Drones, Seeds, & Fires: How DroneSeed Plants Trees From The Sky, by Erika Clugston in Clean Technica, February 8th, 2019
Tree planters are superheros, according to DroneSeed founder and CEO, Grant Canary. In one day, a person manually replanting trees will burn the caloric equivalent of two marathons. Wow. It follows that such labor-intensive work is difficult to implement on a large scale, without the time or resources to match the intensity of forest fires that climate change is bringing about. Tree planters could use a little help.
DroneSeed is on a mission to ‘make reforestation scaleable.’ As CleanTechnica has reported, the Seattle-based startup uses drones, automation, and machine learning in its work to replant in post-fire environments, combat the spread of wildfires, and keep forests healthy.
It’s an exciting time for the company: just this past week, DroneSeed signed a contract with The Nature Conservancy (TNC), the world’s largest land conservation organization, to work on rangeland restoration. “In this project we’re furthering the science behind the aerial deployment of seeds to restore rangeland habitat,” Grant explains. “Carbon sequestration is not directly the aim of this project, habitat restoration and advancement of the aerial planting method is.” However, the company aims to do both. According to TNC’s own research, “Natural climate solutions are affordable, scalable and available right now. They can deliver a third of the solution to climate change needed by 2030. But are less than 1% of the conversation and receive less than 3% of climate funding.” Together with DroneSeed, they are working to make the technology and methodology scaleable to truly tackle these challenges.
We were curious to learn more about the company’s mission, obstacles and successes. In an in-depth conversation, DroneSeed founder and CEO Grant Canary shares his insights into the science behind reforestation, the ins and outs of working with drones, and the facts and figures as to why planting trees is a top priority.
What inspired you to start DroneSeed?
A desire to mitigate the worst effects of climate change. There’s a longer origin story, however, the reason we focused on scaleable solutions for revegetation is that trees and revegetation are the most effective method we have to sequester carbon. The Nature Conservancy published its strategy that we can achieve ⅓ of the global emissions reductions called for in The Paris Agreement with ‘Natural Climate Solutions’ which are replanting trees, rangelands, and fire response and restoration.
However, we don’t have a scalable method to plant and restore. The most sophisticated planting company in the world uses people with shovels because terrain and stumps make tractors useless in most places. DroneSeed is building a better system. We need a better system because we have a scale issue. The UN’s Food and Agriculture Organization (FAO) estimates that 300M acres (129M hectares) have been deforested since the 1990s. By our estimates, the largest forestry company in the world, which is five times larger than its next biggest competitor by acreage, plants around 1M acres a year (400-700M trees). Given current technology — people and shovels — at that rate it would take us 300 years to replant our deficit. That’s if you could find the people. That’s a challenge everywhere in the world because the work is so hard on the body. Treeplanters burn the caloric equivalent of running two marathons every day. They’re superheroes. Yet, we don’t have 300 years and that’s an incredibly inefficient and physically unsustainable way to plant trees. We’re building a solution to allow people to plant more, faster.
Why Climate Change versus any of the other humanity threatening problems out there?
My worldview is that climate change is the problem that all other problems report to. If we fail to mitigate climate change, there is no time left on the clock to work on any other problem. If climatic change is allowed to go to the worst extremes possible with inaction, no social, economic, or political system is going to fare well/survive. Historical precedents of what our future might be like include the Great Famine in China 1959-61 and the Dustbowl in the 1930s in the US. Or you can look to Hollywood to help wrap your head around what the future might look like — Mad Max is a super fun movie, but it doesn’t look fun to live in. Our mission at DroneSeed is to ‘Make Reforestation Scaleable.’ This mission and the opportunity to use technology to make a dent in an incredibly big problem is what I attribute to incredibly top notch talent joining our team.
How do you ensure success with a drone?
Listen first. We sought out, asked questions, and listened to wise old nursery managers and foresters. They are our favorites and have a ton of knowledge gained through decades of tree herding. We have been working with three of the five largest timber companies in the US for over three years and are now also working with the world’s largest land conservation organization. Before we built anything, they asked hard pragmatic questions of us, saved us years of development time with their insights, and today, continue to push us to be better and serve their needs more effectively.
