Field Activity #2: Visualizing Terrain Survey

Introduction

In this second portion of the terrain model activity the goal was to use ArcGIS in order to create a model of the data that we collected in the landscape field activity. The purpose of this activity was to go over our sampling scheme, see if additional points needed to be collected and resurvey those areas where needed. This labs goal was to create a 3D model of our landscape in order to showcase ArcGIS skills along with field methods.

Methods

In order to create a model of the landscapes the data that was initially collected needed to be imported into ArcGIS. This is done by converting the Excel spreadsheet into a dbf file format which works well with Arc. Once in Arc the user simply imports the dbf file into Arc as an xy file with z coordinates as well. These were then exported as a shapefile so they could be interpoled for spatial analysis.

Image 1: X, Y and Z values from Field Activity #1. The grid data was placed into a field notebook then transferred over to Microsoft Excel later, however as the project went on the group decided to place some of the points directly into a laptop for convenience sake.

This data then needed to be interpolated in order to get a continuous surface. The five interpolation methods that we utilized to create those surfaces included: IDW, Natural Neighbors, Kiging, Spline and TIN. Each of these methodologies use different ways to project your data to a surface model. They each have their own advantages and disadvantages. All of these methods can be found using the help portion of the Arc Toolbox and are easily accessible.

The IDW method uses weighted combination sets of sample points in torder to generate cell values. Points must be very dense in this method so as to consume the variation of the surface. The weight that is assigned to each point is a function of the output cell location. Because of this the greater the distance that a point is the less influence a cell has on it.

Image 2: This image shows the IDW output. The data is somewhat choppy and rigid. The edges of each elevation is frayed out and rather inaccurate compared to the real landscape. 

Natural Neighbor also uses the weighted method to interpolate the points. It can be used for both interpolation or extrapolation. It is best used with scattered points that are also clustered in groups. IDW and Natural Neighbor are very similar in their equation. Local coordinates will define how much influence scattered points have on a cell.

Image 3: The above image is the output of performing the natural neighbors interpolation method. This method is smoother than the IDW method, but is still rigid at perceived edges in the data. This image shows a fairly accurate description of the landscape.

Kriging on the other hand assumes that points distance and direction reflect correlation. It is best used when there is bias in direction or distance in the data. It uses very sophisticated interpolation abilities to calculate weighted averages.

Image 4: This image is the product of the Kriging interpolation method. This method appears to do a good job of portraying highs and rounded portions of data, but is still lacking a smooth transition that the group wanted to obtain to best describe our snow landscape.

Spline is used to estimate values using mathematical fuctions. It attempts to create a very smooth surface that passes through each point. It has been said that the spline method is like bending rubber through each point to get a smooth surface.

Finally, a TIN is known for its ability to accomodate unevenly spaced elevation points. A TIN uses irregular triangles to connect points to create models of elevation.

After all of these methods were brought into Arc and altered to best fit the data they were then imported in Arc Scene where data is modified into a 3D version. Some of the methods did a better job than others at creating a surface model of our original landscape. The image below is a reminder of what the actual surface looked like.

Image 5: The final product of our landscape building. Areas of interest on the image above include a small hill at the top right of the image, a mountain/volcano at the center, a plain at the top left, a valley just in front of the mountain, a depression atop the mountain and a ridge near the bottom.

Image 6: The results of the Spline. The best method in my opinion. The original landscape can be seen in image 1 above. The spline method creates a very smooth graphic, which does a great job of portraying the original landscape. The only thing that I noticed was that the digital image appears to be inverted.

In my opinion the spline interpolation method was the best at creating an accurate model of our original landscape. Image 6 above shows the results of the spline method when imported into Arc Scene. Another method that did a good job at giving accurate elevation details was the TIN. The TIN image can be seen below in Image 7.

Image 7: TIN model of the original data from Field activity #1. This TIN image compared to the spline (image 2) above is much more chopping, however does justice to the differing heights of the model. The color scheme seems to stretch better and portray elevation.

When our group revisited our data and box, it had already snowed on our box and ruined any chance of us getting more data where it may have been lacking. While a resurvey could have been justifiably used, I believe that our 5 cm x 5 cm grid data did a superb job of recreating our final results. In essence our group decided that resampling and attempting to gather more data would have been near to impossible. I would suggest that future students try to accomplish the whole assignment in a days time to avoid the snow issue.

When all is said and done, our model came out very clean and accurate of our original terrain model. The spline is probably the most appealing because of its smoothness.

Discussion

I thoroughly enjoyed using Arc GIS and Scene in order to create the terrain models because of the ability to compare the digital image to the original creation. Being able to see a physical landscape in a 3D environment is incredible and could definitely be a valuable tool to model actual regions.

It would have been nice to revisit our landscape to see if we wanted to add additional points, but the weather really put a halt to that. None the less, I believe that our data came out much better than I would have expected. I would say that the group's ambition worked out quite well. As a group it was hard to find a time that worked for everyone and more often than not not all of us were able to attend each group meeting. When everyone was together we were able to work as an efficient team that also was able to joke around.

As I put it in the last blog, I wish we could have come up with a better method to take our elevation, but it seemed to all work out in the end. Higher accuracy in our measurements would have been nice for the perfectionist in me, but our models worked better than expected.

Conclusion

This activity taught me many things about field work. For one, ambition can kill you when it comes to data collection and accuracy. If we were to have a flaw though I would want it to be excessive ambition because it shows that we are truly trying to make the best quality work that we can.

I also learned that having more tools in the field are better than less. Its best to be overprepared. Also that delegation is incredibly important to efficiency in a group environment.

Finally, the Arc GIS and Scene portions of the lab were very helpful in creating surface models and showcasing those results. Some group members were able to learn more than others about the scene portion of the activity. Arc is an incredibly helpful and powerful tool in the Geography world so it is great that the course is already implementing it. I'm excited to see what other new tools and processes that we will be introduced to.