Module 2.2: Surfaces - Interpolation

    In this week’s lab I learned about different surface interpolation methods. I have used some of the methods (IDW and Spline) in previous lab assignments. During the assignment, I learned about some of the pros and cons of each method. 

    I started by exploring thiessen interpolation which is a method that I have not used before. The thiessen output was very coarse and low resolution but simple. This method was not suitable for the Tampa Bay water quality data that we were provided in this lab because of its output’s coarse nature distribution of the sampling stations. This method would be beneficial if you were looking for very high level trends in the data and the output would likely be more informative with data that are of greater quantity and are more equally distributed.

    The next method used was inverse distance weighted (IDW) interpolation. While this method provided a more nuanced output compared to the Thiessen interpolation, there were still limitations. This method assumes that points closer to one another are more alike than points further away. While this is generally true the spatial distribution of the sample stations greatly limited the quality of the IDW output. IDW would serve as a better tool for data that closer together with equal spatial distribution. 

    The last method used was spline. I started with the regularized spline interpolation method. While this method did a better job compared to IDW and Thiessen, it was predicting some very extreme values at each end of the data range. To resolve this problem, I assessed the data’s spatial distribution and averaged stations that were very close together. By doing this, I decreased the variance in the spatial distribution which led to decreased variance in the regularized spline interpolation output. I also used the tension spline interpolation method which provides greater constraints on the interpolation by limiting the interpolated values to basically only within the data range of observed/measured values. 

    Ultimately, I found the tension spline interpolation method to be the most appropriate method for interpolating the water quality data collected in Tampa Bay. If I had to make one slight tweak, I would slightly decrease the limitations of the tension method, so it would predict a little further outside of the data range. Attached below is a screenshot of the tension spline interpolation output.