As an example, in 2015, when we first started, we explained what we were planning. Our advisors were quick to point out that simply dropping seeds and seedlings from the sky had been tried in the 1970s. A drone doesn’t change that. When we started talking about using software to target and deploy seeds to ‘micro-sites’ our advisors finally thought we were on to something useful that advanced the science and practice. Micro-siting is planting a tree in a super localized spot where it will grow well. “Near a stump” is one example of a microsite. The stump provides shelter from sun and wind which keeps more moisture in the soil and the decaying old roots make it easy for new roots to penetrate. In 2016, our foresters and customers suggested we figure out how to control competitive vegetation by spraying. Like a garden, mitigating weeds is the number one way to boost seed survival. We wrote the software to do that in a very targeted way and are ever improving it. Lastly, we had a number of ideas about seed vessels. They gave us feedback as we developed four seed vessels for different eco-systems. Think of seed vessels as bringing the micro-site to the field. This is incredibly important to boosting survival rates as more and more post-burn sites have burned so hot and severely that the topsoil and seeds in it were torched. There’s no soil left to plant in and no seeds to naturally re-establish. In California, 80% of wildfires now burn this severely which hasn’t been the norm.
If we put all the pieces together, micro-siting software, targeted vegetation control, and seed vessels, we have a better response to boost survival rates of our plantings, and a better response for the incredibly massive scale of wildfires.
Traditionally manual planting has been most effective when replanting an area with seedlings (1-2 yr old trees grown in a nursery). Why use seeds?
Seedlings are great if tree harvests are planned and it’s known how many seedlings will be needed in two years. But fires aren’t predictable! There’s a two-year wait for nurseries to grow seedlings. It’s generally assumed you can ‘just order more trees’ but that’s not the case. There aren’t enough orchards and nurseries for the scale, and even if they were, they couldn’t survive the huge swings in demand. Its a large-scale supply chain problem that’s fascinating.
Here’s the math (and more):
7M acres burn each year in the US (ten-year rolling average). That’s 7 times what we estimate the largest timber company in the world plants per year. (Side note: the ten-year rolling average is up from 2.7M burned acres in 1992 which is scary as all f***). Now, it’s a ten-year average of 7M, so consider the swings in supply chain and implications for nurseries: some years we need 5M acres of seedlings, and in some years like 2017, we need 10M acres from nurseries. Wowzer. Nurseries in the supply chain just can’t take that risk and stay in business. At 200-500 trees per acre, we’re talking about 2.5 billion trees that may or may not be needed. Because the trees are small but growing fast, we can’t just hold them for a year as billions of 6-inch trees packed together will become billions of one to two foot tall trees which will kill each other as they compete for light and nutrients.
Can the government take this market risk and grow these seedlings for the public good? It could if it could bear the public wrath of throwing out 2.5Bn trees in some years. It also needs to fund this. Neither of these is likely as the US Forest Service is increasingly spending over 50% of its budget fighting fires. Seedlings can’t scale to this problem. We’ve got to use seeds.
So we use seeds and there are no more issues? Seeds are a limited resource as well. They have to be collected from the wild or cultivated in seed orchards (which take decades to grow).
Unfortunately, while you can just dump seeds everywhere by helicopter, the survival rate of 3-7% for many species means that there just aren’t enough seeds to scale to 7M acres because 93%-97% of the seeds are wasted.
How do we solve this problem then?!?! We have to boost survival rates and better use seeds so we can scale to 7M acres a year, and greater internationally. DroneSeed is doing that with seed vessels, where the seed is packaged and delivered into the ground with things like fertilizer and natural pest deterrents, so that we can spread the same quantity of seeds across more acres as we achieve better survival.
We’ve pioneered four vessels so far and we’re tailoring them to different eco-systems. We want to nix a feedback loop where insufficient seed/seedlings is resulting in a huge backlog of unplanted and unrestored areas. Forests aren’t naturally re-establishing due to fire severity and nature hates a vacuum, so invasives move in. This further reduces carbon sequestration and the invasives are more likely to burn, both of which accelerates climate change and increases fires and fire severity. Uggh. That’s a pretty nasty feedback loop.
We’re incredibly impressed with the foresters and contractors that have been trying to keep the problem in check with the tools and seed/seedlings they have. We want to give them better tools and higher survival rates. We’re listening to what their pain points are, their feedback, and making the most of their experience and suggestions. After every project DroneSeed does, we’re gaining more data to make improvements to the micro-siting software, the vessels themselves, and the aerial deployment. Wish us luck!
