NASA supercomputing study breaks ground for tree mapping, carbon research


Research News


Scientists from NASA’s Goddard Space Flight Center in Greenbelt, Maryland, and international collaborators demonstrated a new method for mapping the location and size of trees growing outside of forests, discovering billions of trees in arid and semi-arid regions and laying the groundwork for more accurate global measurement of carbon storage on land.

Using powerful supercomputers and machine learning algorithms, the team mapped the crown diameter – the width of a tree when viewed from above – of more than 1.8 billion trees across an area of more than 500,000 square miles, or 1,300,000 square kilometers. The team mapped how tree crown diameter, coverage, and density varied depending on rainfall and land use.

Mapping non-forest trees at this level of detail would take months or years with traditional analysis methods, the team said, compared to a few weeks for this study. The use of very high-resolution imagery and powerful artificial intelligence represents a technology breakthrough for mapping and measuring these trees. This study is intended to be the first in a series of papers whose goal is not only to map non-forest trees across a wide area, but also to calculate how much carbon they store – vital information for understanding the Earth’s carbon cycle and how it is changing over time.

Measuring carbon in trees

Carbon is one of the primary building blocks for all life on Earth, and this element circulates among the land, atmosphere, and oceans via the carbon cycle. Some natural processes and human activities release carbon into the atmosphere, while other processes draw it out of the atmosphere and store it on land or in the ocean. Trees and other green vegetation are carbon “sinks,” meaning they use carbon for growth and store it out of the atmosphere in their trunks, branches, leaves and roots. Human activities, like burning trees and fossil fuels or clearing forested land, release carbon into the atmosphere as carbon dioxide, and rising concentrations of atmospheric carbon dioxide are a main cause of climate change.

Conservation experts working to mitigate climate change and other environmental threats have targeted deforestation for years, but these efforts do not always include trees that grow outside forests, said Compton Tucker, senior biospheric scientist in the Earth Sciences Division at NASA Goddard. Not only could these trees be significant carbon sinks, but they also contribute to the ecosystems and economies of nearby human, animal and plant populations. However, many current methods for studying trees’ carbon content only include forests, not trees that grow individually or in small clusters.

Tucker and his NASA colleagues, together with an international team, used commercial satellite images from DigitalGlobe, which were high-resolution enough to spot individual trees and measure their crown size. The images came from the commercial QuickBird-2, GeoEye-1, WorldView-2, and WorldView-3 satellites. The team focused on the dryland regions – areas that receive less precipitation than what evaporates from plants each year – including the arid south side of the Sahara Desert, that stretches through the semi-arid Sahel Zone and into the humid sub-tropics of West Africa. By studying a variety of landscapes from few trees to nearly forested conditions, the team trained their computing algorithms to recognize trees across diverse terrain types, from deserts in the north to tree savannas in the south.

Learning on the job

The team ran a powerful computing algorithm called a fully convolutional neural network (“deep learning”) on the University of Illinois’ Blue Waters, one of the world’s fastest supercomputers. The team trained the model by manually marking nearly 90,000 individual trees across a variety of terrain, then allowing it to “learn” which shapes and shadows indicated the presence of trees.

The process of coding the training data took more than a year, said Martin Brandt, an assistant professor of geography at the University of Copenhagen and the study’s lead author. Brandt marked all 89,899 trees by himself and helped supervise training and running the model. Ankit Kariryaa of the University of Bremen led the development of the deep learning computer processing.

“In one kilometer of terrain, say it’s a desert, many times there are no trees, but the program wants to find a tree,” Brandt said. “It will find a stone, and think it’s a tree. Further south, it will find houses that look like trees. It sounds easy, you’d think – there’s a tree, why shouldn’t the model know it’s a tree? But the challenges come with this level of detail. The more detail there is, the more challenges come.”

