Objective
The raster analysis would be the final step into our ongoing semester long project to locate and determine the most suitable locations for Frac sand mines within Trempealeau county. The objectives within the lab was to generate two separate rasters one which was all suitable land for frac sand mining. This first map included parameters such as land cover, geologic criteria, distance from rail terminals, slope of the area, and needed to included suitable water table depth. The second map was a map generated to look at any potential risk factors in the county. The risks that where being addressed where distances from streams, residential areas, schools, parks and also wanted to make sure the mine would not be on prime farmland. Once both maps were created we then could combine the two datasets to generate a final map showing the most suitable area for a frac sand mine to be located within Trempealeau county.Methods:
The preliminary steps to the project where to set all of our environmental settings to generate easier outputs. Some of the setting that we set where within the raster analysis and included the extent to be masked by the boundary of the southern half of the county, the cell size to match our lowest resolution raster in this case was about 98. and to set the snipping to our county to process all rasters to this outcome.Once our initial environments where set we then could start to generate the model used to locate all suitable land.
For each of the parameters we needed to run similar tools. For most we needed to convert the feature to a raster. Then needed to make sure all of the features we would be using where in the same coordinate system to have all of the units the same. This would allow us to run analysis on distance and have common units. If they where not like as in the case for the rail terminals we needed to project them accordingly. I chose to use NAD 1983 Wisconsin state plane meters.
After for most of the features we needed to either change the ranks to our personally determined set of three (3=highly suitable, 2= average, 1= low suitability). Table of all determined ranks can be found in figure 1. For the others we first needed to run Euclidian distance to determine the bets locations from the features. Once the distance was set we then could reclassify and apply suitable ranks.
 |
| Figure 1: Table of ranks for all suitability models ran within the raster analysis |
The second map was designed to show high risk areas which would equate to low suitability. All areas addressed were ranks in same order but the criteria was based on risk factors, for example a highly suitable area represented by a 3 in rank would be equivalent to low risk area. This rank system allowed us to add all ranked values when combining both maps. As in the case for the first model the steps for each parameter were very similar. We first needed to generate a raster out of the feature class, then needed to run either a distance or reclassification based on previously determined scale. Both final models can be found in figure 2 and 3.
Keep in mind all of the ranks were determined individually by each student based on our own knowledge and may not be the best rank for each parameter because of these maps are complete opinionated and should not be used as real suitable locations.
 |
| Figure 2: suitability model used to determine best location based on land cover parameters. Also includes raster calculation for final best mine locations. |
 |
| Figure 3: Risk model generated with parameters of high risk areas to keep mines away from. All reclassification where placed for most suitable equating to lowest risk. |
Results
After analysis was run on both of the models there was a series of locations generated to show suitable land for the mines to be located. The total raster calculation for both models equated to the bets location being a total of 15 possible points. I then chose to reclassify the maps into three categories to simplify the display. I chose to rank the top three calculation to be most suitable the nest four to be average suitability and the remaining to be low suitability. The first map created (figure 4) displays all suitable areas based on land cover type. the second (figure 5) displays all land cover that is low risk and therefor high suitability. Finally after adding the two raster's together we generated a final map displaying the bets locations for frac sand mines in Trempealeau County (figure 5). |
| Figure 4: displays the raster output from both of the suitability models displaying the best ranked location as highly suitable in dark brown colors |
 |
| Figure 5: final map output locating the optimal location for frac sand mines combing the two suitability models. The best locations are displayed in red. followed by average areas in brown and last all other locations in yellow. |
Sources
Trempealeau county database
Esri online help
NLCD database
University of Wisconsin Eau Claire
