Introduction:
Last week was an introduction to the world of 3D modeling;
we had the opportunity to go out and create our own terrain and find our own
way of surveying the carefully crafted model. The group met, refined, and
re-did the previous exercise. We used various interpolation methods to analyze
the survey we completed and decided the best way to show what we created.
Methods:
We started this activity by placing our original data
into ArcMap. We did this by using the Add X, Y tool and imported the coordinate
pairs and displayed it as a point based feature class. ArcMap is a handy
software that provided us with many ways of representing our data by using
interpolation methods. We entered our point feature class into 5 different 3D
analyst tools; Inverse Distance Weighted (IDW), Natural Neighbors, Kriging,
Spline and Triangular Irregular Network (TIN). (see Figures 2-1 through 2-4
on the Photos Page) These methods use our point feature class and form an
estimated surface from the points using the z-values. They essentially use the
points around each other to form a value, and when in 3D a slope, to satisfy
the elevation variations of the points in the feature class. After creating all
of these rasters it was possible to bring them into ArcScene and view them in
3D. The best interpolation method was chosen to represent our first survey of
the terrain we created, Figure 2-4.
After analyzing the models that were created it was
plain to see there were not enough points collected to properly show all of the
terrain's features. This brought an opportunity to go back out to the sandbox
and create a new survey method to collect sufficient data to help display the
created terrain. After reforming the terrain's features the group concluded on
a larger resolution, 5x5 cm. We set up the grid similarily to our first
exercise, but this time we stretched a string across the length of the sandbox
at intervals of 5 cm. (see Figure 2-5) After pinning down the string
across the sandbox we marked up the side walls every 5 cm and used those
markings to lay a meter stick across the width of the sandbox. Using that meter
stick we took measurements every 5 cm to come out with 22 measurements for each
Y coordinate. (see Figure 2-6.) After the completion of our survey we
had a total of 1056 points, quite the amount to type into a data table. Once
the data was placed into a Microsoft Excel file we brought that table into
ArcMap 10, and ran the Add X, Y tool once again, giving our data a spatial
extent. The point shapefile was ready to use for running the 3D analyst tools
once again.
Discussion:
Our first survey got the designated job done, however
it did not have nearly enough points gathered to give a proper visual
representation of our terrain. After running the interpolations for our first
survey it was interesting to see the results of each method. The method that
best projected our terrain was Spline (see Figure 2-?); it gave our features
the most realistic edges and projected the most properly. IDW gave very
interesting results, the method works in such that it works as a function of
inverse distance, the point measures out a distance and is influenced by
certain points in the immediate vicinity. This gives the IDW method a circular
look to each point location; (see Figure 2-4) this does not work well if
you are looking for a realistic visual representation of the terrain that is
surveyed. Though the 3D models worked very well for the points we collected we
noticed some bad errors, we created a river in the southern part of the sandbox
and as you can see from Figure 2-2 you can barely tell it is there. Also
there is a ridge across the northern part of the map that was not meant to be located
there, you can compare the photos of our terrain from Figure 1-6 and see
there is no ridge in the northern part of the terrain. These errors would not
have happened had we known to make smaller measurements; luckily this
assignment allowed us to go back and recreate our terrain and make more
accurate survey points.
The second time around we decided to make measurements
closer together as stated above, the results were significantly more accurate
to what we created. Once the point feature class and the interpolations were
done it was fascinating to compare the results. The maps do not look much
alike, the new ones were significantly more accurate and when compared to the
final product photos you can actually see what we created in our sandbox. Once
again I chose the Spline method to represent our terrain as it worked very
nicely to give shape to our features created as seen in Figure 2-7.
If this lab was done again I would bring very fine
string to create a full grid on the sandbox to help minimize the error of
figuring the location of each measurement. Another adjustment would be to make
sure all features do not break the top of the sandbox because it would help
alleviate error once again by adjusting measurement methods by having to switch
from measuring down and up from the string level. Otherwise the group was able
to make the best adjustments for completion of the exercise the second time
through.
Conclusion:
This exercise was a great opportunity to produce a 3D
rendition of a real life landscape, look at what you did wrong, and go back,
refine, and redo the entire process. The chances to go back and clean up the
errors you produced or adjust the surveying techniques you thought were good
enough was critical for the success of what we had to accomplish. We didn't
know what we were getting into the first time around and essentially 'winged
it' so it was great to go back and do it the right way. The group performed
very well together, we had no issues distributing the work load and discussing
the best way to accomplish what needed to be done. I learned that there is
always room for adjustment when doing something and that one way is never the
best way of doing it. This activity would be best if it wasn't below freezing outside;
otherwise I thought it was a great opportunity to form our own methods of
completing the exercise.
No comments:
Post a Comment