Frac Sand Suitability Index

Goal and Objectives

Suitable locations and environmental risks for sand mining in Trempealeau County, Wisconsin, were determined using a variety of raster geoprocessing tools to build an index model.  Two index models were created.  The first was an index model of site locations and the second was an index model of environmental risk.  The criteria was hypothetical, derived from Kent Severson and students at the University of Wisconsin-Eau Claire.

Data for this analysis was provided by Trempealeau County, Wisconsin DNR, NLCD and USGS.  Professor Christina Hupy provided a geologic map for the purpose of this project as well.

Methods

Site Location Suitability

Site Criteria included Jordan and Wonewoc geologic features, elevation, slope, Land Use Land Cover, Distance to a rail terminal, and depth to water table.  Non-suitable land cover types were excluded from possible sand mining sites.  Site criteria was ranked from 1-3 with 1 being low suitability, 2 being medium suitability and 3 being high suitability.  The following table exhibits the ranked values for each site location criteria.  The ranks were added together using Raster Calculator.  The output of this tool was multiplied by the binary data for land use land cover to exclude residential areas.

Figure 1: Frac Sand Mining Site Location
Suitability Criteria Table
Trempealeau County, Wisconsin

Figure 2: Site Location Suitability Model

The slope tool was used to determine slope of the DEM for Trempealeau County.  Block statistics were used on the output feature class to avoid the "salt and pepper effect" that is commonly exhibited in slope output data.  The results were reclassified for the elevation and slope output feature classes.  By reclassifying the data, the values could be ranked as shown in Figure 1.  Areas with a steep slope were considered low suitability because mining companies prefer low sloped areas for sand extraction.

Frac Sand occurs mostly in Jordan and Wonewoc geologic formations.  These formations occur at general elevation levels.  The Trempealeau County DEM was reclassified and Jordan formations elevations were given high suitability values, Wonewoc formations were given a medium suitability and all other formations were given a low suitability value.

The National Land Cover Database Product Legend was referenced to determine suitable Land Use Land Cover.  The input raster was reclassified; the output feature class considered Open Water, Developed Open Space, Developed Low Intensity, Developed Medium Intensity, Developed High Density, Woody Wetlands and Emergent Herbaceous Wetlands as low suitability.  Barren Land, Shrub/Scrub and Grassland/Herbaceous were considered high suitability.  Deciduous Forest, Evergreen Forest, Mixed Forest, Pasture/Hay and Cultivated Crops were considered medium suitability.  Residential and wetland land classifications were excluded from the final suitability map.

Frac sand mining facilities must transport the extracted sand via railways.  The distance from a rail terminal was used as criteria for suitable site locations for sand mining facilities.  This was done using the Euclidean Distance tool.  The output of this tool is a raster feature class that accounts for the distance from a rail terminal.  The output was reclassified and the furthest distances were given a low suitability value, medium distances a medium suitability value and short distances were given a high suitability value.

Water is needed for mining processes in order to clean the extracted frac sand.  Depth to the water table was used as criteria for site location suitability.  A water table elevation feature class was converted to raster using the Topo to Raster tool.  The data was then reclassified.  Large values were given a low suitability value, medium values were given a medium suitability value and small values were given a high suitability value.  The closer a location lies to the water table, the easier it is to reach water.

The ranked rasters were added using Raster Calculator.  The excluded LULC feature class was added to the output using Raster Calculator, so these areas would be accounted for in the final product.

Environmental Risk Suitability

Environmental risks criteria included proximity to streams, prime farmland, proximity to residential areas, proximity to schools and visibility from prime recreational areas.  Areas located within a wildlife area were excluded from locations resulting from these factors.  Environmental risk criteria was ranked from 1-3 with 1 being high suitability, 2 being medium suitability and 3 being low suitability.  The following table exhibits the ranked values for each environmental risk criteria.

Figure 3: Frac Sand Mining Environmental Risk
Suitability Criteria Table
Trempealeau County, Wisconsin

Figure 4: Environmental Risk Suitability Model

The first suitability factor for environmental risk was proximity to streams.  Perennial springs were selected from a WDNR hydro line feature class.  The Euclidean Distance tool was used to determine proximity.  The output was reclassed.  Small values were given low suitability ranks, large values were given high suitability ranks.  Water contamination is an environmental risk, so the further a location is from a stream, the better.

Because agriculture is important to Wisconsin's economy, it would not be ideal for mines to be located on prime farmland.  To determine areas of prime farmland in Trempealeau County, a Prime Farmland feature class was used.  The vector feature class was converted to raster using the "farmland code."  This allowed the raster values to account for the type of prime farmland.  The values were then reclassed to indicate suitability ranks.  Areas of prime farmland or farmland of statewide importance were ranked as a 3, low suitability.  Farmland that would be prime if it was drained and protected from flooding was ranked as medium suitability.  Non-prime farmland was ranked as high suitability.

To determine locations near residential or highly populated places, residential districts were selected from a zoning district feature class.  The selected data was then converted to raster.  The Euclidean Distance tool was used to calculate the distance from each residential zone.  Areas within 640 meters were ranked as low suitability; the longest distances were ranked as high suitability.

