Despite thorough preparations, flying the drone is still nervewracking.
Dr. Manoela Machado, a Research Scientist at Woodwell Climate, has double- and triple- checked her calculated flight path over a study plot in the Cerrado, Brazil’s natural savanna. The drone can essentially fly itself, and she’ll be monitoring its speed, altitude, and battery life from her handheld controller on the ground, but many things could still go wrong. High winds, an unforeseen obstruction, loss of connectivity— all could jeopardize the mission, potentially dropping the expensive equipment 40 meters into the woodland canopy below.
Aboard Machado’s drone sits a powerful piece of technology – a LiDAR sensor. Developed originally for use in meteorology, this remote sensing technique now has widespread applications across scientific fields, from archaeology, to urban planning, to climate science. At Woodwell Climate, Machado and others employ LiDAR to create detailed three dimensional models of landscapes, which provide valuable insight into the structure of ecosystems and the amount of carbon stored in them— all with just a few (million) pulses of light.
What is LiDAR?
LiDAR stands for Light Detection and Ranging. Put simply, it is a sensor that uses laser light to measure distance.
Similar to other technologies like sonar and radar, which use sound and radio waves, respectively, LiDAR is an example of an “active” sensor. “Passive” sensors like cameras collect ambient light, while LiDAR actively pings the environment with beams of laser light and records the time those beams take to bounce back. The longer the return time, the further away an object is. That distance measurement is then used to calculate the precise location in three-dimensional space for each reflection.
This process is repeated millions of times during a single scan, resulting in a dense cloud of point locations. With some advanced computing, the data can be assembled into a 3D picture of the landscape.
“It’s effectively three dimensional pointillism,” says Woodwell Climate Chief Scientific Officer, Dr. Wayne Walker, who has been using LiDAR in his studies for 25 years.
Far more detailed than an oil painting however, a LiDAR model can reconstruct nearly every leaf, twig, and anthill on a landscape.
“Once you construct that cloud of millions of points, you get to walk inside the forest again,” says Machado. “When you finish processing the data and see the cloud you go, ‘I remember that tree! I remember standing there!’ It’s mesmerizing.”
For a project like Machado’s, scanning a few dozen hectares, the sensor is usually placed on a drone. Larger study areas require sensors mounted on low-flying airplanes or even satellites, but for small ground-based applications there are sensors one can carry, mount on a tripod, or attach to a backpack. Some newer phone models even have LIDAR apps built in. Regardless of how LIDAR is deployed, it remains a straightforward method of data collection. Just point the sensor at what you want to scan and within minutes, you’ve captured the data for a detailed three-dimensional model of your area of interest.
Estimating the weight of a forest
What Machado and Walker are often after from a LiDAR scan is a measurement of biomass, or the total weight of the organic matter present in an ecosystem. Plants store carbon in the form of organic matter, so biomass measurements are an easy way to estimate an area’s carbon storage.
However, measuring a forest’s biomass directly would require cutting down all the trees, drying them out, and weighing what’s left — impractical and needlessly destructive— so scientists use proxy measurements. Walker likens the process to trying to estimate the weight of a human without access to a scale.
“What are the measurements you might use if you couldn’t actually physically measure weight? You might record height, waist size, inseam, and if you obtain enough of these measurements you can start to build a model that relates them to weight,” says Walker. “That’s what we’re trying to do when we estimate the biomass of an entire forest.”
Raw LiDAR data is only a measurement of distance, but by classifying each point based on its location relative to the cloud, researchers are able to extract the proxy measurements needed to model biomass across the ecosystem. Before LiDAR, these proxy measurements— things like trunk diameter, height, and tree species— had to be recorded entirely by hand, which limits data collection based on human time and resources. The time-saving addition of LiDAR has vastly expanded the possible scale of study plots. While field measurements are still essential to calibrate models, LiDAR is one of the only technologies that can give scientists enough detail and scope to assess carbon stocks over entire ecosystems.
“There is no other kind of sensor that even comes close to LiDAR,” says Walker.
The power and potential of LiDAR
At Woodwell Climate, researchers have employed the power of LiDAR to map biomass and carbon from Brazilian forests, to the Arctic tundra. Outside of the Center, the technology has found applications in archaeological surveys, lane detection for self-driving cars, and topographical mapping down to a resolution of half a meter.