You buy off-the-shelf drones and then modify them to your needs. What kinds of modifications do you make and how extensive are they?
The modifications we make are so extensive, we wonder why we don’t make the drones ourselves. The fuselage is all that remains original (the carbon fiber or molded body) after our brilliant team has optimized it for our needs. However, we don’t want to be a drone manufacturer as they’ll become a commodity. Instead, we’ve built the technology to provide a pay-per-acre service. To best perform that service, we continually integrate the rapidly evolving innovations in hardware across the field to work with our intellectual property: proprietary software, seed vessels, and the aerial deployment mechanisms for our seed vessels.
The drones are automated and use machine learning models in the replanting process. Can you explain how specifically they are using machine learning and how does it impact the process?
The first step in our process is to go out and survey with LiDAR and multispectral cameras to create a 3D model of the terrain and obstacles to plan the many flights of our drone swarms.
The survey identifies the microsites where the trees will grow well and rule out gravel fields, rivers, mature forests etc. Zoom in and we can identify stumps, which we’ve already discussed, which are great to plant seeds next to. Zoom in further and we can use the near-infrared (NIR) to identify if the ground is covered in soil, wood chips, or several feet of branches called ‘slash’. so we can plant on the soil. Today, this is what we pay tree planters to do: be the intelligence that goes into the field and puts the shovel in soil next to the stump. The problem is the best treeplanters can only plant 2 acres a day (800-1000 trees).
Understanding what we’re looking for (gravel, stumps, NIR spots), we then employ ‘supervised classification,’ which is annotation by experts and validation by subsequent surveys, to determine the optimal planting routes and micro-sites in distinct areas. I would rather not go into the exact techniques we use currently, though they would be familiar to anyone versed in machine learning or spatial quantitative methods, but no one else (as far as we are aware) is collecting the data necessary to create an effective model for planting in forest and rangeland environments.
DroneSeed is currently the first and only company in the US that is FAA approved for heavy lift UAS operations with multiple drones simultaneously – congratulations! What was the process like to get this approval and how will it impact your success?
We honestly would have expected any number of corporations in aerospace or backing drone package delivery to have applied for and received this approval first. However, we’re leading this space and it tremendously affects the unit economics of our operations, so we couldn’t wait.
The approval allows us to operate aircraft up to 115 lbs in swarms of five. This gives us 57lbs of payload per drone. More payload means we can service more acreage in a single day. Previously we were limited by regulations to 16lbs of payload. This meant a lot more flights per acre even though our aircraft weren’t even remotely close to their maximum flight time. The increased weight approval allowed us to demonstrate to ourselves and investors how we would make money each day we went out to service acres — the unit economics. That’s a major milestone for a startup.
How do you determine what a burned area needs? Are there some areas that cannot be helped?
Everywhere can be helped, but it’s a survival rate probability game affected by the eco-system and burn severity that affects cost. We have a few principles that guide us:
- Ask the landowner the objectives. This is Forestry 101. Do they want a polyculture of many species for habitat (many diverse species of trees)? Do they want to harvest timber on a super long schedule of 80 years to fund acquisition of more conservation land? Do they want a commercial rotation of 25-40 years? (Side note: seeds significantly cut the supply-chain costs of polycultures)
- One-size doesn’t fit all. Different ecosystems require different vessels. We’ve developed four seed vessels for different forest and rangeland ecosystems with different benefits for arid versus wet climates. We also have different reproduction strategies for rare native species (for which there is low seed supply) versus commercially native species (where seed supply is better).
How many areas have you already replanted and where?
Our first planting project was in October 2018, replanting after the Grave Creek Fire which burned 7,000 acres near Medford, Oregon in 2018.
Over the last two years, we’ve worked with 3 of the 5 largest timber companies in the US spraying to protect trees. Scaleable methods to control competitive vegetation (spray) are vital to seed survival rate, so we spent considerable time figuring out how to do this with drones. We’re very excited about reducing costs for our customers with targeted spraying. We do this by using aircraft and software that enables us to spray just the 3-5 acres of Scotch Broom or Himalayan Blackberries in a 100-acre unit. As herbicides are a major cost center for all our customers they are also quite excited about methods to reduce their use. We keep getting invited to renew contracts and signing new ones, so that’s one indicator that we’re doing good work.
Have you seen any fruit from your labor, so-to-speak?