A parcel data set was used to select parcels that were owned by school districts.  Euclidean Distance was used to calculate the distance from schools.  The output was reclassed; long distances were considered high suitability and short distances were considered low suitability. 

Trempealeau County is home to many well-known recreational areas.  These areas are important for tourism in the area.  Suitable sites for sand mining facilities should be in locations that are not visible from recreational areas.  To determine visibility, a parks feature class was converted to a point feature class.  The Viewshed tool was ran.  This tool results in a raster with either "Visible" or "Non-Visible" values.  The raster was then reclassified as 0 for Visible and 1 for Non-Visible.

Proximity to wildlife areas was also used as criteria for suitable locations for frac sand mining facilities.  Euclidean Distance was used to calculate the distance from a Wildlife Areas feature class.  The resulting values were reclassed as high suitability for long distances, medium suitability for medium distances and low suitability for short distances.

Using raster calculator, all of the output raster were multiplied to create a suitability raster.  The values ranged from 7-18.  A 7 value corresponds with high suitability and 18 with low suitability.

Results and discussion

Figure 5: Locations of perennial streams & prime farmland
Frac Sand Suitability
Trempealeau County, Wisconsin

Figure 6: Suitability and Frac Sand Facility Locations
Trempealeau County, Wisconsin

Figure 7: Residential and School Suitable Distances
Trempealeau County, Wisconsin

Figure 8: Visibility from Recreational Areas
Trempealeau County, Wisconsin

Figure 9: Frac Sand Suitability Index Maps
Trempealeau County, Wisconsin

Conclusions
The results of the site location index model show that areas of northern Trempealeau County are best suited for frac sand mining facilities.  The environmental risk index model shows west central and northern Trempealeau County are the best locations for frac sand mining facilities.  Overall, areas near the Chippewa and Mississippi rivers are poorly suited for frac sand mining. Proximity to residential areas is a major factor in the suitability of frac sand mining site locations.

Sand Sustainability Project Part II

The goal of this Project was to study the transportation routes from frac sand from the mines to a railroad terminal.  These transportation routes could result in significant impacts on local roads.  A network analysis was performed to determine the distance travelled on local roads and the cost of the impacts for each county in Wisconsin.

Data
The addresses of frac sand facilities were used in the network analysis.  I, along with three other class members, located these addresses through geocoding as well as aerial imagery.  Each class member created a feature class of their located mines;these feature classes were then merged into one feature class in ArcGIS.  X and Y coordinates could not be found for two mines.  These mines had to be removed in order to correctly perform the network analysis.  A railroad terminal dataset was provided by our professor, Christina Hupy.  This dataset represents locations within United States for transportation terminals such as bus stations, train stations, marine terminals, and other significant transportation nodes.  Street and county data was supplied by ESRI.  Each of these datasets were crucial to the success of the network analysis.

For this analysis, it was assumed that trucks travelled between each frac sand location and rail terminal 50 times (one-way) per year (100 trips total).  It was also assumed that the cost per truck mile was 2.2 cents.

Methods
A network analysis can be performed using dialog boxes in ArcMap, but this process can be tedious and time consuming.  To perform this network analysis, a model was built in ModelBuilder, an ArcGIS application.  ModelBuilder allows the user to input feature classes and tools into a model.  When all aspects of the model have been added and defined, the model can be performed.  The use of a model presents many benefits.  The process is performed in a single step, yet is editable in order to fix any issues or to change the output of the model.

In order to produce the desired results of this project, the "Closest Facility" network analysis tool was used.  Closest Facility network analysis created a route between a frac sand location and the nearest terminal.  It was used because it most closely resembles the interaction between these features in the real-world.  The model below represents the steps taken to perform the network analysis (Figure 1).

Figure 1- Workflow created in ModelBuilder to perform network analysis

Before the distance travelled per county could be calculated, the SUM Shape Length values had to be converted from meters to miles.  This was done through the field calculator tool using the equation "SUM Shape Length / 5,280."  Once the feature lengths were in miles, the field calculator was used again to calculate the cost per county using the equation "Miles * 100 * .022."

Results
The model resulted not only in a network analysis of the frac sand transportation routes, but also the cost and travel distance per county.  These results are displayed below in Figure 2.

Figure 2-Final results of Network Analysis

Conclusions
The counties that exhibited the most distance travelled were Chippewa, Eau Claire, La Crosse and Trempealeau.  These counties exhibited the highest cost as well.  It can be concluded through these findings that as the travel distances increase between the frac sand facilities and rail terminals, the cost also increases.  Although other factors can contribute to deteriorated road conditions, it can be implied through the results of this analysis that there is a significant relationship between the frac sand transportation, road deterioration and cost per county.

Discussion
The calculations of the network analysis outputs were equally as important as the network analysis. The results of the calculations and the network analysis can be used separately to provide information on the impact of frac sand mining in Wisconsin, but the combination of the results exponentially increase the value of the research.

The real-world replication of network analysis make this tool invaluable to geospatial technology.  It is a tool that not only replicated real-world networks, but also allows the users to define parameters to expand the capabilities of these results.  Many types of network analysis are available and can be used to produce different results.  It is important to know the ramifications of the type of network analysis used; if not used correctly, results may be flawed or inaccurate.