Field Activity #2: Visualizing and Refining Your Terrrain Survey

Introduction

Field activity 2 is a continuation of field activity 1 which our group constructed a surface terrain, coordinate system and recorded data points of the elevation.  Further information about field activity 1 can be found on my previous blog post, Field Activity 1: Creation of a Digital Elevation Surface. This activity has our group utilizing our collected data to produce a 3 dimensional model of our surface terrain using multiple interpolation methods within ArcMap and ArcScene.

To accomplish the task of producing our 3 dimensional model we needed to have or input our recorded data in an Excel file and import the file in to ArcMap and create a feature class of our recorded points.  We had to use 5 different tools within ArcMap to create a 3 dimensional model representation of our surface terrain.  The methods/tools with in ArcMap we had to use were IDW, Natural Neighbors, Kriging, Spline, TIN.  After these tools were used we were to export the results to ArcScene construct our 3 dimensional view.  From the results of the 3 dimensional views we had to decide which interpolation techniques best fit our survey of the surface terrain we created.  Our final step of the activity was to revisit our surface terrain site and record additional points to create a better representation of our surface terrain in the 3 dimensional models.

Methods

The first step of this activity was to import the x,y,z records into Excel.  Once in Excel I imported the table into ArcMap via ArcCatalog, in order to create a feature class of the grid with our measurements.  Prior to this step I had already created a Personal Geodatabase to house all of my data.  We had already entered our measurements in an Excel during the last field activity.  However, when we attempted to create a feature class from the imported table, we could not.  The formatting of our table was incorrect and needed altering.  Instead of having the grid style chart in Excel we needed individual columns for x,y, and z.  ArcMap uses individual rows to create each point, with the grid set up in figure 1, we had multiple points in one row.  Putting the results into individual columns (Figure 2) gave each point its own location. 

(Fig. 1) Incorrect formatting of table

(Fig. 2) Correct formatting of table

Once the formatting of the table was corrected, we imported the new table into ArcMap, created a new feature class, and placed the new feature class in the data frame.

(Fig. 3) The imported table as a feature class in ArcMap

The next step was to run one of the tools from the given list and see what kind of results we have.  I chose to start at the top of the list and work my way down.  I located the IDW tool in the toolbar and successfully ran the tool.  I then imported the IDW file into ArcScene and adjusted the parameters to view it 3 dimensionally.  However, when I did this I noticed all of our features were inversely represented.  Going back and taking look at the data we realized we had all negative numbers for our z-values (height).  We never adjusted the values to represent the elevation within the box.  Our group decided to simply add the height dimension of the board for our box to the z-value to give us the correct height.  If you look at Figure 2 you will see a T column which was the old z-value and the Z column is the recalculated z-value.  After the recalculations we imported the adjusted table in to ArcMap and created a new feature class from the corrected values.  After running the IDW tool again I opened the IDW file in ArcScene to verify the landscape was depicted correctly and at last it was!!

Now that the data is correctly formatted, I was able to use all of the varying interpolation methods to analyze the data from our sand terrain landscape.  As shown below in figure 4, once I ran the tool in ArcMap I had a 2 dimensional image of the terrain.  To create a 3 dimensional view, I had open the IDW raster image in ArcScene.  Once the image was open in ArcScene I had to adjust the layer properties.  Under the Base Heights tab I had to change No elevation values from a surface to Floating on a custom surface: and under the Rendering tab I had to check the box next to Shade areal featues relative to the scene's light position under Effects section.

(Fig. 4) After running the IDW tool in ArcMap this image was the end result.

The following segment briefly displays and describes the 5 interpolation methods used and displays the actual figure constructed.

IDW (Inverse Distance Weighted)

Inverse distance weighted interpolation uses linearly weighted combination of multiple sample points with in an area to determine cell values.  This method gives more weight to points closer to the cell center and points further away from the center are given less weight.

(Fig. 5) This is the surface model produced by the IDW interpolation method.  The IDW method does not give the best representation of our survey due to the number of golf ball appearing holes and bumps in the image.

Natural Neighbor

Natural Neighbor interpolation uses an algorithm to determine the closest set of sample points and applies weight to them based on proportionate areas to interpolate a value.  Based on this definition the Natural Neighbor method will not produce any unrepresented data.  The missing data is averaged with the data around it to give a natural flow.

(Fig. 6) This is the surface model produced by the Natural Neighbor interpolation method.  The Natural Neighbor method produced a far better representation than the IDW, with less bumps and smoother transitions.

Kriging

Kriging interpolation method uses an advanced geostatistical model to generate an approximate surface from a scattered set of points with height (z-values).  This method is a better average of the points in comparison to the above methods.