But the detail that makes LiDAR so powerful can also make the data a challenge to work with. A single scan produces millions of data points. According to Geospatial Analyst and Research Associate, Emily Sturdivant, who analyzed LiDAR data for Woodwell’s Climate Smart Martha’s Vineyard project, that wealth of data often overwhelms our ability to extract the full potential of information available in one point cloud.
“LiDAR creates so much data that when you look at it, it’s hard not to be blown away imagining all the different things you could do with it. But then reality kicks in,” says Sturdivant. “It’s challenging to take full advantage of all those points with our current processing power. It’s a matter of the analysis technology catching up with the data.”
Processing LiDAR data requires large amounts of computing time and storage space, especially when performing more complex analyses like segmenting the data on the scale of individual trees. As machine learning and cloud computing technologies advance however, this becomes less of an obstacle, and the potential insights from LiDAR datasets will advance along with them.
LiDAR can be an expensive endeavor, too. Drones with the right equipment can cost tens of thousands of dollars, as can hiring a plane and pilot and paying for jet fuel, so data sharing has been important in making the method more cost effective. U.S. government agencies like NASA and the USGS have facilitated the collection of LiDAR data through satellites and plane flights, making the data available for public use. Woodwell Climate research has benefitted from these public datasets, using them to inform landscape studies and carbon flux models.
According to Sturdivant, the reliable production of public data has been greatly beneficial to advancing LiDAR-based studies, though it now faces risks from federal cuts to science agency funding.
“One of the greatest advantages of having publicly supported data is the consistency, but that’s exactly what’s now at risk,” says Sturdivant. “Public accessibility has been so important in allowing new scientists to learn and experiment and then help everyone else learn.”
Each new LiDAR scan represents a trove of information that could be used to better understand our changing planet, making it critical to continue supporting and collecting LiDAR data. Its intensely visual and highly detailed nature has made it one of the most powerful tools we have for understanding something as complex as a forest.
“And on top of that,” says Machado “It’s just visually beautiful.”
Under the thick forest of Mexico’s Yucatán Peninsula, the ancient ruins of a Maya City have been uncovered with the use of remote sensing.
Of course, that wasn’t the outcome that Woodwell Climate’s Chief Scientific Officer, Dr. Wayne Walker, anticipated when he and his team collected and processed the remote sensing dataset for an unrelated project nearly a decade ago.
Walker’s team was mapping the region as part of the Mexico REDD+ project, a collaborative, international effort to explore strategies for reducing emissions from deforestation and degradation in the country. Using a remote sensing technology called LiDAR, which scans terrain from a low-flying plane using pulses of laser light, Walker and project collaborators created a comprehensive map of forests—and the carbon they contain—across Mexico.
Walker and team coordinated the flights and processed the raw data for use in the project, uploading it afterwards to a website for public use. But, once the project ended, he all but forgot about the effort, apart from responding occasionally to researchers interested in downloading the dataset for their own work.
One of those researchers was Luke Auld-Thomas, a PhD candidate at Tulane University researching the Classic Maya civilization, which thrived in the Yucatan until the 9th century when much of the region was abandoned, though their culture and languages persist to this day. Because of its unique ability to provide a detailed three-dimensional picture of whatever features are present on the ground, LiDAR imagery is an incredibly powerful tool for a multitude of purposes, from climate science to archaeology. And while the Mexico REDD+ project was interested in documenting the forests, Auld-Thomas was interested in what might be hidden beneath them.
“One scientist’s noise is another’s entire field of study,” says Walker. “In our other projects, like Climate Smart Martha’s Vineyard, we see historical structures like stone walls that aren’t necessarily meaningful to our work but could be of interest to archaeologists.”
In Mexico, the large areas surveyed by Woodwell Climate revealed not just individual human-built structures, but the plazas, reservoirs, and ball courts of an entire, previously undocumented city. The discovery, published in the journal Antiquity, supported the theory that the region was, in fact, densely settled during the height of Classic Maya civilization.