We’ll see results from our 2018 planting project after the snow melts! We’ll see the effects this spring of the 8 projects we did in 2018 spraying to protect trees.
You’ve raised nearly $5M in venture capital (VC) investment from Social Capital and Spero Ventures. Forestry isn’t a typical VC investment space, why for-profit versus not-for-profit?
Spot on. Forestry is not what we typically think of when we think about “software eating the world.” Our investors pattern us after SailDrone, Aclima, and UrbanFootprint. Each have large US government customers and each are working on incredibly valuable problems. These are major pain points for the customer which is trying to do more work with the same budget. An example is SailDrone acquiring oceanic data on acidification and other matters for a fraction of the cost of manned expeditions on boats, and their aquatic drones acquire the data without risk to people for years on end, which wasn’t possible or safe with manned crews. So far, three of the four of these companies have gone on to raise large much much larger rounds. The thesis is that a lot of the low hanging fruit in pure play software companies has been taken and that the next big companies (and IPOs) will come from companies that interact physically with the real world and in regulated markets. Think AirBnB, 23andMe, Lyft/Uber, Bird Scooters/Lime etc.
DroneSeed’s in that space and we’re expanding our offerings beyond the private sector to government entities such as the US Forest Service, Bureau of Land Management, and state agencies. In the immediate, our problem is a valuable one to solve for these agencies: how do we use technology to restore landscapes devastated by wildfires completely, rapidly, and at scale. Manned crews like this which are the current solution aren’t scaleable. Longer term, the problem is growing and becoming even more valuable to solve: We ask ourselves, what would we pay to avoid another dustbowl? What price would we pay to avoid a fishery industry collapse caused by the bleaching of coral reefs and fish spawning areas? Another Great Famine as China experienced between 1959 and 1961 which likely killed as many as all casualties of WWI? The answer should be we’d pay a LOT of money to avoid this. The science is showing in many scenarios we’ll be impacted within the next twenty years, aka within the lifespan of our student debt and mortgages.
This is where the VC model does fit. We need incredibly explosive growth of a technology to scale incredibly rapidly to mitigate a problem (climate change). That’s VC. Successful results will be incredibly valuable for society and the company. That’s VC. However, before anything can happen, large investments of capital are needed from investors with a vision to see what society will need/want before they know they want it. (Think if you really felt you needed an iPhone, or Google Search/Maps/Mail in 1990?). That’s VC. We’re already seeing public opinion shift on climate change as it did on marijuana legalization. Even Exxon believes we should participate in the Paris Climate Accords (1)(2). Right wing conservative groups featuring Trent Lott (Former Senate Republican Leader from Mississippi) and Janet Yellen (former Chairman of the Fed) advanced plans for a Carbon Dividend this year. These are indicators that in the US we’re starting to feel we need solutions. Lastly, one unspoken rule of VC investing: companies making a ton of money rapidly will immediately inspire many many many clones. From an environmental perspective, how awesome would it be to have ten companies competing with DroneSeed for contracts to reforest the planet? We’re happy to have a head start.
What do you see for the next few years of DroneSeed’s future? Expansion plans?
If we’re really going to fulfill the mission to ‘make reforestation scaleable’ by providing a tool to mitigate the worst effects of climate change, we’ve got to get beyond just the US. That means expanding to other countries. I’ve lived 10 years abroad and strongly believe that locals are always more efficient in their home countries than expats. So we’ll put the technology into the hands of locals while setting up either joint ventures, licensing, or royalty structures. We’ve been looking at Canada after the Fort McMurray Fire and already have a Special Flight Operations Certificate from their regulatory agency. We’re also looking at New Zealand where the billion tree program has allocated $245 Million NZD to fund tree planting projects. To enter New Zealand, we’ll need local Kiwis, government agencies, and philanthropic foundations to identify land for planting projects and contact us. We’ll help Te Uru Rākau, New Zealand’s Forestry Agency, get the technology into the hands of locals. Our technology will reduce the caloric cost to the field crews so the work is physically sustainable year after year.
More generally, there are five countries that have ⅔ of the world’s forests: The US, Canada, Brazil, China, and Russia. Additionally, in arid eco-systems, there are the rangelands projects, which sequester a ton of carbon underground in their roots. Between those two market sectors, there’s nothing but opportunity for DroneSeed to connect with partners and make serious headway on sequestering carbon.