Activity Two: Visualizing and Refining Terrain Survey

Introduction

Often in trying to analyze a digital elevation surface multiple methods need to be considered and utilized to come up with an accurate visualization of what exists. As a continuation of last week's grid based survey of our sandbox we revised our methods to include more selective points to better represent the valley area. After analyzing the digital elevation models from our first set of data, we realized there were not enough points to properly show the narrow valley included in our sandbox.

After the revision of the data surveying collection, different interpolation models were used to analyze the accuracy of the features within the sanbox. ArcMap and ArcScene were used to achieve this task.  

Study Area

• Date: 22 September 2015 from 3-5pm 
• Sandy bank of the Chippewea River Valley 
• Conditions: Sunny, Average temp. 65°F (retrieved from Weather Underground)

Methods

Figure 1. Original 8 x 8cm coordinate
       system, 18th Sept. 2015.

Revising the Coordinate System

Figure 1 shows the original coordinate system created 18 Sept. 2015 with the cell size of 8 x 8cm. 

The valley was too narrow to be represented by an 8 by 8cm coordinate system. 

Only x-values from Y12 to Y15 were re-recorded, because we thought the digital elevation models produced an accurate representation of our other features. 

Figure 2. Revised coordinate system 4 x 4cm from Y12 to Y15.

From Y12 to Y15 the coordinate grid was adjusted to include points every 4 by 4cm. 

Push pins were placed every four centimeters around the edge of the rectangle, and then string was wrapped around it to produce the 4 x 4cm squares. 

This area was selected because the narrow nature of the valley. The other features within the sandbox were large enough to be recognized in the original depictions of the digital elevation models. 

Figure 3. Casey Aumann recording the z-values using
the 4 x 4cm coordinate system. 

Data Collection

Again, a meter stick was used to measure the depth of the features below sea level (from the top edge of the rectangle). 

The meter stick was lowered until it was touching the sand and hovered there until a measurement could be read. The meter stick should not have gone below the surface of the sand. 







Digital Elevation Models 

We used the Natural Neighbor, Spline, IDW, and Kriging interpolation methods to analyze how well our coordinate system represented the features in the sandbox. Side by side comparisons are shown below to view the improvement of using the 4 x 4cm grid system through out the valley area. 

Natural Neighbor ( "area stealing" interpolation) 

The Natural Neighbor method places value on the cell depending on what values lie inside of it. 

Figure 4. Natural Neighbor model with
4 x 4cm coordinate system. Made in ArcScene.

Figure 5. Natural Neighbor model withoriginal 8 x 8cm 
coordinate system. Made in ArcScene.

Spline

The appearance of the spline method is smooth because it takes into account the overall surface curvature and tries to minimize it. The points go directly through the input values. 

Figure 6. Spline model from revised data
with 4 x 4cm coordinate system. Made in ArcScene.

Figure 7. Spline model from original data with
8 x 8cm coordinate system. Made in ArcScene.

IDW (Inverse Distance Weighted)

This method values the center point of the cell, making the peripheral values of the cell less important overall. 

Figure 8. IDW model from revised
data with 4 x 4cm coordinate system. Made in ArcScene.

Figure 9. IDW model original 8 x 8cm
coordinate system. Made in ArcScene.

Kriging 

Kriging places high value on the z-values of the cells. It can take many z-values and predict the surface. 

Figure 10. Kriging model 4 x 4cm coordinate
system. Made in ArcScene.

Figure 11. Kriging model 8 x 8xm coordinate
system. Made in ArcScene.

Metadata
Who: Casey Aumann, Ally Hillstrom, Morgan Freeburg
What: Grid based surveying of a sandbox with different features
When: 22 September 2015
Where: The bank of the Chippewa River underneath the UWEC footbridge
Why: To revise the technique of a grid based surveying technique


Discussion

The spline and natural neighbor interpolation methods best represented our elevation surface. The difference from the 8 x 8cm grid system to the improved 4 x 4cm grid within the valley area is clearly visible. The valley has distinct boundaries, especially with the spline. With the  Kriging method , the valley still is much improved and visible however the banks of the river are somewhat blurry and it seems the features run together. IDW does not appear as accurate for our sandbox surface either because of the pixelated nature of how it is display. The IDW is giving the surface the appearance that the features are choppy and disconnected. This is where the spline method comes in between a happy medium of the IDW and kriging. The features run smoothly together, but not so much that they are not distinct. 

The second time around as far as collecting data was simpler in the sense we knew what we needed to improve on and how to go about getting the materials and setting up. On the other hand, it did not take us any less time to collect the data.


