Month: December 2021

Yesterday, a massive wildfire suddenly erupted on the Colorado grassland and destroyed hundreds of structures. It capped a year when wildfires burned 7.13 million acres of land in the U.S., which is actually slightly below the average of 7.47 million acres. Recent years have been much worse, including 2015 when a record-breaking 10.1 million acres burned, and 2018 when the figure was 8.8 million acres, and the “Camp Fire” in northern California killed 85 people.

It’s widely accepted among Americans that global warming is making wildfires bigger and more frequent, and will continue doing so as time passes, and that wildfire damage to infrastructure is unpreventable. To the first point, while it’s true anthropogenic global warming is making fires in many parts of the world worse, its impact is much smaller than the news media has led people to believe. A vastly greater contributor to the problem is human suppression of all forest fires, which allows dead vegetation, underbrush, and dead trees to build up to unnaturally high levels, laying the groundwork for inevitable mega-fires that can’t be controlled.

For example, before white settlement of what is now California, 4.5 million acres of the state’s forests usually burned each year thanks to lightning strikes and Native American land clearing. During 2018, a year that modern people consider catastrophic for California wildfires, only 1.9 million acres of the state burned. Many decades of fire suppression have resulted in the state’s forests having up to 100 times as much woody biomass as they should. Other parts of the U.S. that are prone to wildfires have the same problem.

To the second point, it is actually possible to significantly reduce the damage wildfires do to human infrastructure. Most ecologists recommend having “prescribed burns” (also called “controlled burns”), which are deliberately set wildfires meant to combust excess wood and underbrush. A large area to be burned is marked off, the fire is lit, and firefighters patrol the edges of it to make sure the flames don’t escape. Problematically, these planned fires are expensive, take a long time to get legal authorization to proceed, and are opposed by local people who only care about the short-term impacts of smoke and the threat of the fires getting out of control.

So what solution is left? Well, I have one proposal, and like so many ideas that spring from my mind, it is just as logical as it is crazy-sounding: We use thousands of teams of goats and robots to roam through America’s forests to destroy excess wood and plants. I like it because it fuses something natural and prehistoric (animal grazing) with something futuristic.

Each team would consist of two “sub teams”: a herd of goats that would first eat all the light vegetation and small sticks, and a group of “robot lumberjacks” that would then gather and burn all the larger pieces of wood the goats weren’t able to consume. Between the two of them, they would create a “fuel break,” which is an area where excess biomass has been removed from the ground and the trees have been thinned. Wildfires can still burn fuel breaks, but since there is little combustible material in them, the fires spread through them slowly, don’t get as intense, and are easier for firefighters to extinguish.

The goat/robot teams would roam the forests and grasslands at the fringes of human-populated areas, creating fuel breaks. Since the biomass would either be eaten or burned piecemeal in portable stoves, there would be no risk of the teams’ activities significantly damaging local air quality, or setting fires that got out of control. With mildly intelligent AI, the teams could be almost totally automated, needing very little human oversight.

To understand in detail how the system would work, imagine we’re in a government-owned forest that abuts a suburban development somewhere in California. To make things simple, let’s say the edge of the suburb is a straight line several kilometers long. Looking at it from above, all the back lawns of the houses on one edge of the suburb end at a straight line of trees marking the start of a forest. The part of the forest touching the lawns is divided into imaginary, one-acre squares of land, with each square measuring 63 meters to a side.

A herd of goats arrives in one of the imaginary square acres. Over several days, they eat the living vegetation, leaving behind only dead wood on the forest floor. Their droppings also fertilize the soil.

The goats would wear electronic collars containing GPS locators and metal prods for delivering mild electric shocks. As with an invisible fence meant to keep pet dogs from roaming outside their owners’ yards, the collars would shock goats that started to stray from the designated one-acre zone.

A few different robots would watch the herd and move with it. They would do things like supply the goats with water, help them if they got physically trapped or injured, and fend off predators by using nonlethal means like pepper spray. The machines would also monitor the goats’ health and nutrition status.

