Lab 2: Visualizing and Refining a Spatial Survey

Field Activity #2: Visualizing ad Refining Terrain Survey

By: Joseph Mandelko

Introduction:

This activity was a continuation of the first lab. In the first lab a model of a terrain surface was constructed in a box of snow in the courtyard oh Phillips Science Hall at the University of Wisconsin Eau Claire. Data points were collected in a systemic sample collection method using a 5cm by 5cm grid pattern laid over the terrain. Sea level was based on the height of the box the snow was in and measurements were taken to the positive or negative of the baseline of “sea level”. The additional instruction for this week’s activity was to take the data points collected last week and bring them into ArcMap and ArcScene to visualize the collected points in 3D. The data would be viewed using an Interpolation method, methods used included IDW, Kriging, Natural Neighbors, Spline, and Tin. Though TIN is not really an interpolation method it is a way to view the coordinated using x,y, and z values through the construction of triangular shapes that give the appearance of depth. After viewing the data from the first collection it was clear which areas needed more accurate measurement and the scene was recreated and a grid of 2.5cm by 2.5cm was used to get more accurate readings.

Methods:

In order to view the data in a 3D manner it was necessary to bring the data into ArcMap by adding X Y data. A “Z” value can also be turned on using the third coordinate as an elevation. Once the data is added it appears as it does in figure 1.

Figure 1: x,y data input in ArcMap

From this point an interpolation method can be used to draw lines connecting points with similar Z values creating a 2D image with topolographic lines on it. That file is then exported to ArcScene where the image can be draped over the Z values to add depth and a 3D image is created. All of the interpolation methods can be compared in the visual after the description of the methods applied.

The first interpolation method used was IDW or the Inverse Distance Weighted interpolation. This method uses a weighted combination of sample points to determine cell value. Weight of the sample is determined by how close the Z value is giving it weight and applying depth.

The second method applied was the Kriging method. In Kriging interpolation the software uses the distance or direction between sample points to reflect a spatial correlation that can be used to explain the variation in the surface of the image. The tool uses mathematic functions to match to specific numbers of points, or all points within a defined radius, and determines output value for each point. If there is a known correlated distance or direction bias in the data this is the ideal method.

The third method is Natural Neighbors. Natural neighbors is similar to IDW in that it uses weights applied to points however it finds the closest subset of the input sample points in the direction of similar values . This method uses a query point and looks at sample point values around that. This method is quite accurate and doesn’t tend to create features that are not part of the sampled area.

The next method used was TIN, or Triangular Irregular Networks. Though not an interpolation method it is commonly used to digitally represent surface morphology. TINs are vector based, using lines and points to construct polygons. The surface itself, unlike interpolation, does not have depth but the shape of the triangles creates that illusion. The TIN can then be applied at Z points and depth is given to the image.

The last method used was Spline. Spline estimates values using a mathematical function to minuimize overall surface curvature. This effect is clearly visible on the application of Spline below. The function creates a smooth line that runs through exact sample points to make a surface that is not fractured and broken up at harsh angles.

 

Results:

All of these methods did a nice job at representing the snow landscape that had been created and likely any of them would suffice however when thinking of this activity on a larger scale it may be true that Spline did the best representation. They were compared after exporting the file as a JPEG and orienting it as the data had been taken. Scale is shown above with a one meter line and a north arrow to give orientation. Keeping the files in the same format helped with the comparing of methods. The mathematical functions that create the lines through the Z points makes for a much smoother consistent image than the other files. Not that the other interpolation methods did a poor job but Spline seems to represent this model the most accurately.

Revisit of the Survey:

After places for possible improvement were identified, mostly around the hill and the valley, and data points were taken again this time in a 2.5cm grid. As explained earlier the data was put into ArcScene and run with Spline so the best interpolation method could be compared with the same method but with more points. The second measurement with application of the Spline method did do a more accurate job at capturing the features though not as much as I would have expected. The two look very similar and it would not appear that the second application was needed to represent the model. This would however be different on a larger scale of sampling.

Conclusion:

The concept of creating a grid based survey is an interesting one and seems to have worked in the small sample area however on a larger scale I think it would be much more difficult to put into practice. It may not always be realistic to do such a detailed grid and more may have to be left up for interpolation than was done in this activity. Essentially it is a good concept, using points at even spaces to represent a whole areas topology. In other surveys the data can be collected in the same way but it would not be done with string and a meter stick but with laser technology to get more accurate readings and exact points. This was a much more crude method than would be used in the field but essentially the same idea. Interpolation can be used in similar surveys to represent not just elevation but land cover areas, temperature, and weather patterns. In the end this was an effective activity for visualizing different tools. However no test is perfect and some of the data collected the second time could have been more accurately read if the snow was more packable as it was on the day of the first testing. The powdery snow resulted in more steep angles because the snow could not stick to itself. The difference in the first and second spline readings would have been more visible had the original grid been 10cm by 10cm instead of 5x5. There may have been a more telling difference between accuracies.