Lab 2: Visualizing and refining terrain survey

Introduction

This second lab was a revision and refinement of our first lab in regards to methodology and surveying techniques used to map the terrain of our first sandbox. The goal of the second lab was to assess and alter our surveying/measurement techniques depending on how accurate our initial excel data was and to either revisit our first sandbox or construct a new landscape if our old one was destroyed during the week (which ours was so we created a new study area). Following our in class discussion and demonstration of each groups survey techniques, our group found that our current techniques would suffice for the second rendition of the lab so long as we took the advice of our professor and employed more extensive measurement of key features. We decided that additional points would be measured around sharp changes in elevation to ensure that their topography was as accurate as possible which would result in a more defined map and 3D elevation model in ArcScene. Our first sandbox's excel xy data in ArcMap looked relatively accurate when we ran a spline spatial analyst tool on it because we used a 14 by 15 grid of points (210 total) which would be further augmented by higher point density on key features.


Methods

Our first study area had been under the UWEC foot bridge but the sand was not especially malleable or damp enough to hold its form so we chose for our second sandbox to be set up in the Putnam Hall sand Volleyball court. Luckily for us there had been slight precipitation that morning and it was an overcast day so we had the whole of the court to ourselves and did not have to dodge incoming volleyballs. We set up the 4 feet by 4 feet wooden perimeter as we had before, making sure that the study area was level beforehand and set about creating our hill, ridge, valley, depression, and plain features.

Figure 1: Construction of survey area

We measured and marked our 14 by 15 grid every 4 and 8 centimeters (initial measurements moved inward by 1 cm to allow for whole number grid squares) and this time only laid string across the x axis because we would be using the segmented ruler as a mobile y axis marker for where each measurement would be taken.

Figure 2: Measuring x y grid

While taking measurements on flat or gradual sloped terrain we would record elevation data every 8 centimeters but once we approached our key features we switched to recording every 4 centimeters. This time instead of putting our data into a xy grid akin to battleship we used three separate x, y, and z fields which could more easily be translated into an ArcMap document (we had quite a bit of trouble reformatting our last excel table to ensure compatibility with the software). Once all 238 points had been measured and recorded in the correct table format they were transcribed into an excel spreadsheet which we then could analyze using interpolation techniques.

Once the excel data table was imported into ArcMap we created a feature class from the xy fields which we could then run through several interpolation methods, and after the program had run the interpolation we exported the layer as an image file to view in ArcScene.

The first interpolation method was IDW which interpolates a raster surface from our points by using an inverse distance weighted technique. An advantage of IDW is that it works well with dense point collections, but the disadvantages are that it requires close proximity of points and large gaps between points will cause a generalization of the surface, and also the terrain model has little smoothing between points which creates an unrealistic looking surface. Compared to the real life elevations of our study area this interpolation method does not fit the landscape very well due to its frequent bumps in elevation.

Figure 3: Overhead symbolized view of IDW interpolation in ArcMap

Figure 4: Oblique view of IDW interpolation in ArcScene

The second method was Natural neighbors interpolation, which uses the same basic equation as IDW interpolation but smooths the elevations between data points to give the landforms a more realistic look. The advantage of using Natrual Neighbors is that it is a simple interpolation method capable of large point datasets, and its disadvantage is that it still distorts the elevation model enough to make it look unrealistic. This method fits out real life elevations quite a bit better but elevation between data points is still skewed and looks unnaturally pointed.

Figure 5: Overhead view of Natural Neighbors interpolation in ArcMap

Figure 6: Oblique view of Natural Neighbors interpolation in ArcScene

The third interpolation technique used was Kriging, which takes into account spatial correlation that can help explain variation in the surface. The advantages of Kriging are that it is very useful in the fields of soil science and geology, and its disadvantage is that for it to be accurate you have to know the spatially correlated distance or directional bias. Kriging is not a method relevant to our lab due to how it represents elevation features, despite its usefulness in other fields of work.

FIgure 6: Overhead view of Kriging interpolation in ArcMap

Figure 7: Oblique view of Kriging interpolation in ArcScene

The fourth interpolation method used was Spline which estimates the elevation values using a mathematical function that tries to minimize the overall surface's curvature, which results in a smooth surface passing through the point data's elevation. Its advantages are that it can accurately predict ridges and valleys in data sets, and its disadvantage is that its usefulness is limited to specific fields of work. Spline gave us the most accurate 3D model representation of our point data and mirrored the real life study area's landscape the best.

Figure 8: Overhead view of Spline interpolation in ArcMap

Figure 9: Oblique view of Spline interpolation in ArcScene

The last interpolation method was TIN (Triangular Irregular Networks) which are vector based digital geographic data the are comprised of triangulated vertices connected to other vertices to create triangles, all of which display that vertices elevation in relation to other vertices. The advantages of TIN are that it requires fewer data points to be stored than a DEM, making data input more flexible. The disadvantages of TIN are that it is less suitable for certain GIS applications like surface slope and aspect. This interpolation method created a fairly accurate representation despite its angular vertices because it shows distinct elevation changes and shadows which helps visualize the terrain.

Figure 10: Overhead view of TIN interpolation in ArcMap

Figure 11: Oblique view of TIN interpolation in ArcScene

Of all the interpolation methods we used Spline gave the most accurate representation of the real life sandbox terrain.

Discussion

The terrain elevation data we gathered in our second lab was a much better representation of the actual physical surface than our first lab's attempt because we now knew to hone in on specific features, and how to catalog our data points properly and efficiently. Our study area was much more suitable to our needs this time which allowed us to create an study environment with both higher and lower elevation features. Although we were not able to resample our old study area due to it being destroyed by pedestrian traffic and rain, we used our experience from the first attempt to reduce the number of errors and mistakes made in the new study area. With the addition of the maps produced from running the five interpolation methods we now know of, the analysis of our study area's physical features is much more complex and can be approached from several angles depending on which model we use.

Conclusion

Over the two weeks it took to conduct these labs my knowledge of rudimentary surveying skills has grown and been refined through trial and error, consulting my fellow peers and professor, and tinkering with ESRI programs like ArcGIS and ArcScene. I now have several skill sets that will easily transfer over to other classes I am currently taking such as Web GIS and Computer Mapping, and any future geography classes I will be attending. Having to conduct elevation surveys twice using rudimentary and time consuming methods has given me renewed appreciation for the technology that automates those processes, and has instilled a certain curiosity on how they will work which will hopefully be answered in future labs.