A herd of 30 goats takes roughly six days to eat an acre’s worth of vegetation, and that sounds like a manageable number of animals for the robot shepherds to keep track of, so let’s choose that for the size of our herd. After the goats graze in the 63 meter square of land for six days, the robots visually confirm that the animals have eaten all they can, and then they slowly shift the boundaries of the “electric fence” to drive the herd into the next one-acre box of land to start eating the vegetation there. The shepherd robots move with them.

An all-robot crew then moves into the acre of land that the goats just vacated. Their job is to eliminate any dead wood that remained on the ground, as well as cut down dead trees and excess saplings, and trim all branches up to a height of 3 meters above the ground. A crucial piece of equipment they has is a combination wood-burning stove and steam engine mounted in a small, off-road vehicle. Dead wood gathered from the square acre would be burned in the stove, and the heat from the flames would boil water inside the steam engine, which in turn would spin a turbine and generate electricity. The electricity would recharge the batteries of the vehicle, of the robot crewmen, and of their power tools. The system would be energy self-sufficient, reducing costs. The other robots that were tasked with taking care of the goats would walk to the stove/steam engine to recharge their own batteries when needed, and then go back to the goats.

The all-robot crew’s first on-site task would be to set up the wood stove / steam engine. For obvious reasons, it would probably be moved to the geographic center of the acre. A chimney would be attached to its top to catch sparks and filter the most poisonous gases from the smoke. For the second task, the chimney would contain a removable “catalytic combustor,” which is a common feature in modern American wood-burning stoves.

The robot lumberjacks would then get to work. Thanks to the goats consuming most of the underbrush and low-hanging tree branches, it would be easy for the robots to move around the area, see pieces of dead wood on the ground, and pick them up. All of the biomass marked for removal would be put in the stove and burned. The robots would use electric chainsaws, log splitters, and other tools to cut anything too large to fit into the stove into sufficiently small pieces.

The opportunity would also be used to remove human-created trash from the acre. Anything that was safely combustible would be thrown in the stove while the rest would be bundled in a pile and geotagged for eventual pickup.

Over the course of six days, the robot crew would slowly feed all of the acre’s dead wood and excess vegetation into the stove. They would periodically remove wood ash from the stove, wait until it was no longer hot enough to cause a fire, and sprinkle it on the ground to fertilize the soil. One member of the robot crew would be a small vehicle meant to carry water and spray it on fires accidentally lit by sparks from the stove, as well as douse the ashes before they were spread on the ground. It would use local bodies of water like streams and lakes to replenish its reservoir, and might also provide the goats with drinking water.

What would the other robots look like? To move around over uneven forest terrain, between closely-spaced trees, and under branches, they would need legs, and they couldn’t be much bigger than human adults. Some of the robots in the crew would also need to have body layouts that gave them general-purpose work abilities so they could do things like assemble and disassemble the stove / steam engine, pick up large pieces of wood, replace chainsaw blades, and make minor repairs to themselves and other robots. With those requirements in mind, I think most of the robot crew would be humanoid or centaur-like, and would have one or two pairs of arms and hands for grasping tools and objects. (For a deeper discussion of this topic, read my blog entry “What would a human-equivalent robot look like?”)

Once the excess wood in the area was all burned, the robots would dump the last of their ashes, configure the stove/steam engine and any other equipment for travel mode, and move the system to the next burn site a few hundred meters away, again moving into the new acre square as the goat herd moved out. This process would repeat itself indefinitely, with the goats and robots slowly creating a 63-meter-wide “line” of thinned trees, trimmed branches, and debris-free forest floor. Assuming the goat/robot system spends 50% of its time working and 50% out of action due to adverse weather, maintenance, and other factors, it could transform 26 acres of forest into fuel breaks over the course of one year, making the “line” 1,638 meters long (almost exactly 1 mile).

