What it takes to be a Scientist: A personal account into testing a new method
Editor's note: Trying to find new ways to gather data is never easy - and as Geology student Keeley Chiasson explains in this story, it's a process of trial and error, of learning to overcome frustration and dead ends, and, hopefully at the end, a feeling of reward and accomplishment.
March 2018
We drove up logging roads on the Van Zandt dike, based in the foothills of the North Cascades, Washington. Several turns seemed to lead nowhere. When we reached a dead end, Geoff turned off the car.
“The trailhead is right over there,” he said. It was barely noticeable, overgrown and tight with new growth trees. We slung on our packs and ducked under low-hanging branches to begin our march through the underbrush. I contemplated what it means to be a scientist and how people come up with these ideas.
Geoff Mallick graduated with his masters in geology from WWU in 2017. His thesis was looking at an ancient landslide complex in Van Zandt, WA. I followed him towards his previous study site, which would be my new study site. After about ten minutes following the shadow of the trail, we crested a ridge. I looked over the edge and could see how the terrace below us had once slipped away from the ground on which we walked.
Geoff then pointed out holes in the ground. I tried to see where the holes ended, but they seemed to lead towards the center of the earth. As we got closer to the study site the terrain became increasingly more chaotic. The forest floor had been obviously disturbed by something. I had never seen anything like it. At the site, Geoff showed me how he had found several surface cracks into which he installed motion sensors, called wire-extensometers. These sensors measure one-dimensional expansion and contraction of these surface cracks.
Later, I interviewed Geoff about the significance of the area. He said, “Van Zandt was a big scary landslide that again has active deformation. Local people talk about hearing rock fall and they are concerned... It has a history of multiple failures and is currently active in many ways. You go up there and it’s the landslide textbook.”
I learned that this area stood out to residents of Van Zandt and local geologists because of the giant recent landslide in Oso, Washington which in March of 2014, drowned the town in mud and killed 43 unsuspecting people. This begged the question, how can we understand the timing and mechanics of landslides in order to prevent future catastrophes?
The difference between Oso and Van Zandt, however, is the specific type of landslide hazard. The failure in Oso was related to loose sediment, whereas the problem in Van Zandt lies within the bedrock. In the Van Zandt hillside, tilted layers of strong sandstone alternate with weaker layers of mudstone. When the weaker layers someday erode, gravity will win, pulling the strong layers downslope.
“Really high velocity, [with] long runout, covered valley floor, with 10, 15, 200 feet of debris? If that happens again, you’re going to die,” he said.
June 2018
The sun beat down on my shoulders. I was sweating from carrying a pack filled with unusual supplies: drone, laptop, and miscellaneous tools. Three colleagues and I, with skis strapped to our feet, tramped over bare ground between snow patches. The snowpack is in sight, and Mt. Baker towers ahead.
David Shean, a professor at the University of Washington, has been working with WWU grad student Liza Kimberly, and they needed another assistant to help carry gear to the Easton Glacier, so Liza invited me to join them. Shean is working on an ongoing project looking at glacial change. For this study, he used a field technique called Structure from Motion (SfM), which required three drones and a significant pile of expensive research equipment. He needed help slogging the gear to the toe of the glacier. I was invited to join in on the fun, and I could not turn down an opportunity to do science and ski!
Plodding up the glacier with targets and GPS in hand, Liza Kimberly, my colleague and connection with the professor, and I skied from location to location along the bottom portion of the glacier.
“David is this a good spot?” Liza communicated via walkie-talkie.
“Nope, a little more to the left.”
“Ok, how about here?”
“Looks good!”
Prior to this trip I had never heard of SfM. Scientists have been using this relatively new tool to explore progressive landscapes. Liza explained that the process is straight forward. Take a sequence of overlapping images and upload them into a program called Agisoft. The targets allow the program to connect the dots between images, which then can build a 3-D model. The method is fairly inexpensive, outputs high quality data, and is accessible to the backyard scientist.
There are multiple ways to collect photos. In many situations using a handheld camera, or even an iPhone, attached to a selfie stick is sufficient. We, however, were using drones. We recorded the GPS coordinates to tell the program where the photos are located geographically.
Professor Shean goes out to the glacier a couple times a year, so he can generate several models. He can then stack the models on top of each other to show how the glacier’s elevation changes over time.
At the end of the day, the four of us ski down the glacier and I found myself deep in thought, captivated by the concept of SfM. Can I, I wonder, apply this concept to my undergraduate research at Western Washington University?
April 2019
I was not sure if my idea would work, so I consulted my advisor, Professor Douglas Clark, about the viability of testing SfM in this area. He advocated for me trying it out, saying, “If we can figure out an effective way to monitor landslides where we don’t have to have an extensometer [in the field] all the time, we could go up to the study area for repeat Structure from Motion surveys. It would be a very effective way to track motion over time and is potentially a lot simpler and cheaper.”
I left his office feeling confident that if I could get SfM to work, we would be able to see deformation in this area in three dimensions, allowing for more nuanced information.
I was under the impression that it would be a fairly straight forward process. Thinking back to the Easton Glacier, the most complicated aspect was avoiding crevasses while placing targets, the rest was up to the drones. As my study area is very small in comparison, the process should be straight forward. I would print off targets from Google. Instead of drones, I would use my personal handheld camera. I would clear out distracting debris, place targets, and proceed with taking photos of the surface cracks. Easy.
However, on my first go around with SfM I found it was not that simple. First, lighting was a problem. Blurry photos haunted me. Second, moss growth, vegetation, and trees impacted surveying, distracting the program. Third, the ground itself moves; the program needs at least one stable point to use as a reference for repeat surveys.
These three things created too much error to build a first model. The precision I needed to track millimeters of change was daunting. Professor Shean’s research was more straight-forward: glaciers don’t usually host plants. I questioned if my work was worth the hassle, and if I was on track to being a ‘real scientist.’
In a heart to heart with my advisor, Doug reminded me, “It’s not all on you. Your field site is not ideal, but if you can create a step-by-step procedure on how to apply Structure from motion, this can open up other opportunities for [future] student research.” I appreciated his kind words and reinforcement. Even if SfM doesn’t work for Van Zandt, this process is still beneficial.
June 2019
I drive myself up winding logging roads until I reach the now familiar dead end. I get out of the car, lace up my hiking boots, sling on my pack, and leave the car behind. As I scoot under low-hanging branches onto the now well-known trail, I do a mental check list of the contents of my pack: laptop, camera, Sven saw, drill, and miscellaneous tools.
I trudge through the bushes, crest the ridge, and find my way to the first surface crack. I look around me and set down my pack. All of the debris needs to be cleared away before I can proceed with my second trial of SfM. Considering my site is in characteristically thick, dark Pacific Northwest forest, preparation is going to take a while.
Being a scientist means trial and error. It starts with an idea and sometimes that idea doesn’t pan out as expected. Science can be frustrating, especially when trying to learn a new method and apply it at the same time. But it can also be rewarding. Whether this method works or not, I am still learning a really cool tool, and am providing a stepping stone for future students to use on different projects.
I take a breath, select the Sven, and begin sawing away a dead tree.