“We knew that it was close to a lot of interesting sites and settlements— areas of large-scale landscape modification that had been mapped and studied— but none of the survey areas themselves were actually places that archeologists ever worked, making it a really exciting sample to work with,” said Auld-Thomas.
Auld-Thomas had specifically been on the hunt for a pre-existing LiDAR dataset like the one Walker helped create— a survey conducted for completely non-archaeological purposes and therefore free of any biases. Essentially a “random sample” of the region. That randomness, and the subsequent discovery of an entire city, allowed Auld-Thomas and his colleagues to more strongly argue their point about intense urbanization in the Yucatán.
“If you’re only going to places where you know there’s going to be something, then of course, you’re going to find something significant, right? These random samples, not collected for archeological purposes, are gold in some respects,” said Dr. Marcello Canuto, who co-authored the paper. Canuto directs the Middle American Research Institute at Tulane, where the research for this study was conducted.
The unexpected outcome of the LiDAR survey offers a textbook example of the value of open data access. Sharing data and resources both within and between fields of science can jumpstart discovery and distribute the costs of an otherwise expensive data collection effort.
“Just look at what came out of the moonshot,” says Canuto. Thousands of technologies, developed in humanity’s pursuit of the moon landing, have found unforeseen applications in today’s world— including LiDAR.
“Certainly, many of us have produced datasets that have led to incremental advances in closely related fields,” says Walker. “But here is a special case of open source data advancing discovery in an entirely unrelated field of study.”
Advancements across fields continue to better our understanding of the world around us. And the lessons learned from a civilization like the Maya have very real parallels to today’s climate crisis.
As Auld-Thomas and Canuto show, the Maya densely settled the Yucatán Peninsula, maxing out the capacity of the surrounding environment to support their population. And then the regional climate shifted. A long-term drought settled in, resources became scarcer, governments became unstable, people started leaving the cities, and the infrastructure of the larger civilization collapsed.
“The reason environmental scientists collect LiDAR data of the forest, is that they are trying to understand environmental processes in order to help human societies conserve the landscape,” says Auld-Thomas. “As archaeologists, we try to understand how people in these exact environmental contexts have confronted deforestation and climate change and all of these other things before.”
For Canuto, the lesson to be learned lies not just in the environmental perils, but in the societal ones. Because what complex societies hate— be they the Classic Maya or today’s modern culture— is a lack of predictability. If a system cannot adapt, it will fail.
“The collapse was more than just climate change,” says Canuto. “It was a failure of a political system to respond to climate change.”
“It’s been around a long time, actually,” muses Senior Scientist, Dr. Jennifer Francis. “It’s gotten more sophisticated, sure, and a lot of the applications are new. But the concept of artificial intelligence is not.”
Dr. Francis has been working with it for almost two decades, in fact. Although, back when she started working with a research tool called “neural networks,” they were less widely known in climate science and weren’t generally referred to as artificial intelligence.
But recently, AI seems to have come suddenly out of the woodwork, infusing nearly every field of research, analysis, and communication. Climate science is no exception. From mapping thawing Arctic tundra, to tracking atmospheric variation, and even transcribing audio interviews into text for use in this story, AI in varying forms is woven into the framework of how Woodwell Climate creates new knowledge.
AI helps climate scientists track trends and patterns
The umbrella term of artificial intelligence encompasses a diverse set of tools that can be trained to do tasks as diverse as imitating human language (à la ChatGPT), playing chess, categorizing images, solving puzzles, and even restoring damaged ancient texts.
Dr. Francis uses AI to study variations in atmospheric conditions, most recently weather whiplash events— when one stable weather pattern suddenly snaps to a very different one (think months-long drought in the west disrupted by torrential rain). Her particular method is called self-organizing maps which, as the name suggests, automatically generates a matrix of maps showing atmospheric data organized so Dr. Francis can detect these sudden snapping patterns.
“This method is perfect for what we’re looking for because it removes the human biases. We can feed it daily maps of, say, what the jetstream looks like, and then the neural network finds characteristic patterns and tells us exactly which days the atmosphere is similar to each pattern. There are no assumptions,” says Dr. Francis.
This aptitude for pattern recognition is a core function of many types of neural networks. In the Arctic program, AI is used to churn through thousands of satellite images to detect patterns that indicate specific features in the landscape using a technique originally honed for use in the medical industry to read CT scan images.