To stay effective, fuel breaks need full maintenance once every 10 years, which means that one crew of 30 goats and maybe 10 robots could create and sustain a fuel break 16 kilometers (10 miles) long and 63 meters wide. Put into perspective, it would take 116 crews to make a straight fuel brake extending from the U.S.-Mexico border near San Diego to the U.S.-Canada border near Seattle.

Of course, one fuel break paralleling the West Coast won’t solve America’s wildfire problem. We’d probably need fuel breaks 100 times longer than that, scattered all over the country, and in irregular configurations around the fringes of towns and suburbs, to significantly cut the amount of damage wildfires cause each year. It might sound like a lot, but doing the math, it’s feasible, at least with the technology we’ll have later this century.

11,600 fuel break crews would require 116,000 robots, which on average would be the sizes of adult humans (the water carrier vehicles, which should be thought of as autonomous vehicles designed for off-road use, would be larger and heavier). That might sound like a lot of robots, until you consider there were 245 million passenger vehicles in the U.S. in 2018. If we can afford to build and maintain that many large, complicated machines, then it should be possible to create a vastly smaller fleet of lighter and less complicated machines.

The crews would also need 348,000 goats, which is indeed a large number, but achievable when you consider the total goat population of the U.S. was 2.66 million in 2020. It would take only a few years of more intensively breeding the existing goat population to expand it by 13%–the amount needed to populate the fuel break crews.

Automation would keep the system’s costs low, and it would be rare for human staff to have to travel to work sites (reasons might include veterinary care for the goats, or major repairs to broken robots). The amount of human deaths and property losses averted by the system would, hopefully, more than pay for its costs. According to my own estimates, AI and robotics should be advanced enough to make the first goat / robot crews sometime in the late 2030s. However, due to public skepticism of the idea (if it is even known to a non-token segment of the public by then), I think this idea or any variant of it won’t come to fruition for decades after that.

Links:

1. In 2021, wildfires burned 7.13 million acres of land in the U.S. The yearly average was 7.47 million acres.
   https://disasterphilanthropy.org/disaster/2021-north-american-wildfire-season/
2. ‘By some estimates, many of [California’s] forests have up to 100 times the amount of small trees and underbrush than what grew prior to white settlement. Meanwhile, researchers estimate that prior to 1800, some 4.5 million acres of the state’s forests burned in a typical year — more than the 1.9 million acres that burned in 2018, the most in modern history. Yet in a state with more than 30 million acres of forest, only about 87,000 acres of California land were treated with prescribed burns last year to reduce undergrowth prior to the state’s deadly fire season, according to data from Cal Fire, the U.S. Forest Service and the U.S. Bureau of Land Management.’
   https://www.nytimes.com/2015/04/14/science/californias-history-of-drought-repeats.html
3. There are political and cost barriers to doing prescribed burns.
   https://reason.com/2020/09/14/western-wildfires-can-be-prevented-if-burdens-on-forest-management-are-eased/
4. Goats have long been used to remove overgrowth.
   https://www.npr.org/2020/01/05/792458505/california-cities-turn-to-hired-hooves-to-help-prevent-massive-wildfires
5. Sixty goats can eat an acre’s worth of vegetation in three days, which means 30 goats can do it in six.
   https://www.knoxgoats.com/faq
6. A detailed description of what a “fuel break” is.
   https://www.portugalwildfires.com/what-is-a-fuel-break/
7. Fuel breaks should be maintained once every 10 years.
   https://www.fs.usda.gov/nfs/11558/www/nepa/92563_FSPLT3_3949473.pdf
8. There were about 245 million passenger vehicles in the U.S. in 2018.
   https://www.bts.gov/content/number-us-aircraft-vehicles-vessels-and-other-conveyances
9. The U.S. goat population was 2.66 million in 2020.
   https://www.nass.usda.gov/Publications/Todays_Reports/reports/shep0120.pdf