Introduction
The goal of the final exercise of the year was to build multiple suitability models of Trempealeau County which represented the best and worst locations for sand mines to be built. Multiple factors were used when determining the location of the mines including: geology, land use/land cover, distance to railroads, slope, water table, proximity to streams, impact on farmland, impact to populated areas/schools, visibility of recreational areas, and one area of our choice that we thought should be included in the analysis. Many of the tools used to calculate prime areas or areas of risk were in the spatial analysis tools used on raster. The main tools I used when running the analysis included: euclidean distance, reclassify, and raster calculator. The objectives of the exercise are listed below:
2. Generate a spatial data layer to meet land use land cover criteria
3. Generate a spatial data layer to meet distance to railroads criteria
4. Generate a spatial data layer to meet the slope criteria
5. Generate a spatial data layer to meet the water-table depth criteria
6. Combine the five criteria into a suitability index model
7. Exclude the non-suitable land cover types
8. Generate a spatial data layer to measure impact to streams
9. Generate a spatial data layer to measure impact to prime farmland
10. Generate a spatial data layer to measure impact to residential or populated areas
11. Generate a spatial data layer to measure impact to schools
12. Generate a spatial data layer to measure impact on one variable of your choice
13. Combine the factors into a risk model
14. Examine the results in proximity to prime recreational areas
Methods
As I mentioned above the area of study was Trempealeau County located in Western Wisconsin, some of which borders the the Mississippi River. To speed up the process of running all of these tools in ArcMap, Dr. Hupy, assigned to only use about a half or two thirds of the county as seen in figure 2 below. Some of these tools used in ArcMap take a very long time to run, therefore making the area of interest smaller, the process will speed up. Under Geoprocessing > environments > and processing extent > it was set to the boundary so all processes would output to the same area. One huge error that occurred while I ran my process was my output not being correct. As you can see in some figures below, the output of some tools I ran would stay in a rectangle and not be clipped to Trempealeau County. I found this error only to happen when running euclidean distance, any of the others tools I would run would result in the correct boundaries. I have no idea why this was happening, my processing extent was set and I tried and tried again to change around things and fix this problem, but nothing was working. I ended up just keeping the results and placing the boundary line over the rectangle. I think it defiantly affected my results but not too seriously as I compared with other students in the class.
 |
| Figure 1: Trempealeau County |
 |
| Figure 2: Processing Extent |
Geology:
The first objective was to find suitable rock or land formations. This was done by adding the geology feature class from the Trempealeau County database that was provided to us. It was found that the most desirable geologic formations for sand frac mining are the Jordan and Wonewoc formations. After converting the feature to a raster a re-class was run and ranked to determine where the Jordan and Wonewoc formations occurred. I gave Jordan and Wonewoc a 3 and the rest of the geologic formations a 1. In the first half of the model, I awarded the best locations to mine with a 3, and the worst locations with a 1. Some were even given a 0 because mining was virtually impossible there.
Suitable Land:
The next objective was to find and rank the best locations for a sand mine. Using the land cover raster downloaded from exercise 5 I ran a reclassify tool on the the different types of land uses. Determining which land uses are most suitable for sand mining was the hardest part. When ranking the certain land uses it is important to remember the amount of money it would cost to run a mine in the area. Therefore, picking an area with little forest or shrub cover would most suitable. I determined that barren land and agriculture were the most suitable and awarded a 3 to both of them. They were followed by shrub and herbaceous land, forest/wetlands, and developed land, and finally areas of water, snow or ice. These groups were given a 2, 1, and 0 respectively. After determining the ranks I ran a reclassify tool to sort the suitable land covers.
Proximity to Railroads:
Objective three was the task of finding lands closest to rail terminals. Being closer to a rail terminal means that it will be more cost efficient when transporting the sand. Using the rail terminals from exercise 7 I ran a euclidean distance to find areas closest to the one rail terminal in the county of Trempealeau. I ranked the classified the ranks based on natural breaks and found that the most cost efficient area was within 10,000 meters of the rail terminal.
Suitable Slope:
The best location for a mine is to find flat line, the next objective was to calculate slope to find the least sloped areas. I used the Trempealeau County digital elevation model from exercise 5 to calculate the slope of the area. First a raster calculator was used to convert feet to meters and then the slope tool was run. When running the tool it is common to find a "salt and pepper" effect, therefore running another tool, block statistics, to hide the "salt and pepper" effect was necessary. I used a 3 x 3 cell filter and ranked the lowest slopes the highest.
Water Table:
A sand mine uses a lot of water to cool and compact the sand. Therefore being close to a water table may make an area more valuable. For objective five, the task was to find the areas where the water table was closest to the surface. To do this data was downloaded from Trempealeau County to our folder. Because the data came in an exported coverage (.e00) it had to be converted to a raster. The topo to raster tool was used as it converted the data to be used in ArcMap. After this process was done I used the re class tool and ranked using natural breaks.
 |
| Figure 3: Model 1 |
 |
| Figure 4: Ranking system for model #1 |
The next model had a different angle is it looked at how the sand mining impacted the environment and people surrounding. The ranking system followed this model: For each criteria you want three ranks for impact (3 = High, 2 = Medium, 1 = Low). High impact will be locations where mines should not be located because of high environmental and/or community impact.
Proximity to Streams:
Sand mines have the potential to produce a lot of pollution either by erosion which can cause sediment to run off into streams or by the blowing of sand which can also end up in streams. Because almost the entire county is covered in streams running a euclidean distance would not logically make sense. I ran a select by attribute to grab the most important streams, or perennial streams, to use in my analysis. I then used a raster calculator to convert to meters and a feature to raster to transform the streams to a raster. Then a euclidean distance was run and the re class tool used natural breaks to rank the data. Again in this model 3 is the most influential to the environmental.