Establishing an accurate count of trees in this area provides vital information for researchers, policymakers and conservationists. Additionally, measuring how tree size and density vary by rainfall – with wetter and more populated regions supporting more and larger trees – provides important data for on-the-ground conservation efforts.

“There are important ecological processes, not only inside, but outside forests too,” said Jesse Meyer, a programmer at NASA Goddard who led the processing on Blue Waters. “For preservation, restoration, climate change, and other purposes, data like these are very important to establish a baseline. In a year or two or ten, the study could be repeated with new data and compared to data from today, to see if efforts to revitalize and reduce deforestation are effective or not. It has quite practical implications.”

After gauging the program’s accuracy by comparing it to both manually coded data and field data from the region, the team ran the program across the full study area. The neural network identified more than 1.8 billion trees – surprising numbers for a region often assumed to support little vegetation, said Meyer and Tucker.

“Future papers in the series will build on the foundation of counting trees, extend the areas studied, and look ways to calculate their carbon content,” said Tucker. NASA missions like the Global Ecosystem Dynamics Investigation mission, or GEDI, and ICESat-2, or the Ice, Cloud, and Land Elevation Satellite-2, are already collecting data that will be used to measure the height and biomass of forests. In the future, combining these data sources with the power of artificial intelligence could open up new research possibilities.

“Our objective is to see how much carbon is in isolated trees in the vast arid and semi-arid portions of the world,” Tucker said. “Then we need to understand the mechanism which drives carbon storage in arid and semi-arid areas. Perhaps this information can be utilized to store more carbon in vegetation by taking more carbon dioxide out of the atmosphere.”

“From a carbon cycle perspective, these dry areas are not well mapped, in terms of what density of trees and carbon is there,” Brandt said. “It’s a white area on maps. These dry areas are basically masked out. This is because normal satellites just don’t see the trees – they see a forest, but if the tree is isolated, they can’t see it. Now we’re on the way to filling these white spots on the maps. And that’s quite exciting.”


From EurekAlert!

31 thoughts on “NASA supercomputing study breaks ground for tree mapping, carbon research

  1. So.. the science of the carbon cycle.. was never “settled”

    And yet they say the CO2 in the atmospheric part of the carbon cycle controls climate..

    OOPS !!

  2. As I understand, one of the beneficial effects of increases in atmospheric CO2 is an increase in the growth of C3 plants, which represent the majority of plant types on the planet. However, this increase in plant growth is substantially greater in water-stressed conditions such as deserts and arid regions, because the increased CO2 reduces the size of the leaves’ stomata (or pores) and reduces the amount of evaporation, leaving more water for plant growth.

    • It was my understanding that regardless of size, the stomata are able to close sooner, having obtained a higher concentration of CO2, thereby reducing water lost by evaporation compared to those that remain open longer.

  3. Computer Aided Self Pleasuring – there’s gotta be a word for that.

    “rising concentrations of atmospheric carbon dioxide are a main cause of climate change.”
    The disappearance of the trees is the main cause of climate change

    Quote:”store it out of the atmosphere in their trunks, branches, leaves and roots”
    Wrong again
    The ‘carbon’ is ‘stored in the dirt/soil under the trees
    Work it out:
    Imagine (coz it don’t exist any more) some soil under A Forest (again= another good imagination is requires here also)
    1) Consider the ‘top soil’ (The “A Horizon” if you wanna be scientific and learn more)
    2) The A Horizon can be or should be anything from 60cm to 300cm deep
    3) Take it to be 50% by volume “Organic matter” = dead leaves, twigs, branches &trees.
    4) Give the organic fraction a density of 660kg per cubic metre
    6) Allow 35 cubic feet to 1 cubic metre and thus 12 x 35 ‘board feet’ per cubic metre
    7) Allow your ‘boards’ to have the same density as the ‘organics
    … how much wood (equivalent) have you got?

