Lab 8 - Surface Interpolation

Interpolation...a word that is found throughout GIS analysis readings and studies, but I didn't really understand it until this lab. There is so much still to learn, but I am excited to understand some different interpolation techniques a little more and be able to recognize different surface patterns due to interpolation a little better. While I am often frustrated as my limited math skills, this lab actually left me feeling a bit sad I don't have more advanced math skills right now as I found the exploration of surfaces and results interesting!

For the lab, I first created two DEMs from elevation point data-one using IDW and one using Spline Interpolation. Using the Raster Calculator to find the difference I explored where and why different values occurred from one to the other. Unsurprisingly, the greatest differences were in areas lacking sample points, allowing the different interpolation methods to be more obvious. Another pattern of note, in general, the Spline interpolator created a broader range, or more areas, of lows and highs in the DEM then the IDW interpolator.

The second part of the lab was interpolating BOD concentration sample point data in Tamp Bay using different interpolation methods and collecting statistics for comparison.  Non-spatial statistics were noted, then Thiessen polygons, IDW, and Spline (regularized, then tension) surfaces were created and statistics noted. The regularized Spline surface ended up with a potential false "peak" of high concentration in an area where there weren't any sampling points. By looking at the surrounding data points, I located where I thought the error was (two close points that contained fairly widely different values), removed the lower value that I felt was erroneous (in real life further sampling may need to be done in that area), and reran the Spline interpolations. After creating all surfaces, I carried out a comparison to determine which model I felt captured the data distribution best. While IDW had the lowest standard deviation and matching max and min values to the sampled data, I felt the Tensioned Spline with it's smoother, more gradual range of values was more likely how BOD concentrations would be distributed in the bay. The standard deviation was still low, max value the same as the sampled, and the lower min range was appropriately possible for the study.

Map showing the difference value range for IDW and Spline interpolated DEMs of elevation data.