Objective 2: To determine the importance of characteristics and assign relative weights for MCE.
Geomorphological characteristics of catchments influence their recovery states by controlling delta habitat structure and nutrient and organic inputs to delta areas (Wesolek et al, 2010). Specifically, vegetative cover (NDVI), wetness, and slope are characteristics that are particularly important for dictating the rates of chemical and physical recovery of these delta areas (Wesolek et al, 2010). In addition, stream power has impacts on the concentration of dissolved materials in the water (Caissie et al, 1996).
ArcMap and WhiteboxGAT are utilized to calculated values of catchment characteristics for seven lake sub-catchments, which are displayed in Table 1.
In ArcMap, slope is derived from the Slope tool in the Spatial Analyst Tools extension. Using the Extract by Mask tool in ArcMap, each sub-catchment is extracted and its average slope is derived from the mean of the sub-catchment slope data.
In WhiteboxGAT, spectral bands 4 (red) and 5 (near-infrared) from Landsat8 are used as inputs to compute the NDVI for the study site. After an NDVI raster is created, this raster is opened in ArcMap. Using the Extract by Mask tool in ArcMap, each sub-catchment is extracted and its average NDVI is derived from the mean of the sub-catchment NDVI data.
Catchment Wetness Index and Stream Power
In WhiteboxGAT, the digital elevation model (DEM) created to compute slope is used as an input to compute the flow accumulation grid using the tool Fd8FlowAccumulation, which outputs a specific contributing area (SCA) raster file. Then, the SCA file and DEM are used as inputs to compute the catchment wetness index and stream power using the WetnessIndex tool and RelativeStreamPowerIndex tool, respectively. After each of these raster parameters are created, they are opened in ArcMap. Using the Extract by Mask tool in ArcMap, each sub-catchment is extracted and its average wetness index and stream power is derived from the mean of the sub-catchment average wetness index and stream power data.
Table 1: Values of catchment characteristics for Baby and Daisy lake sub-catchments.
Next, a regression model is developed in Rstudio in order to determine which of the characteristics has the most significant impact on water chemistry elements. Since there are twenty water chemistry elements in total, twenty regression models are created. For each model, there is an order of characteristics values from greatest to smallest. To illustrate this step, the first regression analysis between catchment characteristics and Alkalinity concentration has the following code and results in Table 2:
mychemmodel=lm(data=water_chem_avgs,Alkalinity~water_chem_avgs$`Slope (degrees)`+water_chem_avgs$NDVI+water_chem_avgs$Wetness+water_chem_avgs$`+water_chem_avgs$`Stream Power`)
Table 2: Estimated values of catchment characteristics generated from the first regression model with Alkalinity.
Based on absolute value, NDVI is the most important characteristic to water chemistry, following by Wetness, Slope and Stream power. This order is also the most dominant one throughout the analysis.
Multi-criteria evaluation (MCE) in GIS refers to assigning land to suit a certain objective based on the analysis and investigation of various conflicting factors (Eastman, 2005).Two different factors for MCE analysis are constraints and factors. While the former eliminates alternatives completely, the latter affects the possibility of the relative objective so alternatives are not limited (Lopez-Marrero et al, 2011). Criteria standardization allows users to convert data to numeric scales which then are comparable to each other (Eastmen, 2005).
In order to calculate weights for relative characteristics, a pairwise comparison is utilized first (Table 3; Table 4). Next, the Consistency Ratio is calculated to ensure ranks are consistent (Table 5).
Table 3. A typical nine-point rating scale.
Table 4: Pairwise comparison of criteria factors using 1-7 scale.
Table 5: Consistency Ratio (CR) Calculation.