There is one thing I'm always thinking about: a mesh network of autonomous devices that scattered across a field or a forest and a base nearby. These devices can have temperature sensors and someone can rather quickly see that something is going wrong either by temperature or if some amount of devices went down (say burnt in fire).
I think those things called sensor networks or something like this.
I am not sure how can this scale up, if a sensor can cover X square meters then you need a LOT of them to cover huge forests like Sweden has for example, not to mention the amount of bases, communication channels etc.
If it's on a tower on the peak of a hill, the coverage might be measurable in miles. That's the case with fire towers today. But yeah, even then scaling could be difficult. It might work for realtively smaller, more developed areas prone to fires, such as parts of CA. Maybe put them on existing cell towers?
I think there would be better luck using satellite imaging, both thermal and regular vision. You could then change the satellite to look for fires in different areas based on the risk factors - eg California might have gotten a lot of rain recently so you want to watch Montana. There are some restrictions on the satellite imaging, so that and cost would be a concern.
Fighting or detecting ? I think that the most efficient way of detection is by covering a lot of area at once and satellites are best for that, I see that there are multiple efforts being done around automatic real time detection and alerting of forest fires from above.
Aerial drones with thermal imaging cameras to predict where wild fires will occur.