A Modest Proposal to Address Wildfire Danger in the American West

Wildfire in Montanna, image by John McColgan


I’ve been a California resident since January of 2010. In that relatively short time I’ve seen first-hand the effects of a changing climate on life in the San Francisco Bay Area. When I arrived in 2010, winters were still the foggy season, the Sierras regularly had a snow pack through the spring and early summer, and wildfire was limited to a summer-autumn season. Now, droughts are the norm and the fire-season doesn’t really end in California.

There are many things we need to do collectively in response to the changing climate, including improved forest management and reconsidering the relationship between development and wildland. And while I’m an educated lay person, I’m not an expert in forest management or urban planning and I don’t have specific recommendations in these areas.

Project Loon luanch event, Christchurch NZ 2013

I am an expert in a range of technologies relating to sensing, drones, and large, distributed, systems. I’ve worked on a range of autonomous and semi-autonomous systems over the years, and was the first technical architect of a globe-spanning stratospheric communications network based on high-altitude balloons. I know something about aerospace, earth observation, and engineering systems at the scale of a state or the globe.

It is from this perspective that I believe I have some suggestions about what we can do to make California and the entire West safer in the coming decades of the Anthropocene.

So, what’s the problem to be solved?

The fundamental problem to be solved is the impact of wildfire as measured in the cost of human lives and infrastructure damage. And while there are many ways to approach this problem, one space of solutions involves detecting fires sooner, when they are very small, and responsing faster, while the fire is still a local, easily contained event.

Detecting and responding to fire quickly is particularly a challenge when fires start in remote, difficult to access terrain. Even when fire fighters know where the fire is, they may not be able to get there quickly. Worse, the very conditions that can make fire spread quickly — high winds — makes using conventional piloted aircraft to fight fires even more dangerous.

How do we solve this?

I don’t believe that there is a single, technological “silver bullet” solution to this problem. However, the right technological system deployed in conjunction with human responders and designed in a thoughtful way could have a large, positive impact.

Solar-powered long-duration autonomous aircraft prototype, NASA

Imagine a collection of autonomous aircraft or aerostats loitering in the stratosphere above the wildland of the Western United States. These vehicles have mid-infrared thermal imagers — heat sensing cameras — that focus on the ground, and provide continuous, automatic 24/7 fire spotting. The near-infrared imagers would be able to spot fire through most cloud cover and even light rain. The imagers would have sufficient resolution to spot a fire as small as a cooking fire or a few burning branches and the imagery would be processed automatically and in real-time by cloud-based AI.

As soon as candidate fires or hot-spots are identified by automatic surveillance, a human monitoring team of wildland fire fighters would be notified and provided real-time, high-resolution imagery of the candidate fire in progress. The monitoring team could then dispatch on-the-ground or aerial response resources as needed.

Semi-Automated Aerial Fire Suppression

Photo credit: Daniel Ingrassia

This human-in-the-loop automated surveillance system could be extended with semi-automated airborne fire response as well. During red-flag warnings or other high-risk conditions, autonomous fire suppression drone aircraft could be pre-positioned to general aviation airfields in effected regions. These aircraft might resemble conventional fixed-wing or rotary-wing aircraft, modified to support autonomous piloting and equipped with medium-scale fire retardant dispensing capabilities. When fire starts in steep or inaccessible terrain, these aircraft could be launched with the press of a button and fly automatically to the fire zone, once there to be guided and controlled by remote human pilots to deploy fire suppression.

The guiding principle is that small fires are easy to control, but small fires can quickly grow into catastrophes under the right conditions. If the automated monitoring system could alert human responders within minutes, and dispatch semi-automated aerial resources to arrive within ten minutes, small fires are much less likely to grow out of control. This approach wouldn’t work under all conditions or with all potential ignition sources — imagine a gas line fire, or a burning rail car full of petroleum derailed near dry forest or grasslands — but for lightning-driven wildland fire or many human-caused wildland fires it could make a huge difference.

Infrastructure Monitoring

A further extension would be to increase surveillance and response capabilities along critical infrastructure, such as high-tension power lines. Low-altitude (under 400’ above-ground-level) autonomous surveillance drones could overfly the right-of-way of power lines on a frequent basis, performing automated high-resolution thermal inspection of the lines and the surrounding area. Pre-positioned fire suppression drones or semi-automated tracked vehicles capable of spraying fire suppression could be staged along the power-line right-of-way. The frequent, high-resolution inspection could lead to problems being discovered well before they progress to becoming ingntion sources. If there were an unanticpated ignition, fire would be detected quickly and the pre-positioned response resources could significantly decrease response times.


The bigger-picture solution to the wildland fire risk in Western states and California involves green house gas emissions control, improved forestry and wildland management, improved urban planning at the human-wildland interface, and potentially designating certain regions as uninsurable due to increased fire risk. However, an automated wildland fire detection and semi-automated response system like the one described here is technically feasible and could make a significant impact. Even eliminating one Camp fire class catastrophe would pay the capital and operating costs of such a system for many years.

While technially ambitious, this system is buildable today with technologies that are either commercially available or in an advanced state of development, nearing commercial availablity. I believe that a prototype capability demonstration could be stood up within 12 months for less than 10 million dollars, and a full implementation suitable for covering large areas of the West could be deployed within two years, costing less than a billion dollars. Putting this number in perspective, the estimated cost of the recent Camp fire in California alone is expected to total billions. Creating a system like the one described here is essentially a public good, and something that California and other Western states should invest in, perhaps in the form of a public-private partnership.

Founder of @hypersolve, co-founder of Project Loon. Formerly leader of early-stage innovation activities at X, the Moonshot Factory.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store