Objective 2: To develop the methodology required to quantify the factors used in the predictive model.
One or more layers were created for each factor identified in Objective 1. These layers include soil hydraulic conductivity (ksat), soil group, agriculture landcover, urban landcover, water landcover, natural landcover, slope, and weather (temperature and precipitation). The methods used to create these layers are shown in Figure 3.
Figure 3. Flow chart illustrating the processes used to create the various factors layers
Before these layers could be quantified for the sampling points, the upstream area contributing to these points needed to be determined. The upstream contributing area was determined as this is the land that water flows across and ultimately ends up in the stream. The characteristics of the land within the contributing area would be the factors that impact the N and P levels at the sample points. Determining the upstream contributing area was done by creating overlapping watersheds for each point. Overlapping watersheds were created for each point using tools in ArcHydro. The steps to creating the overlapping watersheds are shown in Figure 4.
Figure 4. Flow chart illustrating the process to create watersheds.
For the predictive model, the factor layers needed to be quantified within each watershed and joined to the sampling points attribute table. The quantification methods for the factors varied and are shown in Figure 5. The hydraulic conductivity and slope were quantified by averaging the ksat values or the slope values within each watershed. The soil group and land cover layers were quantified by calculating the percentage of each soil group or each land cover type within each watershed. The weather layer was linked to the sampling points through spatially joining the nearest weather station to each sample point. The attribute table of that sample points shapefile was then exported as a table. This table was then joined to the temperature and precipitation data using python. The source and scale of the raw data files used for this analysis can be found in the appendix.
Figure 5. Flow chart illustrating the processes used to index the factor layers.
This methodology was followed to quantify the factors in the contributing drainage area of the Ontario Water Quality Monitoring Network sample points. The output table from this analysis was used to develop and validate the model (explained in Objective 3). This methodology was then repeated using 300 randomly placed points along the stream network in the Lake Erie watershed. This process was repeated as randomly placed sample points spread throughout the whole watershed were required to generate predictions for all stretches of streams. The output table from this analysis was input into the finished predictive model and was used to predict the water quality at each of the sample points.