Colorado Firecamp - wildfire training wildland firefighter training Wildfire Blog Engine Boss Apprenticeship Location and Facility About Colorado Firecamp Frequently Asked Questions

Colorado Firecamp - wildland firefighter training
Swiss Cheese Model

swiss cheese slice

The Human Factors Analysis and Classification System—HFACS

Cover and Documentation
1. Unsafe Acts
2. Preconditions for Unsafe Acts
3. Unsafe Supervision
4. Organizational Influences

HFACS and Wildland Fatality Investigations

Hugh Carson wrote this article a few days after the Cramer Fire

Bill Gabbert wrote this article following the release of the Yarnell Hill Fire ADOSH report

A Roadmap to a Just Culture: Enhancing the Safety Environment

Cover and Contents
Forward by James Reason
Executive Summary
1. Introduction
2. Definitions and Principles of a Just Culture
3. Creating a Just Culture
4. Case Studies
5. References
Appendix A. Reporting Systems
Appendix B. Constraints to a Just Reporting Culture
Appendix C. Different Perspectives
Appendix D. Glossary of Acronyms
Appendix E. Report Feedback Form

Rainbow Springs Fire, 1984 — Incident Commander Narration

Years Prior
April 25th
Fire Narrative
Lessons Learned

U.S. Forest Service Fire Suppression: Foundational Doctrine

Tools to Identify Lessons Learned

An FAA website presents 3 tools to identify lessons learned from accidents. The site also includes an animated illustration of a slightly different 'Swiss-cheese' model called "defenses-in-depth."


Given February 1997 for use in the Fatality Fire Case Study Training Course

Fire Narrative

Upon our arrival at the fire around 1345 hours, the District Ranger told me to take charge of the fire. Sizing up the fire was relatively simple. From point A (see attached map), I had an excellent view of the fire. The head fire had just reached the mountain top and the overall intensity was diminishing rapidly. Flames on the left flank were being pushed by a strong southeasterly wind but flame heights were generally less than 2 feet. The rate of spread was approximately 8 chains per hour on the left flank, but as you will see later, the rate of spread was greater along the lower portion of the south facing slope due to lens surface rocks and a much larger accumulation of surface fuels (pine needles and hardwood leaves).

I considered a number of options but never considered a direct attack on the left flank because of the steep terrain and wind gust up to 30 MPH from the southeast. It was also my feeling that the terrain was too steep for the tractor not to mention the inexperienced operator. I was not really concerned about the head fire that had already reached the top of Dallas Mountain and perhaps started down the north slope. The overstory vegetation on the north slope was hardwood with only surface fuels consisting of cured hardwood leaves. The fire would also be backing down the very steep slope that was largely shielded from the strong southeasterly wind. My main concern at that moment was the left flank since the wind was pushing the fire in a westerly direction down the south facing slope of Dallas Mountain.

As I recall, the District Ranger, ORA, Timber Sales Administrator, Tractor Operator and I were the only arrivals. The District Silviculturist was in route. Soon after our arrival a member of the volunteer fire department for the community immediately north of Dallas Mountain arrived on the scene. He informed me that the fire was already burning down the north slope of Dallas Mountain and also informed me that the people with homes along Road 42 were already in a state of panic. I radioed the District Silviculturist who was in route and requested that he drive west on Road 42 and see if he could locate a route into the fire from the north and also give me a report on how far the fire had progressed down the north slope of Dallas Mountain. He reported back a few minutes later that crews coming in from the north would have to travel by foot from Road 42. At the same time, he reported the fire had made little, if any, progress from the top of Dallas Mountain down the north slope. Although that news was reassuring, my brief conversation with the local fire department volunteer had brought back some horrors of the Acorn Fire in which I had to deal with home owners in panic at the same time I was attempting to direct operations on the fire. I certainly did not want the fire to reach the residential area along Road 42 and that sense of urgency most likely influenced some of my decisions.

Click for larger image.

At about 1400 the Poteau Ranger District Tractor Crew was dispatched to the fire. I was expecting a let of handline construction on the north side of Dallas Mountain and had requested an additional 50 firefighters for handline construction and burnout.

My plan of attack was to use our tractor to clean out a primitive road from Point B to Point F and then construct a tractor line around the west end of Dallas Mountain from Point F to the ridge top. In the meantime, I would let the hand crews walk around the right flank of fire to somewhere around Point G. They would anchor to the fire at that point with most of them working west with a handline and doing burnout at the same time. A smaller number of them could work back along the right flank. However, the fire on the right flank was moving very slowly into the wind and I did not consider that a priority. Again, my main concern was the fire moving west along the south slope of Dallas Mountain which I refer to as the left flank. However, due to wind direction and terrain, that could have logically been called the head of fire.

