Field Activity #4: Data Collection for the Creation of a Digital Elevation Surface: Survey Techniques

Introduction

Sampling is a key concept in geography. Since it is impossible to attain an infinite amount of data with the constraints of limited time, money, and human power, it becomes critical to obtain a representative and statistically valid sample that can be reasonably used to draw conclusions about the whole. The strategy in choosing a sampling method that will give reliable data is foundational to any field-based project. Geographers cannot record infinite elevation points, for example, nor could a computer process that data, so instead careful consideration must be given to a strategy to simplify the collection of spatial data.
According to online resources from the Royal Geographic Society (visit the link here), there are three main categories of sampling strategy:

·        Random - sample points chosen at random

·        Systematic - sample points chosen based on a spacing system

·        Stratified - Used when population contains subgroups, samples chosen from designated stratification

As with any simplification, these sampling methods are not perfect. There are tradeoffs to each regarding simplicity of data analysis and collection vs. data accuracy—cost vs. accuracy.

Lab objectives

 The artist at work; Amanda constructing the sandbox landscape.

The goal of this field exercise is for the 
class to develop the ability to create a Digital Elevation Surface using critical thinking skills and improvised survey techniques. This is a two-part field exercise, but the objectives reflected on in this post were:

1)     To understand the pros and cons of the different sampling strategies in order to make an informed decision about which technique works best for mapping the sandbox terrain

2)     Create a unique terrain containing a ridge, hill, depression, valley, and plain in the sandbox provided.

3)     Conduct a survey that provides accurate X, Y, and Z coordinates and compile them in a normalized spreadsheet that can be downloaded to ArcMap 10.4.1

Methods

Students were assigned groups of three. Group 2 was comprised of Jesse, Amanda, and Zach (who was absent for a field trip).  Groups were encouraged to read up on sampling methods and develop a strategy. We chose to use a stratified systematic sampling technique. The advantages of this technique is its flexibility; we could choose to collect more data points in area of higher relief without having to increase our overall point density in areas of low relief. We also could extrapolate the data via comparisons and correlations of sub-sets.  

Figure 1 The red X shows the location of the sandbox site. 

The sandbox site was located east of Phillips Hall on the UWEC campus (Fig.1) The area includes a privately owned home and yard, but directly across the Little Niagra stream from the home is an open grassy area where the class set up two sandboxes. The sandbox terrain and features (Fig.2) were unique to each group. 

Figure 2 The black labels highlight the required features. The red labels highlight the features of the landscape that we were given free reign to create. The landscape is entitled: Voyage to the Land of the Llama king and his Happily Subservient Peasant-People.

Materials

·        Samsung Galaxy S6 Edge

·        measuring tape

·        meter stick
·       push-pins

·        Colored string
·    1 meter x 1 meter sandbox

For the survey, a push-pin was inserted every 5 cm around the rim of the sandbox. Then the
colored string was used to denote the lines of the grid (though this method was soon abandoned as
we decided to draw a grid onto the landscape using the meter stick). Sea level (our arbitrary
“zero” elevation) was the plane formed by the top of the sandbox’s walls. An elevation
measurement was taken in the center of each grid square and recorded as X, Y, and Z coordinates.
Areas of high elevation changes were marked with a push-pin and then measured and recorded
(Fig.3). This allowed for higher detail to be recorded where more feature definition will be
needed. Our coordinates were then transferred to a spreadsheet containing X, Y, and Z fields
(Fig.4). This spreadsheet will be imported into ArcGIS in the next field exercise.

Figure 3 The grid was drawn onto the landscape and push-pins used to denote areas of high-relief.

Figure 4 The normalized spreadsheet that will be used to create the Digital Elevation Surface.

Results and Discussion

Upon conclusion of the terrain sampling, 179 data points had been collected, all with X, Y, and Z values.

Minimum elevation value: -16.00cm

Maximum elevation value: 12.00cm

Mean: -4.71 cm

Standard deviation: 4.64 cm

The chosen sampling method served the group’s purpose, but the data collection definitely could have been more accurate had the grid been squares smaller than 5x5cm. In hindsight, a grid of 1x1 cm would have been better. Also, being down one person made the sampling difficult, but it would have been better to take more data points.  The sampling technique used for the first two rows of grid squares involved using the colored string, but for the rest of the rows, a method of drawing on the grid lines directly using the meter stick was used. Switching procedure during data collection may have been a source of error. There was also uncertainty in whether the drawn-on grid lines were precisely parallel and spaced evenly. This was another likely source of error, though the error was likely innocuous.

Conclusion

The inherent disadvantages of a stratified systematic sampling technique is that the proportions of the sub-sets must be known and accurate in order to maintain data integrity. Our system fell short on that; there were too many approximations in the measurements.

This activity demonstrates the importance of sampling techniques for collecting spatial data; even in a 1x1 meter sandbox it would have been labor-intensive to attain accuracy down to the centimeter level. Studying real-world geographic features requires an accurate, scalable sampling system in order to collect data in reasonable amount of time.
It will be interesting to use the numbers gathered during this sandbox exercise to create a DES. The uncertainty in the data may cause some discrepancy between the model and our actual sandbox terrain. 179 data points may prove to be less than ideal, perhaps something like 500 points (1x1 cm grid) would have been better.

One last look at the llama keychain, the focal point of our sandbox terrain.

Sources

Royal Geographical Society. Retrieved October 5th, 2016 from http://www.rgs.org/OurWork/Schools/Fieldwork+and+local+learning/Fieldwork+techniques/Sampling+techniques.html