Background:
As we ventured out in the field this week our class was
tasked with revising our original field surveys using a different more refined
method of surveying. After all groups had completed the first sandbox our class
collectively went out to inspect and critically investigate how groups surveyed
and overcame any issues with designing and collecting the original data points.
We found that most groups had decided to use a uniform grid survey technique.
We also found that most groups to counteract the negative numbers had added the
width of the board to raise the original base of the box. Both of these techniques
is what our group had used to collect the original data points. Once we had
discussed the ways we had surveyed the sandboxes we now where tasked to find
more precise ways of survey elaborate features to accurately display in 3-D
imagery.
Figure 1.1 original formatting
for collection of data points did not work for importing to Arc
Figure 1.2: Desired formatting
for Arc GIS all columns are same data and each column does not included
equations.
Methods:
After all of our data points where collected we needed to
display our sandbox in Arc scene 3-d imagery. To do this we needed to import
our Microsoft Xcel table to Arc in order to display our points. However, being inexperienced
and new to using our own data our group was not aware that a specific format
needed to be implemented in Arc to be able to import the data. Our original
table was set up to have easy input of each z value so we started our grid in
the top left corner of the sandbox and laid the grid a-e from top left to top
right and 1-14 from top left to bottom left (figure 1.1). We had thought since
this layout was the same as a excel sheet is formatted it would be easier to
input the values. What we quickly found out is that we needed to have a table
with three columns of like attributes for the input process to work. (Figure
1.2). You can also see in figure 1.2 that we changed our alphabetical labeling to
replicate our grid inches to be increments of eight. Although this change now
allowed us to import the table and display the data it was not the table we
needed. We found when we displayed the data that our features in our sandbox
where all inverted. After trouble shooting the issue we found this was due to
our original false origin being placed in the top left corner. If we wanted to
keep the original false origin we would need negative x-values for our grid
which you cannot have in your table so we had to move the false origin of our
grid to the lower left corner of our plot. Lastly our group found it to be
easier to change the table grid to replicate our centimeter values of our
sandbox meaning the table would go 1-120 in increments of eight centimeters. It
was this change that allowed us to make it easier to input more defined values
in our new stratified survey. We had decided to stratify within our original
grid marks which made it possible to get accurate height values on any area
with features without having to survey more points in areas not holding
significant features.
Results:
Once we had brainstormed and resolved all of our
complication of our original data we were able to display our sandbox features.
For the first set of data points we were asked to generate 3-d images in five
different interpolations, consisting of Inverse Distance Weighted (IDW),
Natural Neighbors, Kriging, Spline, and Triangulated Irregular Network (TIN).
Once displayed we needed to determine which one was the best representation of
our survey and why it was the best fit.
IDW: Inverse Distance Weighted is a interpolation is usually
used with surveys that has a very localized area with many data points. It uses
a range of values to interpolate the true features and is limited for it can
only range from the highest and lowest values of your data table. It is because
of this limitation that the IDW was not one of the better choices for our
sandbox display. The display was very inconsistent and had a lot of pore ridges
as seen with the bumpy surface below.
Natural Neighbors: Nat Neighbors
was a little better representation for our model for it does not interpolate
any ridges, pits or hills that do not show up in the data. It uses a notion of
“area-stealing” meaning that it finds an average height from all surrounding
data points to infer what the feature looks like. This allows for smoother
surface but also leads to shifts and variation from original location of
features.
Spline: Spline or thin plate
interpolation is one of best models to use when you find yourself surveying
gently sloping areas and features with gently elevation change as we were with
our sandbox features. Once modeled spline minimizes overall curvature which
gives a gently surface to the display as you can see in the image. It is the
smooth surface and best display that made us pick to use spline for our second
survey.
TIN: Triangulated Irregular Network uses raster and our
defined z points to design and connect multiple triangular points that are
connected to show your features. However, with little data points our TIN became
very rough and didn’t fit our field feature very well.
Second survey:
For our second survey we decided to redesign new feature
since our original design was destroyed by the weather. However, we keep all our
same layout and grid patterns (image 1.3) but with our new knowledge set our
false origin in the lower left and input the data with three columns and grid
of increments of eight centimeters but stratified in all areas where there
where features. The final product was displayed in a spline model for it best
represented our features.
Image 1.3 Grid
pattern of second sandbox trial with new features. We measure all stratified
areas ever 2cm where needed
Conclusion:
Overall this project was very helpful to our
group for it allowed us to see firsthand how critical prep work is and proper
formatting when collect data point in the field. We found that if you try and
cut corners or take an easy route it can cause you a lot of pain and time in
the long run fixing your mistakes. I hope we can continue to get a lot of filed
work for this way of learning is pleasing and laid back way to learn.
Discussion:
The only negative parts to this lab for me was
the challenging landscape in which we were given to build our features. I wish
we could have had a closer location to build and preserve our features. If this
was accessible it would be easier to make more elaborate features knowing you
could keep them same if you needed longer time to collect points.
Sources:
ArcGIS help online
No comments:
Post a Comment