Data science specialist, Dr. Yili Yang, uses AI models trained to identify features called retrogressive thaw slumps (RTS) in permafrost-rich regions of the Arctic. Thaw slumps form in response to subsiding permafrost and can be indicators of greater thawing on the landscape, but they are hard to identify in images.
“Finding one RTS is like finding a single building in a city,” Dr. Yang says. It’s time consuming, and it really helps if you already know what you’re looking for. Their trained neural network can pick the features out of high-resolution satellite imagery with fairly high accuracy.
Research Assistant Andrew Mullen uses a similar tool to find and map millions of small water bodies across the Arctic. A neural network generated a dataset of these lakes and ponds so that Mullen and other researchers could track seasonal changes in their area.
And there are opportunities to use AI not just for the data creation side of research, but trend analysis as well. Associate Scientist Dr. Anna Liljedahl leads the Permafrost Discovery Gateway project which used neural networks to create a pan-Arctic map of ice wedge polygons—another feature that indicates ice-rich permafrost in the ground below and, if altered over time, could suggest permafrost thaw.
“Our future goals for the Gateway would utilize new AI models to identify trends or patterns or relationships between ice wedge polygons and elevation, soil or climate data,” says Dr. Liljedahl.
How do neural networks work?
The projects above are examples of neural-network-based AI. But how do they actually work?
The comparison to human brains is apt. The networks are composed of interconnected, mathematical components called “neurons.” Also like a brain, the system is a web of billions upon billions of these neurons. Each neuron carries a fragment of information into the next, and the way those neurons are organized determines the kind of tasks the model can be trained to do.
“How AI models are built is based on a really simple structure—but a ton of these really simple structures stacked on top of each other. This makes them complex and highly capable of accomplishing different tasks,” says Mullen.
In order to accomplish these highly specific tasks, the model has to be trained. Training involves feeding the AI input data, and then telling it what the correct output should look like. The process is called supervised learning, and it’s functionally similar to teaching a student by showing it the correct answers to the quiz ahead of time, then testing them, and repeating this cycle over and over until they can reliably ace each test.
In the case of Dr. Yang’s work, the model was trained using input satellite images of the Arctic tundra with known retrogressive thaw slump features. The model outputs possible thaw slumps which are then compared to the RTS labels hand-drawn by Research Assistant Tiffany Windholz. It then assesses the similarity between the prediction and the true slump, and automatically adjusts its billions of neurons to improve the similarity. Do this a thousand times and the internal structure of the AI starts to learn what to look for in an image. Sharp change in elevation? Destroyed vegetation and no pond? Right geometry? That’s a potential thaw slump.
Just as it would be impossible to pull out any single neuron from a human brain and determine its function, the complexity of a neural network makes the internal workings of AI difficult to detail—Mullen calls it a “black box”—but with a large enough training set you can refine the output without ever having to worry about the internal workings of the machine.
Speeding up and scaling up
Despite its reputation in pop culture, and the uncannily human way these algorithms can learn, AI models are not replacing human researchers. In their present form, neural networks aren’t capable of constructing novel ideas from the information they receive—a defining characteristic of human intelligence. The information that comes out of them is limited by the information they were trained on, in both scope and accuracy.
But once a model is trained with enough accurate data, it can perform in seconds a task that might take a human half an hour. Multiply that across a dataset of 10,000 individual images and it can condense months of image processing into a few hours. And that’s where neural networks become crucial for climate research.
“They’re able to do that tedious, somewhat simple work really fast,” Mullen says. “Which allows us to do more science and focus on the bigger picture.”
Dr. Francis adds, “they can also elucidate patterns and connections that humans can’t see by gazing at thousands of maps or images.”
Another superpower of these AI models is their capability for generalization. Train a model to recognize ponds or ice wedges or thaw slumps with enough representative images and you can use it to identify the water bodies across the Arctic—even in places that would be hard to reach for field data collection.
All these qualities dramatically speed up the pace of research, which is critical as the pace of climate change itself accelerates. The faster scientists can analyze and understand changes in our environment, the better we’ll be able to predict, adapt to, and maybe lessen the impacts to come.