    The very fact that they can count individual trees indicated a trashed landscape – trashed by our ancient forebears in their pursuit of food.
    There is NOTHING for critters like us in A Forest, any forest
    And when they finished trashing it (take note California – far too late for Australia) via their imagined to be ‘sophisticated fire management practices (who ARE you kidding) – they *had* to move out.
    Landing in frozen he11-holes likes Canadia, Siberia, most of northern Europe, New England and …..(you fill in the spaces)

    How NASA, the self important self promoting bunch of muppets (2nd to Boris Johnson’s Government) is simply indicative of what a hideous mess we are in.
    Proven by the fact that a 20 strong bunch of 14, 15 & 16 years old attempted to trash my car last night, while I and my Autistic 8 year (girl) friend were inside it as we attempted to leave the car park at children’s playground she (used to) love going to.
    One the day her grandfather was buried and the day after her 8th birthday

    Thanks Boris, Michael, Naomi, Maurice, IPCC, Ben etc etc etc
    Thanks for pretty completely trashing 1 (maybe 2) generation(s) of kids.

    Thank you sooooo much

  4. This program will certainly help wood burning power station owners identify new areas to target for cutting

    • It is always nice to know more about one’s planet, but tracking the stored carbon down to each and every tree is getting stupid, particularly when CO2 is not a pollutant, CO2 is plant food, and CO2 is NOT warming the planet. The Japanese started serious forest management centuries ago, keeping track of every tree in their very limited forests. You cannot do that for a world. A waste of time, effort, and treasure.

  5. Test location in Australia around 31 deg S, 130 deg E.
    Test – what is the name of the nearest town? Geoff S.

  6. I’m almost certain that NASA understands satellite digital imaging data very well, however, the story does not show them utilizing the available power of various satellites. QuickBird pixel size is 60 cm, so a “tree” measuring three (3) meters in width (limbs and foliage) is about nine (9) pixels. QuickBird multi-spectral band 2 is the color green. Also, chlorophyll reflectance spectra is notorious for flooding near infrared bands and can be detected by any number of imaging satellites. Why NASA went through all of these very expensive super computer one year training cycle efforts is almost beyond me. Follow the money?

    • They used pansharpenend NVDI

      NIR-Red / NIR +Red = NVDI

      ‘Why NASA went through all of these very expensive super computer one year training cycle efforts is almost beyond me. ”

      because tree detection is not a simple problem

      • Yes and how much carbon is in any detected tree has a whole other range of errors. So whatever number you come up with is just a stupid massive range. How do you propose we verify the data … one idea comes to mind lets burn every tree down 🙂

        It’s meaningless junk. We call it #MeScience because it’s lazy science by Me Generation when you want answers now but aren’t prepared to do the actual science work.

        A real scientist would do is find a couple of large area of trees and actually count them . Then work on being to able to detect the correct number of trees in the areas from the space images. Thus you can verify and quantify the tree detection error.

      • NVDI is “normalized difference vegetation Index”, and is a mathematical treatment of data from the Sentinel-2 satellite, which has a 10 meter pixel size. We mineral exploration geologists struggle to get rid of tree detection because the reflectance of chlorophyll in the near-infrared band masks important reflectance bands, like jarosite, which has a reflectance peak near 750 nm. I could find trees, especially in an arid environment, quite easily. Show me the money?

  7. If there are more trees than was previously thought then the CO2 problem is far worse than anyone feared as they can only grow when theres an abundance of CO2.

    If there are less trees than was previously thought then the CO2 problem is far worse than anyone feared as the atmosphere is in such bad shape that not even trees can grow.

  8. “It will find a stone, and think it’s a tree. Further south, it will find houses that look like trees.”
    By an amazing coincidence, this is an apt analogy for what climate “scientists” often do, thinking stones or houses are trees. When all you have is a hammer…

    • Turns out the rock is made of calcium carbonate and contains more carbon than trees depending on size.

