 2006). The soil erodability factor (K) is based primarily off the soil textural class which relates to infiltration and structural stability (Brady & Weil, 2010). Soils with finer particles which stick together, such as clay, have smaller values as they are less erodible than soils with poor structure and less cohesion, such as sand (Brady & Weil, 2010). In all cases an increase in organic matter decreases the R parameter value as organic matter works to increase the structure and infiltration, effectively decreasing the erodibility of a soil (Franzluebbers 2002). The topographical soil factor (LS) is a combination of slope and the length of the slope. An increase in slope is directly correlated to less stable soils as well as the tendency of water to move faster and more vigorously over these surfaces. This increases the potential for the detachment of particles, while a longer slope length also increases the surface area being used for transport (Renard et al. 1997) justifying the positive correlation of values for Slope Length in Table 1. Land management factor C, correlates to the crop type. Cropping year-round reduces the amount of time a soil is left bare compared to annual crops which when harvested can leave the land bare for extended periods (Brady & Weil, 2010). The parameter values are increased for vegetation types which offer the most protection. Any surface where erosion is less likely to occur (urban, bedrock, or dense forest) was given a value of zero, while bare land has no preventative vegetation and was given a value of 1. The final value for the USLE, P, relates to any crop or tillage support practice which would prevent erosion (Stone & Hilborn 2000, Erdogan et al. 2006). Tillage or crop cover practices can alter the erosion on land, such as reduced tillage which preserves soil structure (Brady & Weil, 2010), altering ploughing technique and increasing crop residue coverage to help protect the land can also reduce erosion (Zhang et al. 2007). Although this is an important part of the USLE little to no literature or spatial data was available to assign a value for this parameter. Therefore, a value of 1 was used for the entire watershed, implying that there are no support practices currently in place on this land.
DRASTIC as described in Objective 2, is a groundwater vulnerability assessment tool developed in 1987 by a team of researchers at the USEPA (Aller et al. 1987). Although there are several groundwater contaminant models, the factors outlined in Objective 2 that constitute the input parameters for the DRASTIC model best fit the available data for the study area discussed in this report. Additionally, DRASTIC is one of the most widely used models applied to investigate below surface contamination potential (Merchant 1994). A number of sources were used to determine specific parameterisation of the DRASTIC model for use in this study; these parameters are outlined in Table 2. Values of between 1 and 10 are assigned to each of the input factors, 10 being assigned to factors likely to increase contaminant potential and 1 being assigned to factors less likely to increase aquifer vulnerability. The parameterisation as well as the weighting (Table 3) was accomplished following work done by Al-Zabet (2002) and previous to that the rating system as developed by Aller et al. (1987). The DRASTIC model was designed to be applicable anywhere in the entire United States; due of the geologic and geomorphic similarities between Southern Ontario and the northern United States, it is believed that no further alteration of parameters was warranted.
Table 2: Outline of parameterisation values used in the DRASTIC model (Aller et al. 1987)
Contaminant travel from a point source to groundwater occurs through the vadose zone via infiltrating water, and hydrodynamic dispersion (Bear & Cheng 2010). Additionally as contaminants persist in the environment they become subject to attenuation processes. The values of the DRASTIC model factor take these spatial and temporal processes into consideration. Aivalioti and Karatzas (2006) have shown that contaminant concentrations decrease exponentially with depth. This is due to the exponential relationship between concentration and the dispersion coefficient which is used to calculate contaminant transport (Fetter 1980). Additionally the dispersion coefficient is based on hydraulic conductivity of the zone media (Bear & Cheng 2010). Recharge rate affects the temporal variability of the contaminant. In areas of high recharge, the contaminant plume can quickly be washed into the aquifer, which occurs much faster than the dispersion rate. This relationship is also related to the depth to groundwater, as this increases the travel distance. Contaminant travel is highly correlated to storm events, with contaminant movement exponentially dropping off after storm events (Bosch & Truman 2002). The aquifer media, soil media, and vadose zone media all follow a similar subjective rating scheme. Aller et al. (1987) based this scheme on the increasing relative porosity of materials. Highly porous or fractured materials will more readily transport contaminants, both spatially and temporally. Paralange et al.(1999) noted that the determination of such factors is highly variable both spatially and in scale, additionally Aller et al. (1987) admits this variability, however, no modifications were made to the paramertisation of factors as there is no reason to believe the geology of Southern Ontario has atypical features which would bias the model.
As mentioned above the MCE consists of two factors, a surface factor which is based on soil erosion and a groundwater vulnerability factor. The USLE is effective at locating sites vulnerable to soil erosion but it lacks a transport function which is essential for determining the potential for contaminants to enter surface waters. By incorporating a distance to streams component to the USLE it was possible to determine the potential for contaminants to enter surface waters and in turn created a USLE-plus-Transport (USLETrans) layer. The outputs of both the USLETrans and DRASTIC models were converted to a percent, with 0 representing no erosion, or groundwater vulnerability and 100 representing high rates of soil loss or high groundwater vulnerability. The standardized USLE was then entered into the MCE along with the standardised DRASTIC output, accounting for the weights assigned in Table 3 (Aller et al. 1987). The surface and groundwater models have equal weights, both account for 50% of the overall MCE model. Both surface water and groundwater is of importance for the specific study area chosen and both were determined to be equally significant. The output results of the integrated MCE analysis, as well as the standardised USLETrans and DRASTIC factors were classified based on low, medium, and high contamination potential for easy visualisation.
Table 3: Values for Drastic weights (Aller
et. al. 1987)
