Exercise 8: Raster Modeling

Introduction

For this assignment the goal was to look at suitable sand mining locations in southern Trempealeau County, Wisconsin; sand mining risk model; and ultimately use the two models to determine locations for sand mining with minimal environmental and communal impact.

Methods/ Results


Suitability Model

The first step to assessing suitability for a frac sand mine was to determine the different criteria being assessed. This criteria consisted of having Wonewoc or Jordan formation quartz sands, suitable landcover, nearby railroad terminals, a gradual slope, and a shallow water table. This way, the mine has the right type of sand, easy removal of top soil, minimal travel distance from the mine to a railroad terminal, an appropriate surface incline at the site, and easy access to water, all of which are crucial to operating a successful frac sand mine. Figure 1 shows the model for calculating these factors into one *comprehensive suitability model (*shown in the far right corner of figure 2).

Figure 1: Suitability analysis ModelBuilder tool.

Figure 2 shows a map of all criteria that went into determining suitable areas within the study area.

Figure 2: Map showing different characteristics of suitability as well as comprehensive suitability model.

In figure 2, for the "Suitable Geology" map, the prime frac sand values represented Wonewoc and Jordan quartz sand formations, while the other values were non-suitable geologic formations.

In the "Landcover" map, suitable values were areas of either barren land (rock/sand/clay), grassland/herbaceous, pasture/hay, or cultivated crops as these were landcovers that could easily be temporarily re-purposed for frac sand mining. The poor values were areas of woody wetlands, shrub/scrub, forests, developed land, or open water as these landcovers would be difficult to access the underlying sand deposits.

For the "Distance to Rail" map, the distances were calculated using the euclidean distance tool and were reclassified using the "natural breaks (jenks)" classification to produce three proximity values.

In the "Suitable Slope" map, the slopes were calculated with the "slope" tool and, again, reclassified into steep and gradual classes using the "natural breaks (jenks)" classification method. The gradual slopes class shows optimal areas for frac sand mining as extracting resources across a steep slope is more difficult than mining across a gradual slope.

For the "Water Table Elevation" map, a water table elevation contour coverage file was used to create a digital elevation raster model (DEM) using the "topo to raster" tool. The resulting DEM was the basis for a following rank and reclassification, using the "natural breaks (jenks)" classification method, which produced three classes, ranging from shallow to deep water table elevation. Due to the amount of water used in frac sand mining, it is important for a mine site to have easy access to water (usually from wells reaching down to the water table) and having a shallow water table depth facilitates this need better than a large water table depth.

Lastly, the "Calculated Suitability" map was created using the "raster calculator" tool to obtain the sum of all the rasters used in this analysis. Then that sum was multiplied by the "Landcover" raster to produce the final map. Poor areas, shown in blue, represent areas of  non Wonewoc/Jordan geologic formations, non-suitable landcovers, far away from railroad terminals, steep slopes, and/or deep water tables. Best areas, shown in brown, represent areas of Wonewoc/Jordan geologic formations, suitable landcovers, near railroad terminals, gradual slopes, and/or shallow water tables.

Impact Model

Once suitability was calculated, the next step was to calculate potential risk of environmental impacts. The criteria assessed for this model included: How far away the mine site is from perennial streams, how much impact will the mine have on farmland, how far away the site will be from residential areas, how far away the site will be from schools, and how far away the site will be from sanctioned wildlife areas. The environmental and cultural impacts of these factors were then used to create a comprehensive* impact model (*shown in lower right corner of figure 4) in Model Builder (figure 3).

Figure 3: Impact analysis ModelBuilder tool.

Figure 4 shows a map of all the criteria that went into determining the environmental and cultural impacts of frac sand mining.

Figure 4: Map showing different characteristics of impact/ risk as well as a calculated impact model.

In figure 4, for the "River Proximity" map, the blue values (3) represent high-risk areas that are within 984 meters of any primary or secondary perennial streams over land or into water. The brown values (1) represent low-risk areas that are greater than 4,048 meters away from any primary or secondary perennial stream over land or into water. These values were chosen due to their presence year-round and were determined by using the natural breaks (jenks) classification method.

In the "Farm Proximity" map, the blue values (3) represent high-risk areas that are within 775 meters of any primary farmland. The brown values (1) represent low-risk areas that are greater than 1,946 meters away from primary farmland.