(Fig. 7) This is the surface model produced by the Kriging interpolation method.  The Kriging method has the best representation of the first three methods.  The valley, ridges, and hill look very good less the very small points on the top.

Spline

 The Spline interpolation method estimates values using mathematics which reduces overall surface curvature, resulting in a smooth surface.  This method is one step better at averaging the surface features compared to the Kriging method.

(Fig. 8) This is the surface model produced by the Spline interpolation method.  The Spline method is far better than the IDW and Natural Neighbor method.  This method is smoother than the Kriging method, so I feel this method is better than all of the above methods.

TIN (Triangular Irregular Networks)

TINs are created by triangulating a set of vertices with vector data.  The vertices are connected in a series to create a network of triangles.  This method though accomplishing the same task is done differently so the image appears to look odd but the representation is pretty accurate for the method.

(Fig. 9) This is the surface model created by the TIN interpolation method.  The TIN method produced a better representation than the IDW method but it has a far number of ridges due to the use of triangulation to construct the model.

The final step in the process was to fill in the Metadata in ArcCatalog.

(Fig. 10) Metadata for the feature class.

Discussion

Determining the best of the 5 interpolation method was fairly easy as I felt there was one true winner.  The Spline method (Fig. 8) best depicted the actual design we had created in the previous exercise.  The smooth transitions from low elevation to high elevation and the lack of odd high point or bumps on the terrain made it the clear cut winner of the 5 methods.

The Kriging method ranked 2nd on my list for representation of the original terrain.

Natural Neighbors ranked 3rd for representation of the landscape in my eyes.

The IDW interpolation gave a general idea but was the furthest from the actual thing of the 5 different methods.

The TIN method (Fig. 9) has the principle of the other 4 methods but is constructed differently.  I feel it did a fair job of displaying how the landscape appeared.  However, due to the spacing we used for our coordinate system, the gaps were to large and created a ridges due to the triangulation method.

After  running all of the interpolation methods on the original data part of our assignment was to assess where our data was lacking in representation of the actual terrain model.  After determining the locations which needed better representation we were instructed to revisit our site and record additional points to improve the terrain representation.

After analyzing our data models we determined that the area around the ridges, valley, and the hill could use some additional data points to assist showing their true shape.  The accuracy of this part of the assignment was going to be less than the first activity.  Since the second measurement took place a week after the original creating, the terrain has endured a couple rainstorms and bicycles had tracked through it.  My group reconstructed the terrain the best we could and attempted to place the box in the same location we had the week before.  We felt the y-axis where the ridges and valleys were in need of the most attention and by adding points we would improve the appearance of the hill in the same process.

(Fig. 11) The feature class in ArcMap after adding the additional points from the second measurement.

After adding the additional points to the original table and creating a new feature class in ArcMap I ran all of the interpolation tools again to see if we had improved the results.  The results were improved, however an issue with the second measurement was visible in the interpolations.  It was apparent the alignment of the box was shifted on the y-axis a little bit. This shift gave the appearance of ridges being wider than they actually were and did not smooth out the surface as we had originally planed.  The misalignment is apparent in figure 11 in the left side ridge and the left side of the hill.  In the perfect world we would have been able to leave the box in place and had a cover for it to protect the terrain from the elements.  However, this was not the case, but my group handled the challenges to the best of our ability.

(Fig. 12) The Kriging Interpolation method had the best representation with the second measurements added.  You can see the miss alignment in the left side ridge and the left side of the hill.

Conclusion

After running all of the interpolation methods the Spline method gave us the best representation of our surface terrain.  When the second measurements were added in the misalignment of the box skewed the results of the Spline method more than the other methods.  The Kriging interpolation method gave the best representation with the additional points and was less affected by the shift of the box than the other models.

Using the top of the box for z-axis measurement in activity one gave us a solid measurement to work from.  However, it required us to make the alteration of adding the dimension of the board to the value to achieve a positive number which was required to portray the terrain properly.  Having recorded and entered the data incorrectly for use in ArcMap, we learned a valuable lesson which will benefit us in every activity in the future.

The second measurement gave us a number of variables out of our control.  Our group worked very well together to overcome the obstacles and achieved the desired result in the end.  Knowing what I know now I would have added some form of locator markers for the corners of the box to make sure we got the box placed in the exact same location.

Overall, I was very pleased with my group and our ability to, work together, coordinate times to meet, and complete the tasks at hand.  When you don't get to pick your group partners you never know how the chemistry will be or if everyone will work well together.  My group was great to work with and I would be happy to work with either of them in the future.