My Photo

July 2014

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    


« 'Public Pension Ponzi' – Some Reflections (updated) | Main | Ruminations - 19aug2009 (updated 20aug09) »

18 August 2009


TrackBack URL for this entry:

Listed below are links to weblogs that reference Wildfire Containment Percentage – Say What??!!:


chip carman

Call Pascale at Yubanet. They know their stuff!

George Rebane

Have sent them an email. Thanks Chip.

Dan Getz

Well, now I'm interested. :) My guess is that is correct, meaning that the statistic is pretty useless except at that static point. A group of "percentage contained" data points, while not letting you know if the fire was growing or shrinking, would let you know whether the fire has been growing faster than it could be contained.

Bob Hobert

George, I'd be very interested in getting our out-of-control US Congress 100% contained. That would free up funds for fire prevention.

George Rebane

Dan, agree with your interpretation. But even if that part of it is correct, it still doesn't tell us what it means on the ground, let alone allowing us to calculate it from field data.

Bob, I hope you're not holding your breath :)



Containment is calculated by the miles of controlled line around a fire. If a fire has a 10 mile perimeter and they built 7 miles, the fire is 70% contained.

On the Yuba Fire, 4 miles of line remain to be built as of this morning.

Pascale - YubaNet


I agree with Pascale. However, determining the perimeter of an amoebic blob can be difficult.

Nolan Love

I agree with your point, Dr. George, that it is not very useful to give simply the sole metric of percent containment (say, as a fraction of the perimeter which is under control) without also offering the length of that perimeter, and the area it contains. Let me offer that saying "A wildfire is actively burning an area of 50,000 acres with a perimeter of 18 miles, 6 miles of which are under control; 10,000 additional acres have already been razed," would give a better snapshot of the fire's status and progress. This description falls short, however, when the perimeter itself is unevenly convoluted: a fire burning along the famously tortuous Maine coastline has a very long perimeter which is contained by the ocean, and perhaps a much shorter perimeter which threatens the inland. A standard polygon smoothing algorithm like the one used in Adobe's vector graphics programs could better normalize the descriptions of footprints of various wildfires, and improve the accuracy of a perimeter/area approach to communicating their progress and containment.
Your overarching frustration seems to come from the media's use of a single metric to describe the state of a multivariate system. If we really did need to confine ourselves to one number that would communicate absolute progress toward containment (for the sake of expediency), then here's an idea: Using aerial maps of the fire divided into acre-sized squares, count the number of unburned acres directly threatened by the fire (e.g. adjacent to uncontrolled fire). Add probabilities in for fractional acres if you like. What you get is one number which should decrease proportional to overall containment, and is ultimately what we care about in terms of stopping the wildfire: what is still at risk. Call them Risked Acres and your news might say "Wildfire at 15K RA's, down from 100K yesterday." Perhaps a more useful single metric than percent containment.

George Rebane

Yes, Pascale's answer is the direct and easiest to understand containment, especially if you assume that 'contolled line' is a part of the perimeter that has zero probability of fire penetration to unburned real estate. Things get a little more complicated when you have non-contiguous fire areas, and areas that have one or more doughnut hole in them, and areas that have unburned areas adjacent to a burning uncontrolled perimeter which areas will with high likelihood burn at a high rate of area coverage, etc. All these realworld considerations depend on to what use a 'containment' report will be put. If it's a snapshot feel good/bad assessment for the news media, then the one Pascale gives would serve. However, if it's to be used for allocating fire fighting resources in a complex of regional wildfires, then something more sophisticated using expert probabilities/beliefs and expectations need to go into the containment metric algorithm. I guess it's still worth noodling about.

George Rebane

Exactly so Nolan, that's a good summary and gives an idea of what our 'board talks' were like years ago. The simple algo from Pascale (and much appreciated) will hide a fire that has become much more dangerous while its containment metric has increased. I would like to believe that CDF and other firefighting agencies are ahead of the curve on all this. If not, I wouldn't mind getting back into harness and helping them develop such an interactive system that does both real time assessment and optimum resource allocations using both expert input, and a stream of various unreliable field observation inputs. Great fun.

ps. My guess is that you would too.

Kasey Allen

Natural resource metrics are occasionally vague and undefinable, like percent contained. My opinion is that the lack of extremely accurate and timely data are often the reason. One could even question the reported size of the fire. Did 100% of the 3,500 acres completely burn? Probably not. There are all sorts of holes and gaps and sometimes spot fires.

Just because an accurate and telling metric doesn't exist doesn't mean that it soon wont. Nolan's idea is interesting...


The containment percentage on fires is given as a progress report. The incident management teams are way ahead with contingency planning, resources at risk assessments, and resource allocation distribution systems. It is fully spelled out every day for all the firefighters in the 30-page briefing report issued at the beginning of every shift.

George Rebane

As discussed above, the current metric presented by Pascale need not represent "progress" in containing the fire. Nolan's algo is a much better one. If memory serves, our consideration of risk acres included what we called something like 'time extended risk area' which involved land that would be burned with high likelihood in the next specified time interval (say 24 hrs) given only current resources for containment. The calculation also had an a 'time extended expected risk area' that used a more refined and expanded set of probabilities. But the main idea was a given risk time horizon that was a function of a resource deployments. Playing with various deployments would help find an optimizing, or at least a satisficing solution.

