Assignment #2: Development of survey data into a Digital Terrain Model & comparison of model to actual surface and resampling of surface

Introduction

This project was a continuation of the first project.  The idea was to take our data that we had collected and generate a map of the terrain.  There were several options available for how to generate the map, as the grid points we had collected did not provide a complete picture of the elevation; the spaces between the grid points needed to be filled in, and there are several methodologies that can be applied to do this.

Methods

To start, we needed to get our elevation point data into ArcGIS.  The data file that we had created in Microsoft Excel first had to be saved with the extension .dbf (Database File).  This allows it to be imported into ArcMap.  Once in ArcMap, the x, y, and z data had to be interpolated.  Interpolation gives data to the areas between the grid points, known as cells.  Interpolation will make our map have a continuious surface.  ArcMap has several options for how the interpolation is to be done.  The five methods are IDW, Natural Neighbor, Kriging, Spline, and TIN.  We chose to use each of the five methods to generate five separate maps, and then determine which gave the best looking results. 

The IDW Method

Inverse Distance Weighted interpolation, or IDW, draws on grid points that are close to the cell more than on grid points that are far away, as it makes the assumption that the nearer the points are to the cell, the more accurately they can be used to predict the cell’s value.  It does, however, draw on grid points that are farther away than the immediate area when making its calculation.

The map generated by the IDW method.  This map is very choppy, and gives the appearance of many little pointed hills all over the landscape surface, which were not present on the actual landscape surface.

The Natural Neighbor Method

Natural Neighbor interpolation defines a local area around the cell and uses only those grid point values to calculate its cell value. 

The map generated by the Natural Neighbor method.

The Kriging Method

Kriging interpolation, unlike the other methods, examines the z data for the surrounding grid points when making its cell value calculation. 

The map generated by the Kriging method.

The Spline Method

Spline interpolation uses a mathematical function to minimize surface curvature when creating its cell value.  This results in an overall smoother landscape. 

The map generated by the Spline method.  The smoothness is clearly evident.

The TIN Method

TIN interpolation creates triangles out of three grid points and determines cell values based on these triangles. 

The map generated by the TIN method.  This map has a striking visual appearance that separates it from the other methods, owing to its use of triangles.  This map clearly shows the classes of elevation that it generated.  This makes it easy to see how the features relate to one another.  However, it makes the top of the ridge appear to have three little unconnected peaks in the yellow class, when the actual feature had a more continuous area of that elevation.

Discussion and Results

Ultimately, the Spline method produced the best map for our purposes.  IDW and TIN produced maps that did not replicate the actual conditions of the smooth, snow landscape.  The Kriging and Natural Neighbor methods produced maps that were almost identical, but were not as good as the Spline method.  The Spline method produced a map that most represented the actual landscape terrain that we had created.

This project was immensely valuable.  With x,y,z data from any landscape source (including actual life-sized field work) one can employ this methodology to generate a proper map showing the elevation.