For the "Residential Proximity" map, the blue values (3) represent high-risk areas that are within 640 meters of residential areas. This high-risk distance was determined by the minimum distance requirement for all mines according to standard zoning laws and is what is known as a "noise/ dust shed." The brown values (1) represent low-risk areas that are greater than 28,594 meters away from residential areas. The medium and low-risk values were re-adjusted using the natural breaks (jenks) classification method after the high-risk value was set. The residential areas were derived from the Trempealeau County Zoning Districts layer. The residential areas selected consisted of: Residential Public Utilities, Residential - 20, Residential - 8, and Rural Residential fields.

In the "School Proximity" map, the blue values (3) represent high-risk areas that are within 640 meters of all schools. These high-risk distance, again was chosen due to the minimum distance requirement for residential areas. The brown values (1) represent low-risk areas that are greater than 10,875 meters from all schools. Again, the medium and low-risk values were re-adjusted using the natural breaks (jenks) classification method after the high-risk value was set.

For the "Wildlife Proximity" map, the blue values (3) represent high-risk areas that are within 15,298 meters of any official wildlife area. The brown values (1) represent low-risk areas that are within 26,790 meters of any designated wildlife area. These values were also determined by the natural breaks (jenks) classification method.

Lastly, the "Calculated Impact" map was produced by using the Raster Calculator tool to add up all of the criteria. High Impact areas, shown in blue, represent areas that are: within 984 meters of any primary or secondary perennial stream over land or into water, within 775 meters of any primary farm land, within 640 meters of a school or residential area, and/or within 15,298 meters of any designated wildlife area. Low Impact areas, shown in brown, represent areas that are greater than: 4,048 meters away from any perennial stream, 1,946 meters away from any primary farmland, 28,594 meters away from a school or residential area, and/or 26,790 meters away from any designated wildlife area.

Figure 5 shows an excel table that gives detailed distances for each risk characteristic displayed in figure 4.

Figure 5: Impact data values.

Once a suitability and impact model were made, the raster calculator tool was again used to add up the two models and create a final risk model. The data flow model is shown in figure 6 and the resulting map is shown in figure 7.

Figure 6: Calculated risk ModelBuilder tool.

Figure 7: Calculated risk and suitability model.

Viewshed Model

One thing much of western Wisconsin relies on for income and overall enjoyment by its citizens, is recreation. The appearance of a frac sand mine could be potentially harmful to certain tourist attractions and therefore is important to avoid mining in these areas at all costs. For the next model and map, a scenic bike route, chummily named "Breathtaking Highs and Lows" by the Trempealeau County tourist site*, was selected and the "viewshed" tool was run based on that loop. Figure 8 displays the data flow model and figure 9 displays the resulting map.

Figure 8: Viewshed ModelBuilder tool.

Figure 9: Viewshed map.

The Trempealeau county digital elevation model, in conjunction with the bike loop, was used as inputs in the "viewshed" tool. In figure 9, blue values represent non-visible areas from the bike loop, whereas beige areas represent visible areas from the bike loop. Although this map was not incorporated into the risk and suitability models, it very well could be and would further help the mine site selection process.

Discussion

Looking at the results it's clear that raster analysis can be extremely beneficial in determining site location for something such as a frac sand mine. Seeing all of the factors come together and be used to develop a map that shows prime locations and poor locations, really puts the importance and power of this technology into perspective. Figure 9, which shows the viewshed of a bike trail in Trempealeau county also supports the claim that analysis is never fully complete. The viewshed analysis wasn't even factored into the final risk and suitability model, but very well could've been, and in doing so, would've further narrowed down the areas suitable for frac sand mine locations. What about network analysis? Air particulate modeling? Return on investment analysis? These additional analyses, along with countless others, could all be factored into a comprehensive risk and suitability model, but were not. So then it's a matter of what to include and what not to include. For the purpose of this assignment, as well as time constraints, I do feel as though the factors chosen for this analysis were sufficient and relevant to the project's end goal - to find suitable and low impact locations for frac sand mines in southern Trempealeau county.

I found the raster calculator tool to be extremely useful and a fascinating method of analyzing multiple factors. The tool works by overlaying multiple rasters with classified values to create one raster with all of those inputs factored in. In figure 7, the effectiveness of this tool is shown, as the areas that are more suitable and have less impact (specifically in the southern and western portions of Trempealeau county), correlate heavily with the layers that went into making the final model.

Overall, the final risk and suitability raster is effective in serving its purpose. It would make for a great map to be used by mining companies, resources management companies, or government agencies in locating acceptable and beneficial locations for frac sand mine sites in Trempealeau county. Although there is a potential for more investigation to be done for this particular topic, the analyses completed on the given data does a good job of showcasing the power of GIS in site location and produced a well-thought-out model of good mine locations.