Example 5: 



The Sundance Fire of September 2, 1%7, was reported by 

 Anderson (1%8) and illustrates a wind-driven fire moving from 

 a fire line 3 to 4 mi (6.4 to 9.6 km) wide. The fire size and 

 shape were examined at 1500, 1700, 1900, 2100, and after 2300 

 hours (fig. 13). The fire shapes do not fit well because the fire 

 on this day began its spread from a line rather than a point. 

 Spotting began some time near 1500 hours and continued 

 through at least 2000 hours. Crowning started after 1500 and 

 played a significant role after 18(X) hours. Midflame winds on 

 the fire front began exceeding 12 mi/h (19.3 km/h) at about 

 1800 hours and appear to have been near 16 to 17 mi/h 

 (27.4 km/h) until after 2300 hours. Up through 1700 hours the 

 fire was generally a surface and shrub fire and the wind reduc- 

 tion coefficient was 0.18, indicating overstory material was slow- 

 ing the wind's movement. After 1700 the wind reduction was 

 less because the fu'e spread was more through the shrub and tree 

 crown material. The earlier fire advance to the west is not 

 included. 



Even though the double ellipse model doesn't match a line fire 

 during its intermediate growth stages, a projection of the final 

 size and shape after an extended run appears possible. Figure 13 

 shows this in the projection of the size after 2300 on 

 September 1, 1967. The average wind on the flame used to 

 estimate a total run, must consider the windspeed variation over 

 time. It also must consider how the free-stream windspeed 

 above the vegetation surface is reduced as the location of the 

 flame front is reached. In this case the average windspeed was 

 11.5 mi/h (18.5 km/h). For fires with wide fu-e fronts, it is 

 probably better to represent each edge of the fire as a point 

 source, project the fire advance from these points, and inscribe 

 the combined area. 



Other considerations that can be made with the material 

 developed on fu-e size and shape include estimation of the 

 backing and flanking rates of spread, interpreting the change in 

 size and shape over time, and using the quadratic equation to 

 determine any one of the three properties defining size and 

 shape knowing the other two. 



In appendb( II, the Freeman Lake Fire in Idaho illustrates a 

 fast moving crown fire that covered 20,000 acres (8 094 ha) in 

 12.5 hours (example 8). Jemison (1932) reports that by the 

 morning of August 4, 1931, the fu-e had covered an area 5 mi wide 

 and 11.5 mi long (8.0 km by 18.5 km), with some spot fires 15 mi 



(24. 1 km) from the origin. Use of the double ellipse model and 

 the wind reduction concepts suggests this fire had a ^/w ratio 

 of 2.3: 1 for an average wind on the fire of 6.8 mi/h 

 (10.9 km/h). The wind measured at the 150-ft (45.7-m) level at 

 Priest River Experiment Forest was an average of 14.9 mi/h 

 (23.8 km/h), so the wind reduction factor was 0.46, which is 

 reasonable for the upper surface of the vegetation layer or tree 

 crowns. Solving the quadratic equation for perimeter yielded a 

 value of 1,8(X) chains (27.3 mi or 43.7 km), which agreed with 

 the double ellipse solution for an area of 20,(X)0 acres 

 (8 094 ha). Fires with similar rapid spread and growth have oc- 

 curred in various regions of the United States. Jim Miller^ 

 notes that several fires in Wisconsin had the same features — the 

 Brockway and Five Mile Tower Fires of 1977 and the Oak 

 Lake Fire of 1980. The Mack Lake Fire in Michigan^ has 

 features of fire behavior that are like those of the above fires. 

 Records of these types of fires will help evaluate the model of 

 fu-e shape, (examples 6 and 7). 



The Big Scrub Fire of 1935 on the Ocala National Forest in 

 Florida — provides a data point at the high //w ratios shown in 

 figure 7. This fire traveled 18 mi (29.0 km) in 3 hours and had 

 an estimated area of 10,0(X) acres (4 048 ha). A wind shift 

 resulted in another 25,0(X) acres (10 120 ha) burning before 

 rains put it out. Winds were reported to be 60 mi/h 

 (%.5 km/h) from the southeast, but the i/w ratio computed 

 from the quadratic equation, if the perimeter is estimated at 2 

 times the spread distance, was found to be 16: 1 . This cor- 

 responds to an average wind on the fire of 25 mi/h 

 (40.2 km/h). The fire was reported as a crown fu-e in sand 

 pine, P. clausa. The calculated wind reduction factor of 0.42 

 indicates the average wind was exerted at some point near the 

 upper surfaces of the tree canopy, according to Albini and 

 Baughman's (1979) presentation on estimating windspeed, 

 Big Scrub Fire is given in example 9, Appendix II. 



Interestingly, the windspeed of 25 mi/h (40.2 km/h) is essen- 

 tially the windspeed McArthur (1966) found associated with the 

 maximum rate of spread in grassland fuels. The above wind- 

 speed at the 33-ft (10-m) height above the vegetation was found 

 to be associated with fu-e shapes having i/w ratios of 6: 1 . If 

 the grass is 1 ft (0.3 m) deep and the flame height is 1 to 1 .5 ft 

 (0.3 to 0.5 m) above the upper surface of the grass, the average 

 wind on the flame is computed to be between 15 and 17 mi/h 

 (24. 1 and 27.4 km/h). This i/w ratio and windspeed match 

 closely equation 17 results (fig. 6). 



Jim MiUer, personal correspondence of October 26, 1981, on file at Northern 

 Forest Fire Laboratory. 



"The Mack Lake Fire," by AJ Siinard and others, in preparation. 

 From notes by Ocala National Forest Ranger John W. Cooper. 



17 