      • Around Basalt Colorado, one finds lots of basalt coated with a white substance on the side facing the sky, even if covered in dirt. This can reach a half inch thick or more, and is probably calcium carbonate. Who is counting this carbon? All this CO2 mongering must be having an effect, I haven’t heard that ‘diamonds are a girl’s best friend’ in some time.

  9. Great, one of the main organizations who has pushed butchered data, and still does, and makes evidence-free claims of doom…

    I am sure this will be all above board and not used to push the narrative.

    As far as Climate goes, NASA’s reputation is in tatters

  10. So follow the logic you use sonic tomography to cross section and calculate the amount of carbon in a tree with error range in 20-40%. Only a couple of tree species have been done so you average the result and assume it’s the same for all species which now has an unknown error range.

    Then you get image entire areas and use a super computer and AI to pick trees. Then you use that number of trees and your error loaded carbon estimate per tree and what do you get?

    Answer: An unverifiable number with massive error range that is about as useful as a visa card at a cash only sale.

  11. Come back in a year, or two, or ten…

    How fast do they think the vegetation in dry climates changes? The only rapid changes in such territory are people caused. Over a year, or two, or ten, the tree isn’t going to change at all on the pixel scale in effect, and so any “measured” changes in storage are swamped by uncertainties in measurements.

  12. Flight center mapping trees ?
    Next, forest state departments here and there might start to fly things around. Now I’m concerned and so shall we all.

  13. From the above article:
    “Scientists from NASA’s Goddard Space Flight Center in Greenbelt, Maryland, and international collaborators demonstrated a new method for mapping the location and size of trees growing outside of forests, discovering billions of trees in arid and semi-arid regions. . . Tucker and his NASA colleagues, together with an international team, used commercial satellite images from DigitalGlobe . . .”

    Hmmmm . . . one can only in stand in wonderment as to why this team of scientists did not first make use of the medium and high resolution multispectral imagery from GOVERNMENT satellites built specifically for the purpose of mapping resources (including vegetation) on Earth’s surface. You know, satellites such NASA’s Landsats, NOAA/USGS’s Advanced Very High-Resolution Radiometer (AVHRR) sensors flying aboard sixteen satellites, NASA’s 36-band Moderate Resolution Imaging Spectroradiometer (MODIS) flying onboard the NASA Aqua satellite, and JAPAN’S Advanced Spaceborne Thermal Emission and Reflectance Radiometer (ASTER) flying on the NASA Terra satellite (refs: and )

    All of the above-mentioned multispectral imagining spacecraft/sensors are noted for their ability to perform mapping of ground vegetation because reflectance values in the visible red and near-infrared bands are used to derive a Normalized Difference Vegetation Index (NDVI), which is a widely used measure of photosynthetic activity.

    And it would be helpful, given the above facts, to see hard evidence that the COMMERCIAL imagery provided new data leading to the discovery of “billions of trees in arid and semi-arid regions”.

    What was the real reason to use imagery from the commercial DigitalGlobe satellites?

  14. “MAPPING THE LOCATION AND SIZE OF TREES GROWING OUTSIDE OF FORESTS,” does not tell you much about forest does it? Oh by the way most so called tress in the AZ desert are really bushes, do they know that? The “trees” on my yard a really bushes, can their satellite image tell the difference? I think not!

  15. All very interesting.
    Maybe I’ll call it powerful.
    This fits with “powerful” supercomputers, and
    “powerful” artificial intelligence, and a
    “powerful” computing algorithm.
    None of these uses makes much sense, in contrast to something such as the Bruneau-Jarbidge event of the Middle Miocene.

    I also like
    “Now we’re on the way to filling these white spots on the maps.”
    Hic sunt dracones

  16. This is a beautiful example that something that is absurdly simple for the Mk I Eyeball is well-nigh impossible for a supercomputer.

    A couple of billion years of not being eaten by predators you didn’t see in time, not starving because of missing potental prey and not failing to procreate because you didn’t notice a suitable male/female makes for DEEPER learning.

Comments are closed.