This is Part 2 of a 3-Part series on an ambitious summer research project exploring how trees manage their water supplies and respond to drought. Part 1 is here and Part 3 is coming soon!
by Sydney Widell – “The thought of this breakfast is what’s kept me going,” said Brian Schlaff, when we arrive at the Outdoorsman Restaurant in Boulder Junction at 8:00 a.m., after five weary hours of dawn patrol field work.
I’m with Brain, Emiliano Rosel and Mike Krellwitz, undergraduate researchers working for PhD candidate Dom Ciruzzi at Trout Lake Station. Also joining us is Professor Steve Lohide, Dom’s advisor who is in town for the day from Madison, and Linden Taylor, another researcher who just happened to be awake.
Dom studies the way different species of trees use groundwater and respond to drought, and his research could lend scientists and conservationists important insights into forestry and water management.
He has been studying individual trees from the ground all summer, but that’s another story.
Today is a pretty big day for Dom. A plane is flying up from Madison to collect aerial imagery of the nearly 300 square miles of forests he’s researching, in an effort to monitor the water status of those trees from the air.
The plane will begin to collect images at noon, when trees are converting the most sunlight into energy — and also using the most water. If everything goes well, the flyover will produce the most highly resolved imagery of the Trout Lake watershed ever collected.
Dom has been planning for this flyover all summer, but at this point in the morning, it’s impossible to say if it will actually happen.
Weather has a major impact on the flyover’s success, and right now, the forecast is not looking great. A sheen of haze covers the sky, and is expected to remain for the rest of the day, according to Dom. He’s had his eyes fixed on the radar all morning.
Everyone on the team seems a little on edge — a lot of planning has gone into this morning, but so many factors remain uncertain.
“Clouds have been a tricky problem to deal with because we can’t really anticipate them until the day before, or even the day of,” Dom said. “Any time there’s a cloud casting a shadow on the forest and we take an image of the area that’s shaded, that data becomes unusable.”
When Dom attempted to coordinate a flyover last month, high winds and cloud cover forced the plane to turn around before it had completed its survey.
The day started at 3:00 a.m., with a round of pre-dawn sampling. Since Dom will have to collect leaves from his sample trees around noon — when the flyover is supposed to happen— to calibrate against the remote imagery, he also wants to compare leaves before and after they begin to capture energy from the sun, which is why we start our sampling in the dark hours before sunrise.
“I think this morning was flawless,” Steve said. “We got more done than we had hoped to — one extra site even. Every tree we needed leaves from gave us leaves.”
The early morning sampling doesn’t always go so smoothly. In the dark, it’s hard to spot the target leaves, and sometimes, the sample trees themselves elude our flashlights.
In the forest, Mike and Emiliano felt their way along familiar trails suddenly more foreign to them in the darkness. The woods were nearly silent when we began to sample. The only sounds were the distant growls of semis racing north up Highway 51 toward Hurley.
Brian and Steve sampled at a different site, and ultimately, Dom and I would meet back at the lab to process the leaves as they arrived from the field.
After our work in the woods was finished, Brian, Mike and I drove to the nearby Crystal Lake campground, where we set giant, reflective tarps in an open field. Because they reflect light at known wavelengths, the tarps will help calibrate the imagery the plane produces when it flies up from Madison in a few hours.
Instead of just registering visible colors, the sensor collects a multitude of wavelengths that will give researchers information about the amount of water moving through the trees, Steve explains as we sip our coffee and wait for our breakfast to arrive. Where the camera on your phone can capture imagery on three spectral bands, the sensor uses more than a thousand.
Specific trees reflect light at specific wavelengths, meaning each tree has a “wavelength signature” that the sensor can detect. If conditions are right, Dom can identify and analyze individual trees from the air.
The remote sensor collects imagery a lot like the panorama function on your phone, only in this case, the “panorama” is a swath 17 miles long and ¾ of a mile wide – an area that translates into 300 million pixels. Each pixel contains information from thousands of spectral bands the sensor can register. Dom estimates that he could generate some 50 terabytes of data this afternoon — incredibly precise images.
“The point of this project is learning how to scale up these really intense observations of individual trees to the entire forest, because we want to know how all the trees are doing,” Dom said earlier. “Satellite images contain information on many trees, which is fine, but we would like to find a way to get data on individual trees everywhere.”
The existing satellite images Dom could have used have a resolution of 30×30 meters. That means one pixel of imagery would have contained 30 square meters of forest, or roughly seven trees. Dom needs enough pixel resolution to discern individual trees at his research sites.
To get an idea of how this works, picture my mug of coffee on a table with an area measuring one square pixel. That table-sized pixel would contain my mug, but it will also contain Steve’s omelette and Emiliano’s hashbrowns. Taken together as one pixel, that image wouldn’t yield very specific information about anyone’s breakfast.
“I can see this mug is here,” Steve said. “But if the whole table is just one pixel, and that pixel includes a lot of other stuff, it’s not going to give us a pure image of this mug.”
Dividing the table into smaller pixels means each pixel will contain a clearer, resolved image. If my mug of coffee alone spans nine pixels, say, at least the center pixel would contain pure “mug” information — some high resolution coffee.
Smaller pixels mean better image resolution, but they also mean that the plane has to fly closer to the ground and add more flight lines to its flight plan. If there are clouds on the horizon, longer time in the air could mean more exposure to bad weather.
“There are a couple of trade offs between how high you fly, and what pixel size you get,” Steve explained. “We want high resolution pixels, so you can’t have them fly too high, but also there’s a limit to how high the plane can fly, and the higher you are, the more the winds affect your flight path.”
No matter how well Dom plans, so much of the flyover’s success hinges on conditions beyond his control. But since it’s too early to know how the afternoon will play out, the plane will be taking off from Madison soon, and we will return to the field soon for a full day of sampling.
Even if the flyover is unsuccessful, Dom says the massive leaf collection effort will still generate valuable insights into the way groundwater moves through the forest.
“Regardless of the flight, the field campaign is highly valuable for our research in general,” Dom said. “Those data can be leveraged irrespective of the flyover.”