New Tree Tech: Real-time, long-term, high-tech reforestation monitoring

This story is the fourth article of a four-part Mongabay mini-series exploring the latest technological solutions to support reforestation. Read Part One, Part Two and Part Three.

“Plant a tree!” has over recent decades become a hackneyed ecological rallying cry for numerous mega-reforestation projects hyped in the media by celebrities, companies and governments.

But these projects have by and large achieved poor results, not due to a lack of enthusiasm by shovel-wielding, seed-planting volunteers, but because of what happens after the video cameras go away. Which is often nothing, or next to nothing.

Many people see reforestation as a quick fix to the climate emergency, but tree-planting projects often fail to put in place the long-term monitoring programs needed to track newly planted trees through their early stages of development, and to offer prompt targeted support when those trees show signs of struggling.

“A lot of those projects fail pretty early on for that reason,” said Cynthia Gerlein-Safdi, an ecohydrologist at the University of California, Berkeley.

Laborious measurements

One reason for all the failures: Monitoring an entire forest over a number of years is a huge undertaking, requiring laborious effort, money, and the amassing and analysis of big data sets.

Monitoring typically requires repeated measurements of the diameter of planted trees at a consistent height above the forest floor. These measurements are used to generate crude estimates of total height, the volume of the tree, and the amount of carbon stored in its tissues. This individual tree data — added together for the entire site — is crucial to not only determine whether a newly planted forest is thriving or failing, but also to accurately assign value to the ecosystem services it provides, such as carbon sequestration.

Measuring thousands of trees by hand is repetitive, time-consuming, expensive and can be dangerous if the restoration site is in an inaccessible area. It also tends to generate low-quality data, says Esthevan Gasparoto, CEO of Treevia, a forest-monitoring startup in Brazil.

The cost and time involved means that even when quality monitoring programs are in place, tree measurements are usually collected only once every five or 10 years, making it difficult to respond quickly if the forest is struggling.

“I think the number one issue is just the sheer size” of reforestation plots, says Gerlein-Safdi. “That’s a lot of time-consuming effort to cover these large areas” — a challenge that’s compounded if the reforested site is particularly difficult to access. “That’s where remote sensing can help,” she says.

Remote sensing is a term encompassing a group of technologies that, when combined, can gather detailed data without setting foot in a forest. It includes monitoring with satellites, drones, planes and forest sensors.

Monitoring from above

Satellites are now ubiquitous in Earth orbit, capturing imagery and surface characteristics almost constantly. Scientists have put these vast data sets to many uses, including monitoring forests — both natural and planted.

“The great thing about those techniques is that you can monitor a very large space,” Gerlein-Safdi explains. “They’re very powerful in that respect.”

They’re also not limited by the remoteness of a site, because satellites are constantly mapping and remapping the entire planet. “You’re going to get information wherever the planting happens,” she adds.

Satellites are fitted with a variety of monitoring equipment, so can provide different types and scales of information. Many collect optical data, which can offer a relatively simple measurement of forest growth. “It’s just like if you were taking a photo from the sky looking down at the Earth and seeing the forest being green,” explains Gerlein-Safdi.

Some satellites are equipped with more advanced sensors that provide more detailed information. Using multispectral or hyperspectral cameras, scientists can compare the intensity of light measured at different wavelengths and generate a vegetation index, which measures how green the forest truly is. This index can be compared with other forests, or the same forest at a previous time.

“If we see an upward trend of the forest becoming greener, then we can assume that this means most of the trees, at least, are healthy and thriving,” said Gerlein-Safdi.

Seeing photosynthesis and carbon storage from space

Hyperspectral cameras can also take advantage of light emitted by leaves as they photosynthesize, using a technique known as solar-induced chlorophyll fluorescence (SIF), to monitor forest productivity.

“When they’re doing photosynthesis, sometimes they get too much energy from the sun … and so they release photons in the near-infrared [wavelength], and we can see those photons from satellites,” explains Gerlein-Safdi.

SIF is particularly useful for monitoring reforestation projects in landscapes where the vegetation is less green. For example, Gerlein-Safdi’s research lab has used the technology in arid environments, where plants have fewer and smaller leaves. “You don’t see a huge green signal, but you can still see those photons being emitted by the leaves when they’re photosynthesizing,” she explained.

In 2018, a major advance came with the launch of the Global Ecosystem Dynamics Investigation (GEDI) program, which mounted lidar laser scanners on the International Space Station (ISS) to produce high-resolution 3D maps of forest canopies.

Lidar bounces laser light off the Earth’s surface and measures how long that light takes to return to the ISS. It can estimate the height of the trees across an entire forest, providing a much more detailed analysis of forest growth and carbon sequestration.

However, satellite monitoring isn’t a complete forest-monitoring solution. For starters, this kind of data doesn’t reveal much about the very earliest stages of tree growth. “At the very beginning, it’s very hard to know whether or not [tree planting] has been successful” using only satellite imagery, said Gerlein-Safdi.

For the first few months after planting, until saplings start to grow leaves, “they can look just like sticks.” But “that initial period is really crucial” to plant health, so alternative monitoring methods are needed.

In addition, different satellites collect data at different resolutions, ranging from each pixel covering an area as small as 30 by 30 meters (about 100 by 100 feet), all the way up to one pixel representing half a degree of longitude (about 55 kilometers, or 35 miles on a side). At these lower resolutions, “it’s not always easy to zoom into a small area and assess [a reforested site’s] health because the pixels are a little too big,” Gerlein-Safdi explains.

Even with high-resolution satellite imagery, the data generated averages multiple trees, giving an overall forest growth impression, but telling little about individual trees.

