Visualizing and Refining Terrain Survey

Field Activity #2

    This field activity is a follow up from the previous weeks field activity of surveying the terrain surface. In the previous field activity we were asked to create a terrain surface in one of the planters boxes in the courtyard. Since it is winter time out, we used snow instead of sand when construction our surface terrain. After construction our surface, it was now time to figure out the best way to survey the area. We used a systematic sampling method, which allowed us to overlay the terrain with a grid so we could take measurements for all of the different types of terrain. Figure one shows the grid system that we used. It was constructed with yarn and thumbtacks.

 Figure 1

While we were getting measurements for our entire terrain model, we were also recording them onto a piece of paper. We got the x the y and the z coordinates for 180 different points in our flower box terrain model. We then were able to transfer all these points into an excel file so we could eventually bring those points to life in ArcMap and ArcScene.

    Now the real fun began of being able to display these points in ArcMap allowing us to render a 3D display of all our points. When we first brought our points into ArcMap and Scene, all that came across were the points in a grid format. We used five different interpolation techniques that would allow for us to show these points in a 3D format. The first interpolation technique that we used was the Inverse Distance Weight technique. IDW determines cell values using a linearly weighted combination of a set of sample points. One reason why I didn't like IDW was because it doesn't take into effect the areas in between two points. For instance, look at the top of the mountain in figure 2. Our terrain model was more of a plateau on top of the mountain, not two different peaks.

Figure 2 shows the Inverse Distance Weight interpolation technique.

The next interpolation technique we used when dealing with the data was the Natural Neighbors method. This method finds the closest subset of input samples and puts them into a query point and applies weights to them based on proportionate areas. This technique works best when there are data clusters in the data set points. Figure 3 shows the natural neighbors interpolation method.

Figure 3 shows the Natural Neighbors interpolation technique. You can see how the surface of the terrain is much smoother in areas compared to the IDW method. It's also worth noting that the top of the hill is also more of a plateau like it actually was in our terrain.

The third interpolation technique we used was the Kriging method. This method is Gaussian process regression that models the points based off of their prior covariance's. The assumption is that there is some type of correlation between the points. Without there being any correlation the points would be lost and would be put randomly onto the map. Figure 4 shows the data in the Kriging interpolation method. This is one of my favorite techniques however when I turned the 3D model onto the side the hill and the ridge ended up being rather pointy, and this is not how it was modeled in the actual terrain.

Figure 4 shows the Kriging interpolation technique.

The fourth interpolation technique we used was the spline method. This method estimates values using a mathematics function that minimizes overall surface curvature, resulting in a smooth surface that passes through exactly the input points. To me this was the most visually please technique that we used.  Figure 5 shows this technique.

Figure 5 is the Spline interpolation technique. You can see that there is no pointy areas rather everything is smooth and blends into one another. This was one of the main reasons why I believe that this one was the most visually pleasing technique we used.

The fifth and final interpolation method was used was the TIN method. A TIN is a Triangular Irregular Network that uses non overlapping triangles allowing to show the surface features. This tends to be very rough and not pleasing to the eye, but it does display surface data. Figure 6 shows the TIN created from the survey grid.

Figure 6 shows the TIN interpolation technique used.

After creating and using these different interpolation techniques, we were asked to look at our data again and to see if there were any areas that we felt would benefit from having more points in that area. Of course there is always room for improvement when looking at data like this so we decided to concentrate on the hill, portions of the valley, and the ridge in the back.

We devised a way from looking at our data that we already have and the area in the planters box outside that we were able to get 16 more points of x,y, and z data for our terrain. After getting this data we put it into another excel sheet and imported it into Arcmap which would allow us to overlay the data onto the previously displayed data points. Figure 7 shows the new point data we collected from our planters boxes.

 Figure 7, With this data being in between other points, it will allow us to give a better representation of the areas that have steep slopes.