Module 5 - Special Topics in Geographic Science
This week we worked on surface interpolations. The problem we were given was to derive Biological Oxygen Demand (BOD) levels within Tampa Bay (Florida). BOD levels are commonly used to measure water pollution. Surface interpolations uses spatially related data to create surfaces in the form of rasters. Interpolations will analyze relationships between points based on the distances from each other to come up with a final product (raster). The resulting product could be used to carry out predictive analysis.
We used three different methods of interpolations to predict the BOD levels within Tampa Bay. We used Thiessen Interpolation, Inverse Distance Weighting (IDW) Interpolation and two forms of Spline Interpolations (Regularized and Tension). Thiessen interpolations also known as Nearest Neighbor Interpolation is the simplest form of interpolation. Thiessen interpolation uses an simple equality function and only considers one nearest point to assign values to areas with no data (Bolstad 2017). IDW interpolations is more complex than Thiessen interpolation in that IDW "estimates values of unknown points by using sampled values and distances to nearby points" (Bolstad 2017). Values at unknown points decrease as the distance to the reference point increases hence the inverse relation in IDW. The final technique we used was Spline Interpolation. Splines use polynomial functions and go through each point to create smooth curves. Splines could be regular meaning there is no weighted values or they could be tensioned meaning results may be coarser as lines must pass directly through the control points (ESRI).
During this exercise I found the that the IDW interpolation method worked best at predicting the BOD levels within Tampa Bay. I believe that IDW worked the best because it had the lowest standard deviation in the data that was derived from the interpolation and the numerical values produced were similar to other interpolations and they seemed to be in the middle as well. The numerical values produced for BOD landing in the middle when compared to other interpolations led me to believe that the numbers were at least comparable and not outliers. The small standard deviation between data produced told me that there was less of a spread within the data and therefore it is likely the data is more accurate. Below is a screenshot of my IDW interpolation raster. Higher concentrations of BOD are represented by lighter colors. The layer in pink shows land boundaries in the area.
No comments:
Post a Comment