Module 3.1: Scale Effect and Spatial Data Aggregation

     This final module explores the effects of scale and resolution differences on vector and raster data. We were given a data set of hydrographic vectors rendered at three different scales: 1:1200, 1:24000 and 1:100000. After exploring the dataset I found that larger scales such as 1:1200 can capture smaller details and finer geometric shapes better than smaller scales such as 1:24000 and 1:100000 since detail is lost as intricate features are simplified. These smaller scales can also cause small features to be completely excluded as the scale makes them comparably insignificant. 

    We also examined how the Modified Area Unit Problem (MAUP) can effect spatial data by using a nonwhite % vs. below poverty line % dataset as a case study. The data shows that the Scale Effect of the MAUP can cause results to vary if the units of analyses are differently sized spatial units (Zip Codes v.s. Counties, for example). The smaller the spatial scale, the more divergent the results can be.

     Lastly, we examined a map of the U.S. Congressional districts for gerrymandering. Gerrymandering is a political process in which a group in charge of redistricting draws districts in a way that would devalue an opposing group's vote. This can be done by clumsily clumping a target population's entire area into a single, messy district, resulting in them only winning one district. Another method of gerrymandering is done by drawing the districts in a way where the target population's area is heavily split up so their votes get drowned out by the other voters, resulting in them winning no districts.

 

     Pictured above is an example of one of the worst cases of gerrymandering in terms of compactness. Compactness is defined as the area a feature occupies in relation to its area center, an can be scored using the Polsby-Popper Score. This score is calculated using the formula 4π(Area)/(Perimeter)^2, resulting in scores between 0 and 1, where lower scores are considered bad. The district in the screenshot above achieved a very low score of 0.08.

Module 2.2 - Interpolation

        In this week's module we examined the differences between various surface interpolation methods. We went through the steps to create rasters using the Thiessen, IDW and Spline methods, then we compared the results.

     This screenshot is of our study area, Tampa Bay, where we were given sample points representing the BOD of the bay. The first raster I made was the Thiessen raster. This involved using the Create Thiessen Polygons tool and converting the resulting shapefile into a raster. Thiessen interpolation creates proximal zones around the sample points and it can act as a starting point for greater data collection. 

    The second method involved using the IDW tool to create a DEM of the BOD points. IDW interpolation can be more sensitive to outliers than other interpolation methods. We also used the Spline method in this module. 

   We had to modify the original dataset in order to allow the Spline tool to function correctly. The northern portion of the bay had overlapping sample points, causing an area of unusually high BOD concentration to be generated by the Spline tool. To rectify this, I averaged the results of the overlapping points, overwrote one of the results with the average and deleted the other point. This way, the data would not be simply disregarded.

     The Spline method can be split into two different methods: the regularized method and the tension method. The screenshot included in this blog post was created using the Tension method, which is a method that creates an output that stays closer to the known sample points when compared to the regularized method.

Module 2.1 - Surfaces - TINs and DEMs

    This week's module involved comparing TIN and DEM elevation modules. TINs (Triangulated Irregular Networks) consist of a network of triangles. They can be good at representing areas where the surface is highly variable. A DEM (Digital Elevation Models) is a graduated raster grid where each cell has an elevation value. They can be easier to process than TINs due to their comparative simplicity. 

    Contour lines can be generated from both of these elevation models. The TIN contour lines were easy to generate as they were found as a check-box in the Symbology pane. The DEM contour lines had to be generated using the Contour tool.

    TIN contour lines are shown in red and black, and the blue contour lines were generated from a DEM. These contour line types can have remarkable differences, depending on the study area and type of analysis performed. In this case, the study area had spots of great elevation changes, and the differences between the two elevation types can be seen there. This module highlights the importance of choosing the right elevation model for the data and analyses required.