In this module I learned how to carry out various data interpolation techniques ArcGIS Pro. I performed the Thiessen, Spline, and IDW methods to gain a better understanding of how these algorithmic methods affect data representation. I performed these interpolation techniques on a dataset provided by UWF for educational purposes only.
The data provided for this exercise included a feature layer of sample points containing multiple values. For this exercise I was to focus solely on BOD concentrations. These points were to have an assumed random distribution with no spatial correlation. The points spanned over the Tampa Bay area and I was provided a raster covering the area as well to aid in analysis.
To begin this assignment, I first made sure that the provided raster was set as the Processing Extent and Mask within the project's Environments. Creating the interpolated rasters from then on was fairly straightforward, so long as the correct tool was used and the parameters for each tool set properly. For the Thiessen technique, the Create Thiessen Polygons tool was first used, followed by the Feature to Raster tool. The IDW technique required the IDW spatial analyst tool, and the Spline spatial analyst tool for the Spline technique. For each method, the output cell size was set to 250 as this was the grid resolution for the provided raster covering Tampa Bay.
With each technique of interpolation properly carried out, I was then challenged to compare them and think critically about how well each method fits for the data provided. I determined that the Spline method fit best as this method passes directly through the sample points and the values it interpolates can exceed the maximum and minimum of the sample value's range. Alternatively, the IDW method would work better had the sample points not been randomly selected and had distance from the point been a logical variable affecting BOD concentration. Thiessen interpolation, while excellent for providing a rough idea of how well certain areas of the surface are spatially correlated, is simply that: a rough idea.
The Spline technique was also the most interesting to perform as it required some critical thought to do so accurately. After carrying out the regularized Spline interpolation, I was challenged to look at one area of the resulting raster and reason out why this area was represented with such drastically different values than neighboring areas. It was clear that there were two specific sample points that were very close together with two drastically different values that was influencing the result. The Spline tool sees this sudden change almost as a measure of slope and will be influenced to present the data as increasing similarly along its curvature. I enjoyed this mental exercise and it provides a great deal of insight into what to look for when preparing data and choosing interpolation techniques for a given scenario. Once the data was cleaned up, I performed the Spline interpolation once more, this time with the tension method. The result of which you can see below.
Comments
Post a Comment