“Satellite data can provide the condition of [a] newly planted forest at the regional level, but it is difficult to focus on the individual trees from satellite images,” explains Feng Wang, a dryland ecologist at the Chinese Academy of Forestry in Beijing.

Drones give foresters a bird’s-eye view

To get more detailed data, the monitoring equipment needs to get closer. This can sometimes be achieved by flying light aircraft over a forest, but that can be expensive and requires fuel that contributes to global climate change — not ideal if your goal is to plant forests that will combat the problem.

Over the last decade, drone technology has advanced to the point that it’s suited to this task, bridging the gap between on-site measurements and distant satellites. “Compared with field measurements and satellite data, drones can give a detailed picture of every individual tree from newly planted forest at the landscape level,” Wang explained.

Just like satellites, drones can be kitted out with a range of sensor types that provide different information about a forest, including high-resolution optical cameras, multispectral cameras and lidar, allowing scientists to calculate vegetation indexes that quantify a forests’ greenness, estimate the height and volume of trees, and approximate carbon storage.

A major advantage of drones over satellites or planes is they can survey the forest from below the canopy, flying between trees to get more accurate measurements of tree density and size.

However, while drones are well-suited to surveying relatively small areas, it’s difficult for them to cover forests hundreds of square kilometers in area because the measuring devices have limited battery life and range. Drones also need human assistance, either from a pilot to control flight, or a supervisor who monitors autonomous drones as they navigate over or through a forest. These drone experts need to work from locations relatively close to the reforested area, making long-term drone monitoring challenging in the most inaccessible areas.

The offsetting pros and cons of satellite versus drone monitoring means both should be seen as complementary approaches. Drones allow scientists to “really get at the fine details of what is happening in the area,” explains Gerlein-Safdi, whereas satellites “give you more of an average” over a much larger area.

‘The internet of trees’

Moving yet closer, in-situ sensors can be used to directly monitor individual trees, relaying the data wirelessly to people who analyze the data from a safer and easier-to-access location. If many sensors are fitted to trees within a forest, they can form a distributed network that returns detailed data on the growth of the whole forest within huge reforestation plots.

This is technology that’s already in use: Brazil-based Treevia has developed sensor technology fit for the mammoth task of monitoring natural or planted trees over the long term. Weighing just 50 grams (less than 2 ounces) each and measuring around 4 centimeters (1.5 inches) tall, the sensors are small enough to be attached directly to saplings and track each plant’s growth over long periods of time.

“Once [the sensor] is installed, we are able to monitor the growth of the diameter of the tree,” says Treevia’s Gasparoto. The sensors also collect environmental measurements of temperature, humidity and rainfall. “We can collect a new data set every day,” he remarked proudly, which is “mind-blowing for people from the forestry sector.” This data, as well as helping foresters and conservationists understand the state of the forest, can also be used to predict the risk of wildfires and detect climate change impacts as they are happening.

Each sensor has an elastic strap that can be fitted to saplings as small as 1 cm (0.4 in) in diameter and that can stretch and stay on the tree until it reaches 1 meter (40 in) in diameter, allowing the sensor to “remotely monitor the forest growth over years,” Gasparoto explains. It can take years for the tree to outgrow the first elastic strap, but when it does, the team needs to return and install a new sensor with a larger strap. Still, this is a relatively small effort considering the richness of the data the devices generate.

Treevia has also developed a cloud-computing system called Smart Forest that receives the raw data transmitted by the sensors and processes it. Gasparoto calls Smart Forest the “internet of trees.” Using artificial intelligence algorithms known as machine learning, the system can turn terabytes of raw data into information useful for foresters to make real-time management decisions, such as when to apply fertilizer. It can also estimate the potential future growth of the forest — useful for valuing carbon sequestration.

The company has collected extensive measurements in the field to ground-truth its technology and ensure that the remote measurements are accurate and reliable. It has already deployed 4,000 of its sensors across nearly 200,000 hectares (roughly 494,000 acres) of forest plantation in Brazil and Uruguay, with ambitions to take its technology global in the coming years. That includes projects for the pulp and paper industry, as well as projects monitoring the Amazon and Cerrado biomes for carbon credits.

Remote-sensing technologies are an important tool in the reforestation toolkit, making large-scale, long-term monitoring of forests possible. Satellites, drones and in-forest sensor networks each have their strengths and weaknesses, but taken together, they offer an unparalleled view of the natural world. Combined with other data, they could bring the once-formidable task of forest monitoring within reach.

“Combining these different types of information, like remote sensing, climate data, lidar,” concludes Gasparoto, “that is the way to do the right forest measurement in the future.”

Banner image: Landsat views the forests from high above the Earth. Image courtesy of NASA.

Citations:

Gerlein-Safdi, C., Keppel-Aleks, G., Wang, F., Frolking, S., & Mauzerall, D. L. (2020). Satellite monitoring of natural reforestation efforts in China’s drylands. One Earth, 2(1), 98-108. doi:10.1016/j.oneear.2019.12.015

Buters, T., Belton, D., & Cross, A. (2019). Seed and seedling detection using unmanned aerial vehicles and automated image classification in the monitoring of ecological recovery. Drones, 3(3), 53. doi:10.3390/drones3030053

Li, X., Zhao, N., Jin, R., Liu, S., Sun, X., Wen, X., … Zhou, Y. (2019). Internet of Things to network smart devices for ecosystem monitoring. Science Bulletin, 64(17), 1234-1245. doi:10.1016/j.scib.2019.07.004

Tao, H., & Zhang, H. (2009, November). Forest monitoring application systems based on wireless sensor networks. In 2009 Third International Symposium on Intelligent Information Technology Application Workshops. IEEE. doi:10.1109/IITAW.2009.66

FEEDBACK: Use this form to send a message to the author of this post. If you want to post a public comment, you can do that at the bottom of the page.