Module 3.1: Scale and Spatial Data Aggregation

    In this week's lab, I explored how scale affects shape geometry of vector features and how resolution influences raster statistics and sampling. Additionally, I learned about spatial data aggregation and its influence on statistical analysis. Last, I utilized methods to identify gerrymandering of US congressional districts. 

    From this week's lab, I learned the relationship between scale and geometric calculations of various features. As the scale increased (1:100000 to 1:1200), area, perimeter, length , and count of features increased. As more vertices are used to draw features at higher scales, the features become more nuanced (e.g. more curves) and less generalized. This ultimately leads to greater length, and area measurements. Further, as scale increases smaller features may be included which explains the increase in feature counts. 

    Similar to scale for vector features, resolution has a large influence on raster features. I resampled a 1 meter resolution to a 2, 5, 10, 30, and 50 meter resolution raster to explore how a lower resolution DEM would influence the measure of slope. As the resolution decreased (1 to 50 meters) the average value of slope also decreased. As the DEM became more generalized the variation in elevation values decreased which influenced the measure of the average slope. 

    Last, I explored gerrymandering which is the manipulation of political districts to skew the results in the favor of a particular group based on the features within a specific area. Gerrymandering is an example of the modified area unit problem (MAUP). By altering the area that gets polled politicians from certain parties can increase the likelihood that they will win an election  and in some cases this bias is quite significant. I started by assessing features that were split into separate polygons despite there being no physical or geographical reason to separate the areas. Then I explored compactness as a means to measure gerrymandering. To do this, I calculated a Polsby-Popper score for each congressional district using the following equation:

Polsby-Popper score = (4π * area of feature/(perimeter length of feature)^2)

Then, I assess the lowest score which indicates the least compact districts. An example of a district with a low Polsby-Popper score is highlighted in the attached screenshot below.