The 1993 Malibu fire burned many tens of square miles and ended at approximately 250 yard fire line after burning for the better part of the week. This fire line ran through our property near the intersection of Saddle Peak and Tuna Cyn Roads, and was fought with five heavy lift helicopters that dropped salt water on that ridge, including our house. (The orbit included diving down Tuna to the ocean, filling up, climbing up Las Flores to make the attak run. Our driveway was a saltwater creek during those hours.) But here's the point.

'Pascale's metric' would have yielded 99.999% containment due to the fire's huge perimeter, and I recall something like that was reported on the media. Yet if that 250 yd fire line did not hold, then the fire would have dived down Fernwood-Pacific into the community of Topanga containing about 800 tightly packed houses in brush and vegetation that had not burned for over 30 years.

The firefighters (and all of us on the ridge) knew this, and that is why that line had everything available thrown at it that November afternoon. (The fixed wings were fighting another fire in Redlands, else they too would have been there.) Had not that line held, in addition to the loss of the houses, an estimated 4,000 acres would have been added to the burn area in the next 24 hours. That happens to be the real world, and it made clear to me that the simple containment metric would serve neither the public, residents peripheral to the fire, nor the managers of the fire fighting effort.

And all this occurred so fast that none of the thousands of Topangans in Topanga Cyn knew what was really happening. They thought the danger had long passed and were getting ready for a return to normalcy instead of a very rapid evacuation.

Nolan Love

Yes, yes. That makes even more sense. Good idea. I was thinking of sampling at even time intervals, but even better would be to include the dimension of time when assessing and reporting the risk.


George, I remember that fire very well as one of the folks that stayed and took good care of personnel from three engine companies from Central Calif parked in my driveway for four days. In fact, the fire did dive into Topanga on the second afternoon, halted by the firefighters' efforts about 50yds west of our house, right where Tuna Canyon turns into Fernwood Pacific. Kate Talamantez' house next door would surely have been lost but for having been completely bubbled in foam. At one point that afternoon, we were surrounded by fire areas, NSEW. Unforgettable.

The only thing those blessed firefighters wanted was the use of a shower every morning and I was happy to oblige. When Barbara returned on the final day, she treated them all to a huge spaghetti dinner with a great Caesar salad and we thanked them as best we could. The had spent 16 hours on the road struggling up over the grapvine and arrived at about 2AM when I was on the roof hosing everything down. About noon the second day, I ran into a bunch of Idaho smoke jumpers who had fought the fire all night down the east side of Saddle Peak and were almost trapped by a chain-link fence at the bottom. They cut through and wound up behind Kate's, and I found them sleeping on the ground, not hearing the roar of helicopters dumping all around. Ah, memories... Nice to live now in a place that has no such problems, nor earthquakes,hurricanes,floods,tornados,snow,sleet,volcanos,
tsunamis, avalanches. Only the most beautiful lightning storms imaginable...and the occasional "economic refugee"...

George Rebane

Larry, good to hear your memory of that fateful week. After the initial fire storm, Jo Ann and I also invited two engine companies to make our house their headquarters. Our water was out, but Jo Ann kept making 'tons' of tuna and lunch meat sandwiches and serving soft drinks from our ample food pantry. With two very large engines parked in the driveway and about ten or so firefighters in the house with their radios remoted, we felt very safe when the fire returned up Tuna Cyn two days later. In the interval the firefighters had a warm place to sleep, plenty to eat, and two tables around which to play cards while they were on standby. They were a little apprehensive when I told them that our boa constrictor Annabelle was loose somewhere in the house. I found her behind the books on a low bookshelf that indicated how low the smoke filled the house as the living room burned. Normally she always hid herself on the very top bookshelf when we let her out for her 'walks'. Everyone got a kick our of her, but felt much better when I put her back in her terrarium.

With respect to containment, and expecting to be branded a skeptic, what might be even more interesting is how much containment is prescribed for any particular fire. Along with that would have to go, who makes these decisions, what are the basis for those decisions, what factors other than forest management are taken into account such as air quality and public health, and how the property value factor gets applied to the equation?

After that it would also be interesting in seeing the "forest management" statistics on previous fires compared to the total cost of that campaign and the total cost in property loss.

George Rebane

Well said AItOawmNWbVtDRGQBpPJEsSqSh1AePLXrZrp3tc (may I call you '3tc'?). That indeed is the larger question that should be addressed in a macro-utility function which would then presumably be used to generate response policies/tactics addressing the points you make. Most parts of the country are now over-vegetated do to modern fire supression methods and policies. The fuel loadings (BTUs/acre) are highest since the stuff started growing eons ago. Attmepts, such as controlled burns, to reduce such loadings and prevent the inevitable wildfire disasters have been foiled by our legal industry.

Melissa Rhyner

Oh my goodness. This is so fascinating. I'm in Colorado with the now contained High Park fire and the still uncontained Waldo Canyon fire in Colorado Springs. After hearing containment updates for the last several weeks, I was just as curious and thank you so much for this thorough but technologically digestible post and discussion. It's great watching the containment % climb, but so appreciated to have an understanding of what it really means.

George Rebane

Welcome MelissaR, and I'm glad this discussion was at least interesting to you. We hope that you and your family are not in the reach of any of these fires.

The comments to this entry are closed.

Blog powered by Typepad