Raster Modeling

Objectives

    The goal of this lab was to use various raster processing tools to build models for both sand mining suitability as well as the impact on the environment it could potentially have for Trempealeau County, Wisconsin. These models would take a look at the best locations for sand mining with minimal environmental and community impact.

Data sets and sources

    The data sets that were used for this project came from the Trempealeau County geodatabase along with the USGS. To narrow down the output of these models I only concentrated on the southern 2/3's of Trempealeau County. This would give me a more distinct accuracy of areas that this portion of the county could hold, and flourish with a frac sand mine. 

Methods

    The first model I created was the suitability model for the region. The data that was included in this model was the NLCD land cover type, geologic unit, slope of the region, distance to rail terminals, and the water depth table. I had to convert all of these from a vector dataset to a raster, which would allow me to calculate the suitability much easier. I also had to reclassify all of these raster datasets. The reclassification was a scale from 1 to 3, with 1 being low suitability and 3 being high suitability (figure 2).

Fig. 2 Suitability model for Trempealeau County

    The first criteria that I looked at was the NLCD land cover type. By using the reclassify tool, I was able to determine what types of land cover were best for a sand mine. The best land cover for a sand mine was the barren land or herbaceous land cover. These lands were mostly open land which wouldn't take much to put a new mine in these areas. (figure 3a,b)

Fig. 3a Reclassify tool for land cover

Fig. 3b land cover suitability for Trempealeau County

   
    The second criteria of the suitability model was the geologic unit for Trempealeau County. The best Frac sand in the area is found in either the Trempealeau group or the Wonewoc formation. There are many other different distinct groups in this region, but they are considered not suitable. To classify these two formations, I only used a numbering system of either (1) for suitable or (0) as not suitable. (figure 4a,b)

Fig. 4a Reclassify tool for the geologic unit

   

Fig. 4b Suitable land for Trempealeau County

    The third criteria that I took into consideration for the model was the slope of the area. Seeing as it wouldn't be feasible to put a sand mine on the side of a steep slope, I had to take this into consideration when calculating this criteria. The most suitable slope for a sand mine in this area was less than 10 degrees. Anything above 10 degrees and it would be difficult to have a sand mine in that area. It would cause stress on everyone involved if the sand mine was located in a steeply sloped area. (figure 5a,b)

Fig. 5a Model breakdown of the reclass slope

 

Fig. 5b Suitable land based on slope breakdown

    The fourth criteria on the suitability model is the distance to a rail terminal. Seeing as the frac sand is loaded onto railways for easy shipping, it would make sense to have a new sand mine located in close proximity to a railway. The most suitable railways for a new sand mine would fall within 12,500 meters of the sand mine. Although it would awesome if the sand mine could be located directly next to a railway, this isn't always the case. With a railway within the highly suitable area, trucks can always transport the sand to these railways, or if it is close enough they can all use a conveyor system to transport the sand to the rails. (figure 6a,b)

Fig. 6a model used to create rail distance

   

Fig. 6b suitable land based on distance from rail terminals

    The final criteria of the suitability model is taking the water table depth into consideration. Water table depth is important because once the sand is mined it needs to be washed to remove fine particles. This means that the closer to the surface the water table is to the surface, the easier the access is to water for washing the sand. A high suitable water table would have a depth from 0-820 meters deep. This would allow for companies to drill a well for the wash plants. (figure 7a,b)

Fig. 7a model used for the water table depth

   

Fig. 7b Suitable land based on water table depth

    After taking all of these criteria into consideration, I could then use the raster calculator tool allowing me to generate a suitability index for the entire model. (figure 8a,b)

Fig. 8a raster calculator tool used to create suitability model

   

Fig. 8b Suitability Model

    The second model that I created was the sand mine risk model index. There were four different sets of criteria I used when creating this model, distance to streams, areas of prime farmland, distance to residential neighborhoods, and distance to schools. Just like the first model, I also had to reclassify all the criteria into categories of 1 being high risk, to 3 being low risk. This would allow me to create the risk model index much easier. (figure 9)

Fig. 9 Risk Classification Ranks

    The first criteria I used was the distance to streams. This would allow me to show areas that are within close proximity to streams. If a mine is located to close to a stream, it could pose a problem with contaminating the stream. The first step I did when calculating the stream distance was a select by attributes tool. This allowed me to only look at the streams I wanted to base my model off of. I decided to only look at perennial streams as they are around year long and pose the highest threat of being contaminated. After creating a feature class of just of the perennial streams, I could then run the Euclidean distance tool. I wanted to have this tool broken down into three categories. The closest category would have a value of "1", which would pose the highest threat. The farther from these streams the better. (figure 10a,b)

Fig. 10a Risk model based on streams

   

Fig. 10b Risk based on stream proximity

    My second criteria of the risk model was looking at prime farmland. This data came from the Trempealeau County database. This would allow me to see which areas in the county were considered to be prime farmland, and which areas are not. A sand mine shouldn't take priority over prime farmland seeing as it is just as important if not more important to continue to have agriculture in the region for producing food and income. First I looked at the attribute table and create a new field for the re-class values. I wanted all the prime farmland to have a value of "1" and all other areas to have a value of "0". This would allow me to show the areas of the prime farmland. I used the select by attributes tool to do this conversion. After selecting all the prime farmland, I had to then use the polygon to raster tool to create the raster for this dataset. Since it was in a different projection then the rest of my datasets, I then used the define projection to change its coordinate system. (figure 11a, b)

Fig. 11a Risk model based on prime farmland

   

Fig. 11b Risk based on prime farmland

    The third criteria that I used in this model was the distance to residential neighborhoods. I used the zoning distance feature class from the Trempealeau County database to base my findings from. Once again, I added a new field and used the select by attributes tool to give all the zones of residential a "1" value and all the other zones a "0" value. I then used the feature to raster tool, which I can run Euclidean distance from. This allowed me to run the distance tool based off of this new field. I then had to reclassify the zoning distance into these two new value fields. (figure 12a, b)

Fig.12a Risk model based on residential neighborhoods

   

Fig. 12b Risk based on residential neighborhoods

    The fourth criteria that was used in this model was looking at the distance to schools. Since there is no schools feature class in the Trempealeau County database, I had to use the parcels feature class to look at them. I built a query statement to locate the parcels that were owned by a school district. After selecting those schools I used the polygon to raster tool to create my raster of the parcels owned by the school districts. I was then able to use the Euclidean distance tool to create a map showing which schools are within close proximity. (figure 13a,b)

Fig. 13a Model used to show the proximity of schools



Fig. 13b Risk based on Proximity to schools



Conclusion
    This assignment taught me a lot about the proper ways to use raster tools and how to create different models for rasters. Some of these tools were hard to run as the computer continued to stop working at times, but the overall end result allowed me to show areas of land that would benefit from a Frac sand mine in Trempealeau County. By looking at not only the geological make up of the land, but taking into account streams, rail lines, schools, slope, and much more, I was able to determine areas of land in the county that these sand mines would be best situated in.