At about 1415 our tractor crew had started cleaning out the primitive road moving west from Point B. Our new Timber Sales Administrator was leading the tractor but had not been red carded or even received any training for tractor boss. The District Ranger and ORA were serving as line scouts. AT 1510 the Poteau Tractor crew arrived with Paul Keener as tractor boss and James Frizzell as tractor operator. At the time they arrived, I did not know where to send them. I had a scout checking the right flank of the fire to see if they could operate in that area. A few minutes later that scout informed me that he did not feel the tractor could work the right flank due to rocks and steep terrain. In the meantime, the Mena Tractor Boss informed me that the Poteau tractor could work in his area. I directed Keener to take his tractor down the freshly bladed road (left flank) until he made contact with the Mena Tractor Crew. It was only a few minutes later however at about 1521 when the Timber Sales Administrator informed me that they had reached the mountain top with the tractor line. I recall being very surprised that they had made that much progress in the relatively short time period he had been working on the line. Although we did not have a topo map at the fire, I had estimated the distance to the west end of Dallas Mountain to be about 1/3 mile from where they started at the origin of the fire. I thought the mountain dropped off at about the letter D (in Dallas) on the attached map. I made that assumption because from my location at Point A the ridge top appeared to be dropping off. I was also using my memory of where the mountain ended from my travel of the primitive road at a much earlier date. Unfortunately, the drop off was into a saddle and the actual distance to the west end of Dallas Mountain (Point F) was closer to 1 mile. Had I been aware of the longer distance, I may have called for the tractor line to be located in the same general area that it was located. However, I would have been very concerned about having people on the steep south slope working that close to the main fire, hopefully my actions would have reflected that concern. With that approach it would have been extremely important that the tractor line be anchored to the fire at the top of mountain and the line be constructed downhill with a very timely burn—out of that line. I would have also made certain the line scouts were closely monitoring the main fire. Whatever the strategy, I had very serious concerns about any of our people working the steep south slope in unburned fuels on the left flank of the fire. I spent the entire afternoon from initial attack until the accident thinking that none of our people were working in those conditions.

When I received word from the Timber Sales Administrator that they had reached the mountain top with the tractor line, I did not question their progress or the actual location of the tractor line. Instead, I told him to start the burn—out moving slowly back down the tractor line. Moments later he called to inform me that the backfiring torch was not on the tractor and no one had thought to bring a torch or any other firing device.

The employee under investigation that I mentioned earlier had joined me at Point A. Although he showed signs of mental stress and did not appear to be in good physical condition, he had passed the step test and was technically a qualified firefighter. Since no one else was available, I gave him a drip torch and told him to take it to the Timber Sales Administrator. From what I was told later, he stopped to rest several times along the way and did not reach his destination in a timely manner, but as it turned out, that may have been a blessing. Had he delivered that torch only a few minutes earlier it could have brought additional people into the blow-up site adding to the number of fatalities or serious injuries.

Most of the hand crews had arrived at Point A around 1530. Instead of sending them around the right flank as previously planned, they were told to use the tractor line and join up with the District Ranger and Timber Sales Administrator. The plan was to use some of them to burn out and hold the tractor line on the left flank but the majority would be used to construct and burn out a hand line working east from the tractor line that ended at the ridgetop. That line would completed a line around the fire and hopefully provided containment.

At about 1605 our fire detection aircraft flew over the fire and the aerial observer told me the fire had cooled down considerable from the time he had observed it earlier in the afternoon. That received some discussion during the subsequent investigation and review but the information from the aerial observer had nothing to do with my decisions. I knew the weather conditions had not changed and could see no reason to be complacent. The aerial observer was not a fire behavior officer and I did not feel he had the knowledge and experience to accurately evaluate the fire.

At 1610 the District Ranger informed me that he had scouted the right flank of the fire and felt the tractors could be used to construct part of that line. I called the Poteau Tractor Boss and requested that he bring his tractor back down the tractor line to my location at Point A. From there he would receive further instructions.

What I was not aware of was an incomplete section of tractor line at the fatality site (see attached map). One of our employees told the Poteau Tractor Boss to connect that section of line on his way down the mountain. That action held them up at the fatality site.

At about 1620, a member of the local volunteer fire department was located at Point E and observed a spot fire at Point D that he later described to be about 1/2 acre in size but spreading rapidly up slope toward the fatality site. He had no radio communication with Forest Service personnel but would not have known what was about to happen anyway. The spot fire moved rapidly up slope until reaching the area of preheated fuels. The influence of the mountain saddle resulted in a much higher wind velocity in that same area. The blow-up immediately south of the saddle as shown on attached map was so intense that it is difficult to describe. I was alerted by the very loud noise even as far away as Point A. It was very frightening to me although I had no idea the Poteau Tractor Crew members, Paul Keener and James Frizzell, were directly in its path. There is no way to describe the feeling that came over all of us when their bodies were discovered a few minutes later.

As indicated earlier, the rate of spread was much greater along the lower part of the south slope. The fire had moved to Point C probably around 1615 and it is likely that a very strong gust of wind carried a burning ember all the way to Point D. We had an Ignition Component of 78 but did not know that until after the fact. However, had there not been a spot fire, the main fire had already reached the base of the canyon located immediately east of the fatality site. It would have been only a matter of minutes until that fire made a run up the canyon with the same consequences, or worse. It could have been worse because the burn-out operation was getting ready to start. That would certainly have brought more people into the area. We should also remember that the hand crews used that same tractor line to reach the north side of the mountain only a few minutes earlier. As terrible as it was, it could have easily been much worse.

<<< continue reading Rainbow Springs Fire, Lessons Learned >>>


©2005-2006 Colorado Firecamp, Inc. home scheduleblogENGBfacilityabout usFAQ's