GIS4930, Lab 6: Scale and Spatial Data Aggregation
This was one of the more interesting labs I've done in this program. Here, we explored how scale and resolution affect the amount and fidelity of spatial data. In the first exercise, we compared the number of waterbodies and the length of hydrographic lines (i.e., rivers and streams) visible at different scales, and graphed out how much is lost between each one. Then we made simple maps with a layer for each one, to visualize what was present or omitted.
In the second part, we took a Digital Elevation Model and used the resample tool to change the pixel size of the raster and created layers of different scales, then used the slope tool to create a slope grid of the resampled DEM, and then created histograms to determine the average slope for each layer. Starting with an original dataset but progressively lowering the resoltion produces a clear downward trend in average slope.
In the next, we explored the Modifiable Areal Unit Problem. To do this, we took socioeconomic data and explored the relationship between race and poverty using the Ordinary Least Square tool, at different levels of analysis: zip code, congressional district, and county. We then graphed the slope and Y intercept of our OLS output, showing how the same dataset can produce different trends or interpretations based on the levle of analysis.
Finally, we looked at all congressional districts in the lower 48 and the phenomenon of gerrymandering. We started by selecting out all districts that had multiple polygons (i.e., were broken up into multiple areas), and manually filtering ones where this was reasonable (i.e., part of the district included an island). From there, I used the Polsby-Popper formula to determine how compact each district was, which is used as a test for gerrymandering. I did this by entering the perimeter and area data from the remaining districts into Excel, and using a formula to calcuate it. The "winner", below, is a district in Maryland:
In the second part, we took a Digital Elevation Model and used the resample tool to change the pixel size of the raster and created layers of different scales, then used the slope tool to create a slope grid of the resampled DEM, and then created histograms to determine the average slope for each layer. Starting with an original dataset but progressively lowering the resoltion produces a clear downward trend in average slope.
In the next, we explored the Modifiable Areal Unit Problem. To do this, we took socioeconomic data and explored the relationship between race and poverty using the Ordinary Least Square tool, at different levels of analysis: zip code, congressional district, and county. We then graphed the slope and Y intercept of our OLS output, showing how the same dataset can produce different trends or interpretations based on the levle of analysis.
Finally, we looked at all congressional districts in the lower 48 and the phenomenon of gerrymandering. We started by selecting out all districts that had multiple polygons (i.e., were broken up into multiple areas), and manually filtering ones where this was reasonable (i.e., part of the district included an island). From there, I used the Polsby-Popper formula to determine how compact each district was, which is used as a test for gerrymandering. I did this by entering the perimeter and area data from the remaining districts into Excel, and using a formula to calcuate it. The "winner", below, is a district in Maryland:
Comments
Post a Comment