Software:
The most interesting
project that Ernie has been nibbling at involves creating wildfire training modules within video
games that already have good terrain modeling. We noticed on a
workshop trip to the American Southwest that the terrain modeling in the game "Fallout New Vegas" is eerily accurate to the real area.
The Fallout series game developers have made parts of their code open-source, allowing fans, players, and third-party companies to publish "mods" that change or expand on gameplay. There's a Weather Mod, for example, that allows you to have somewhat realistic and random weather and try the same scenario under different conditions (with factors like visibility or cold that might affect outcomes).
We got very interested in exploring this platform for interactive wildfire-fighting scenarios. These could be fun, raise awareness, and if carefully designed with attention to details like tactical rules of thumb, might also serve as winter training / recruitment tools for actual wildland firefighters.
The modding community has been playing with this idea at Ernie's request. But in attempting to make something that looks and feels realistic, they repeatedly come up against limits of processor power and graphics. Fire modeling at scale tends to crash either the player's computer, or the game server.
I've been encouraging Ernie to start with the basics - use known methods of simplifying the fire behavior for notepad calculations, and don't worry about the accuracy of the graphics. Develop something that is tactically useful and accurate
enough to get the wildfire community interested, if not as graphically impressive (think Minesweeper or Atari game graphics).
The fire incident command teams have been faced with this problem for a long time (predicting fire behavior, at least on a gross scale, effectively enough to allocate resources and plan daily tactics). Prior to the widespread availability of handheld computers (smartphones) and apps, a lot of this was done with pencil and paper, and those reference tables might be a reasonable place to start for a simplified ruleset.
The three main factors are terrain, fuels, and weather.
Terrain (slope) affects fire speed and behavior, as well as crew work speed, risk levels, and escape speed. Radiant heat trapped in canyons or chimneys; safety zones, road access, backup routes, community assets and hazards, etc.
Fuel types affect how fast fires move, how much heat is produced (risk to structures & permanent assets), how tall/long the flames are, how fast moisture levels change, and how long the fire burns intensely, all of which have to be considered for effective strategy and tactics.
Weather affects fire speed, drying conditions, crew work pace (heat illness), etc. Accurate 3-day weather predictions are a critical tool for planning large-scale fire incident response.
Here are some existing tools from a cursory search:
Very simplified FLAME model (4 fuel types):
https://training.nwcg.gov/pre-courses/s290/S-290%20Student%20CD/FLAME%20in%20a%20Nutshell.pdf
Standard fuel model (13 types) and factors for slope:
https://training.nwcg.gov/pre-courses/s490/s490pcw/1992%20Fire%20Behavior%20Field%20Ref%20Guide.pdf
https://training.nwcg.gov/pre-courses/s290/FHB_Appendix%20B.pdf
Existing computer-based tools:
BehavePlus fire behavior modeling:
https://www.frames.gov/behaveplus/home
There are also recommendations for homeowners on preparing their homes for defensibility. Though this might be less exciting for the usual Fallout audience (consequence-free mayhem is kind of the market niche for most first-person
games), preparedness "contests" could be strategically more relevant as a public take-away from this effort.
https://www.nfpa.org/Public-Education/Fire-causes-and-risks/Wildfire/Firewise-USA
There are also courses, intended for command and meteorological staff, on fire behavior calculations: S-390 and S-490 in the NWCG training catalog. We could probably get access to course materials for programmers interested in working on efficient modeling for training scenarios.
Knowing these details gives you very meaningful trigger points and goals - "stop the fire before the wind change" or "before it reaches the other side of the valley (slope/wind alignment)" or "before it escapes from the forest into the grassland."
There are many thousands of wildland fires every year that don't make the news, because
local forces stop them quickly and effectively. Most of the state and national forestry agencies I've worked with in the West have targets like keeping 97% of fires to under 1/4 acre - which they routinely meet and exceed. Mini-games with modest-size fires (less than 1/10 acre) could be a good starting point to practice specific tactics, such as working from the heel of the fire toward the head (anchor and flank), or structure protection, or tactical prioritization that could be roughly turn-based (so many hours to order a dozer vs. going to resupply with
water vs. time it takes crews to dig fireline by hand).
It's a game. These things won't be perfect, but they could be a fun and interesting contribution to helping communities respond effectively to potentially devastating risks.
You can PM Ernie Wisner if you're interested in playing in this space.
9
