Surfaces: Interpolation

In this lab, we learned about the different methods for interpolating surfaces using several data sets. In the first part, we compared the interpolation between Spline and IDW for elevation data from Kansas. In general, the difference between the two techniques was not that great. Over half of the pixels fell within 3 m of each other in the models.

For the second part, we used water quality data from the Tampa Harbor to compare Thiessen, IDW, and two different methods for Spline interpolations. 

Thiessen uses the exact values from the points, creating large blocks of the study area through a nearest neighbor technique and assigning them these values. This create a not very nuanced surface model with abrupt changes between the areas.

IDW, or inverse distance weighted, compares each point to the five closest points around it, working under the assumption that things closer together are more similar. This resulted in a surface with circles around many of the individual points, with areas disconnected by slopes rather than flowing into each other.

Spline runs through each point to create a smooth surface while minimizing the total curvature. The regularized method generally creates more elastic surface that will frequently have values that lie outside of the sample data range. The tension method will create a surface that is flatter and force the estimated values to stay closer to the known sample points. Because spline tries to make nice smooth surfaces, we had to fix the data where several points were clustered together with varying BOD measurements.

In the end, I think I was happiest with the Spline interpolation using the tension method. It was a great exercise to see the results from each.

Here's an image of the Spline interpolation using the tension method.