Conclusion

We were more confident in our abilities to create and execute grid based surveying system. We did not realize we could have selectively chosen the valley area to include more points in the first trail, we thought this would inflict some bias. Looking at the surface and features that were created it only made sense to subjectively select this area for more surveying points.For this extension we challenged our critical thinking skills and were able to overcome the obstacles we had in the first activity. 

References
ESRI. (2015). Comparing interpolation methods. ArcGIS Software (Version 10.3) [Software]

Activity One: Creation of Digital Elevation Surface

Introduction

This activity was to serve as an introduction as to how to go about creating a coordinate system and to evaluate a digital elevation surface. In order to do this it was required to include a linear peak, valley, hill, depression, and a plain. In the real world digital elevation models are used in ways ranging from ecologists estimating wildlife distribution to natural disaster specialist to see the impacts of that disaster (NOAA). Bamber and Gomez-Dans (2005) assed two different methods to analyze the accuracy of the digital elevaion model of Antarctica. Reading through this article shows how intricate these digital models must be and how critical thinking and problem solving plays a huge role in determining the accuracy of digital elevation models. Our task was much simpler and on a much smaller scale, of course, but none the less to think spatially and problem solve. 

Study Area

• Date: 18 September 2015 from 1:30-4pm 
• Sandy bank of the Chippewea River Valley 
• Conditions: Overcast, Average temp. 60.7°F (retrieved from Weather Underground)

Methods

In the sandy bank of the Chippewa River a 122cm by 125cm hollow rectangle was placed and leveled. To level the box we placed a level on each corner of the rectangle. We decided we would fill sand up to the bottom edge of the rectangle. To level the sand we placed a level on top of a meter stick placed in the center of the rectangle. We measured the levelness along the x-axis and y-axis. 

Figure 1. Students Casey Aumann and Ally Hillstrom building the features.

Construction of features 

Once the sample was leveled and ready to go, the features were constructed. A linear peak, plain, depression, valley, and two hills were shaped from sand as well.

Figure 2.  The final constructed features. 

Definition of features

Linear peak:  A ridge with a high point that continues 
Hill: A peak with an isolated high point with a rounded top
Depression: The opposite of a hill, where there is a low point
Valley: An elongated lowland channel
Plain:  A flat expanse of roughly the same elevation

Figure 3. The coordinate system 8 x 8cm squares. 

Making the Coordinate System

First, we placed push pins every 8cm around the frame of the rectangle. String was wrapped around the pins to make 8 x 8cm squares within the rectangle. 

Second, the coordinate system was labeled using numbers along an x and y axis. For example, the squares were labeled X1, Y1; X1, Y2 and so on. 

Figure 4. Casey measuring the z-values and Ally entering the data. 

Data Collection

All measurements were taken from the upper right-hand corner of each square using a meter stick.

This coordinate system produced 210 points and the measurements were recorded directly into an excel spreadsheet (shown to the left).   

Figure 5. Data in excel. 

The z values are negative because the top edge of the rectangle was assigned sea level. When transferring the data into ArcGIS we will multiply all of the z values by -1 in order to create positive values. 

Discussion

This activity allowed us to get our hands dirty and experience the challenges of creating a coordinate system and trying to establish a digital elevation surface.The challenges we came across ranged from choosing where sea level should be to selecting the best spot to represent the whole square in terms of elevation. Sea level was set at the bottom edge of the rectangle because then all of the measurements would be negative, and it would make measuring the features in the middle of the rectangle easier. For example, it would be difficult to get an accurate measurement of the middle of a linear peak without destroying the feature. It would be interesting to see how this is achieved successfully using different methodology. Another problem that occurred was with choosing how big to make the squares within the rectangle. Eight by eight squares were chosen because they were small enough to ensure a representative survey of the features within the rectangle. However, there were sections of the rectangle we disregarded in order to have a perfect square because we did not know how to correctly record a value from less than a whole square. While recording the z values some error could have occured because two students took turns measuring. Also, the values could have varied a bit because the meter stick could have gone into the sand instead of representing the surface of the feature. The conditions allowed us to work without problem, with that said it took a substantial amount of time to work through the difficulties mentioned above and then to measure all 210 points. 

Conclusion

In conclusion the difficulties we experienced helped us develop our critical thinking skills and our ability to think spatially. This activity taught us that developing and evaluating a digital elevation surface takes a great amount of time and serious consideration of the available tools. It was a great starting point to introduce this topic and to introduce us to thinking spatially. 

References  

Bamber, J. & Gomez-Dans, J. L. (2005). The accuracy of digitally elevation models of the Antarctic continent. Science Direct, 273 (3-4). 516-523. doi:10.1016/j.epsl.2005.06.008

NOAA National Centers for Environmental Information. Digital Elevation Model (DEM) Discovery Portal. Retrieved from: https://www.ngdc.noaa.gov/mgg/dem/