Impact to Prime Farmland:
Keeping farmland that is critical to farmers to area is very important, therefore, we do not want to destroy this land by building a sand mine in the area. The prime farmland feature class was obtained from the Trempealeau County database and it was converted to a raster. From her I used the re class tool to rank the types of farmland. I ranked my farmland accordingly: prime farmland had a value of 3, prime farmland if had a value of 2, and not prime farmland had a value of 1.
Impact to Populated Areas:
Sand mines can cause a lot of noise and dust traveling in the air. For this objective the task was find areas of highly populated areas to find areas where it is not okay to mine. The distance a mine needs to be away from a highly populated area is 640 meters. I used the zoning feature class and queried out all residential areas because these areas will have high populations. Then converting it to a raster and running euclidean distance. I then used the re class tool and ranked my data. 0-640 was given the number 3 and numbers 2 and 1 were broken up using natural breaks. Instead of using meters I left the data in feet and converted the 640 meters to 2,100 feet. I did this for schools and parks as well.
Impact to Schools:
Again because sand mines can cause a lot of air pollution and noise it is important to keep them away from schools. Querying out the schools from the parcels feature class was the first step. Then converting the feature class to a raster and running a euclidean distance to find the distances from the schools. Again using the, 640 meters away, I ranked my re class based on that assumption. 3 being most vulnerable and 1 being least vulnerable.
Impact to Parks:
One factor of the sustainability model was of our choice. I choose to look at parks in Trempealeau County. Using the parks feature class I converted it to a raster, using the feature to raster tool. Then running the euclidean distance tool followed again by the re class tool. AGAIN I followed the model of being 640 meters away from a sand mine and ranked my re class accordingly.
Visibility from Prime Recreation Areas:
The last objective of the model, was to run a viewshed analysis of a prime recreational area. A veiwshed analysis is the the area that can be see from the human eye when walking or standing in the area. Being in a special recreational area a person is not going to want to be able to see or hear a huge sand mining in the area. I choose the recreational area of Lakes Coulee Wildlife Area, located in the northeast area of Trempealeau County. It is home to many hunting and hiking trails, and also has a high population of trout. It is monitored by the State and an area of high importance to the county. The first step of the process was to query out lakes coulee from the recreation feature dataset. Then convert the polygon to a point because viewshed can only be run with a point or polygon feature class. The viewshed was then run using the digital elevation model used above to calculate slope along with the lakes coulee point. The re class tool was then used giving the areas visible from lakes coulee a 3 and every where else a 2.
 |
| Figure 5: Model 2 |
 |
| Figure 6: Ranks for model 2 |
First I want to explain some errors that occurred while making the suitability models. As you can see below in the images not all of maps intersected within the boundary desired. This happened when I ran the Euclidean distance tool. As I mentioned earlier, I tried and tried again to fix this and nothing worked, therefore I left it and placed the boundary line over the map. This did have a minor impact on my final products but nothing to dramatically change the outcome. Two other errors that occurred when mapping had to with my ranking system. For the first models I ranked 3 as the best locations and 1 as the worst but then reading later in the model for the 2nd part, I read that 3 is the worst and 1 is the best. It made my ranking system flip. Also I messed up the rankings for slope and proximity to streams. Giving a 3 where a 1 should be and a 1 where a 3 should be. I only noticed the mistake when mapping and it did have a an impact on my final models. I switched the rankings for figures 10 and 13 to what they should be.
 |
| Figure 8: Land Cover Suitability |
 |
| Figure 7: Geology Suitability |
 |
| Figure 10: Slope Suitability |
 |
| Figure 9: Proximity to Rail Terminals Suitablity |
 |
| Figure 12: Suitability Model 1 |
 |
| Figure 11: Water Table Suitability |
The
first suitability map, figure 12, gathered and ranked data based on what sand
mines need. Figure 12 represents the best locations with a highest number
in red. Most locations are in the North West part of the state and border
a river. Because you cannot have a sand mine on a river these ranks were
given a 0 as you can see in the dark blue above. It makes sense that the best locations of the first model would be along streams and rivers because they are close to water, the water, table and flat land. However being close to rivers can also lead to environmental problems.
 |
| Figure 13: Proximity to Streams Suitability |
 |
| Figure 14: Prime Farmland |
 |
| Figure 15: Populated Areas Suitabilit |
 |
| Figure 16: Impact to Schools Suitability |
 |
| Figure 18: Lakes Coulee Suitability |
 |
| Figure 17: Parks Suitability |
 |
| Figure 19: Model 2 Suitability |
The second model represents the worst locations to place a map base on environmental impacts. These locations are represented with high numbers in purple. Most of these locations are in the south to south east. The best locations for a sand mine with little environmental impacts are represented with low numbers and a green color. Looking at figure 19 it seems that the best locations are in the north corners and one line striping in the middle of the county. If I were to combine both models I think the best locations for a sand mine would be in the North East South East portions of the county because both models represent great locations with easy access to resources but not too environmental degrading.
Conclusion
Completing the final exercise of the year was very rewarding. Using all the information and tools run from the beginning of the semester to now was cool to see how they were all used. I think I am now an expert on certain aspects of sand mining and especially Trempealeau County. This exercise in particular was the most difficult I have ever encountered. I ran across many errors but used my knowledge and classmates to pass through them to get a results. However, it did teach a lot about the spatial analysis tools used with rasters that will be very valuable next semester and the rest of my career.
No comments:
Post a Comment