Sandbox Elevation Model Continued

 Introduction

In this project, the preliminary data collection was analyzed using 3D tools, working with the different types of interpolations showing where improvements could be made. With that, the evaluation of how a re-survey refining our data would better work was made. The second collection was made and the data manipulation with ArcMap, ArcToolbox and ArcScene resulted in a consistent representation of the landscape designed inside the planter box.

Methods

After the data collection, it’s necessary to format the table using Microsoft Excel so each point has x, y and z columns. With this format ready, the creation of a feature class using this coordinates is possible in ArcCatalog or ArcMap.

With the point feature class in ArcMap, it’s time to create a raster related to the Z value of the points. Five interpolation methods were used with ArcToolbox to create rasters: IDW, Kriging, TIN, Natural Neighbor and Spline. These raster files are necessary to provide a 3D view using ArcScene.

The 3D view allows the analysis of the accuracy of the results in comparison with the real model in the planter box. With that, it’s possible to decide what might be changed in terms of density of points and which interpolation method best represent the features designed in the box.

With this in mind, a second collection is necessary to refine the data and obtain better results. For this survey, strings were used to create the grid where the points would be taken. Nails were placed with a hammer (Figure 1) to fix the strings, avoiding the latest problem where the negative temperature compromised the effectiveness of the tape that was being used. Also related to the temperatures, the day chosen to collect the data was warmer than last time and the strings were cut inside the building to minimize the amount of time outside.

Then, the same computer process was applied: formatting table, creating a feature class, transforming it into a raster and visualizing it in three dimensions with ArcScene. The final result will be a digital model that reasonably represents the landscape designed in the planter box with accuracy.

Figure 1 - Use of nails and hammer to fix the strings.

Results

As it was seen in the last report, the note-taking was made following the box shape, instead of having three different columns x, y and z for each point. Hence, it was necessary to create a table (Figure 2) in this format to allow the creation of a feature class with the data.

Figure 2 – Fixed format of coordinates.

Then, the feature class was created and used to create five rasters using the different methods of interpolation. Using ArcScene, the visualization in 3D (Figure 3) made it visible which methods represented better the surface and which features in the landscape needed more detail.

Figure 3 – Results in three dimensions.

Unfortunately, none of the methods were able to represent one of the features designed in the box: the river flowing in the right-upper corner of the Figure 4 was supposed to be represented with minor elevations in the left-lower corner of the 3D images in the Figure 3. Some points in this area does have the pattern of low elevations, however, they don’t have continuity as the river should have.  

Figure 4 - Landscape in the planter box with the river.

Therefore, the density of points being collected was the first change for the second survey. At the first survey, the point collection was being done starting from the origin with 6 inches intervals to east and north. Thus, the new collection was done starting from the origin with 8cm intervals. (Figure 5). The International System of Units was now applied since it fits better the scientific purposes of the project. (Figure 6)

Figure 5 – Comparison between data collection methodologies.

Figure 6 - Use of SI to measure the intervals.

Then, the collection was made collecting points of the strings’ intersections. (Figure 7)

Figure 7 - Elevation data collection.

With the data collected in the notebook, the table was formatted in Excel (Figure 8) and used to create the feature class. Using ArcToolbox, the different interpolation methods created five raster files that were visualized in 3D in ArcScene. (Figure 9)

Figure 8 - Notebook data formatted to an excel table.

Figure 9 - Visualization in 3D of different interpolation methods.

Conclusion

The analysis of the 3D model results indicate that the spline method was the one that best represented the surface designed in the box. The reason for that is that there’s a higher generalization in the unknown areas that smoothly represents the landscape. In the other methods, it’s easy to notice some geometric shapes that are repeated over the surface, which doesn’t represent the real-world. Besides the spline method, the natural neighbor method also represented reasonably the landscape.

Also, the increase of points collected clearly shows the improvement in the representation over the surface. The river is now well lineate and the heterogeneity of the surface is more apparent.

The project as a whole showed how a preliminary collection can improve fairly the final collection and result.