The effects of overprotection from fire in the mixed conifer forests are most pronounced between 4000 ft and 5000 ft in elevation. There the overstory is predominately ponderosa pine with incense-cedar. The forest has understory fuel types consisting of bear clover, pine needles, or incense-cedar reproduction. On the floor of Yosemite Valley, the pines stand above an understory of large and small incense-cedars. Investigations were made in each of these four types.
Fire behavior and subsequent effects are influenced by the direction that a fire is burning. Fires burning with the wind or upslope, called headfires, spread more rapidly and are hotter. Backfires spread against the wind or downslope and are slower and less intense than headfires. Both methods of burning were evaluated.
Prescribed burning can be accomplished only within certain ranges of fuel moisture, wind speed, relative humidity, and air temperature. Considerable variation exists within these ranges, however, and burning under various combinations of these variables will produce fires with different characteristics and effects. Closer examination of the variables shows that they are somewhat related. Air temperature and, to a lesser extent, wind speed affect relative humidity. Fuel moisture content is closely related to relative humidity and is one of the most important factors affecting fire behavior. Fuel moisture content serves as an indicator of the combined effects of both air temperature and relative humidity and is easily measured with fuel sticks that are commonly used for the California Fire Ranger Rating. The safe, effective range for fuel stick moisture content is from 6 to 20%. Four levels of fuel moisture within this range were chosen to study.
The fire characteristics considered in this study were rate of spread, fuel energy, intensity, and scorch height. To calculate the amount and rate of energy released, it was necessary to measure the amount of fuel consumed and the caloric values of the fuel.
Prescribed fires decrease fire hazards primarily by reducing the amount or weight of ground fuels. Fuel weight is distributed between fine fuels (duff, litter, small branches), heavy fuels (large limbs, logs, stumps), and vegetative fuels (grasses, forbs, low brush, tree reproduction). A direct measure of the reduction of fire hazard is provided by changes in the weight of the various kinds of fuel after a fire. Fuel weight reduction was used as a measure of fire hazard reduction.
Changes in plant composition were evaluated by measuring density by size class and basal area for each species in the understory before and after burning.
Ideally, the study plots should have been distributed throughout the mixed conifer zone of Yosemite National Park. Practical and administrative considerations dictated that burning be confined to areas southeast of Wawona and in Yosemite Valley (Fig. 2).
WAWONA. The Wawona area is typical of the mixed conifer zone in the park. It is characterized by an overstory of ponderosa pine, with some incense-cedar and an occasional California black oak.
The elevation is 4600 ft and the slopes are primarily southwest-facing. The understory on these slopes consists of a mosaic of fuel types. The pattern of the mosaic is about 0.1 to 0.25 ha in size. Patches of bear clover are interspersed with clumps of incense-cedar reproduction and areas with only pine needles. Heavy fuel was sparsely scattered over the area although there were some concentrations of downed "insect" trees. Figures 3, 4, and 5 show the three fuel types at Wawona.
YOSEMITE VALLEY. The valley floor presents a unique situation within the mixed conifer zone. The floor is practically flat and is surrounded by steep canyon walls. Because of the high walls, the south side of the valley is more moist than the north side and the character of the vegetation manifests this fact. Ponderosa pine and canyon live oak (Quercus chrysolepis Liebm.) dominate the northern stands. The designated burning area was on the south side of the valley near Sentinel Creek.
The understory is sparse consisting primarily of larger incense-cedar trees, with an occasional white fir or Douglas-fir. In the early summer there is also some herbaceous cover. In general, the nature of the Yosemite Valley area is most similar to the areas at Wawona that have only pine needles (Fig. 6).
There is considerable heavy fuel in the valley because of the number of trees that have been cut due to insect damage or Fomes annosus root rot. The heavy fuel, flat terrain, and cooler local climate contribute to a different fire behavior environment than at Wawona.
The models used for analysis were the analysis of variance model for the prefire and postfire fuel and vegetation measurements and for the caloric values, and the analysis of covariance model for the fire characteristics.
A three-way factorial design was used to analyze the effects of fuel type, fuel moisture level, and direction of burning on the fire, fuel, and vegetation variables. The covariance design, in addition, included wind speed. This increased the sensitivity of the model by removing wind speed effects from the other effects. F-tests were used to test the significance of the main effects, the interaction effects, and the covariate effects. Details of the models have been discussed by van Wagtendonk (1972).
Units of 5-10 ha were selected for study in each of the two areas. The Wawona unit was first stratified into understory fuel types by running parallel line transects through it at 20-m intervals and recording the fuel types on a map. The valley unit was also transected to locate a relatively homogeneous area.
The number of plots needed for the project was dictated by the experimental design. Burning at four fuel moisture levels with two methods of burning at each level necessitated 10 plots per stratum, including 2 control plots, or a total of 40 plots.
The plot size was determined by the minimum area necessary for an adequate fire to burn without being influenced by "edge effects." Experience with previous fires indicated that 100 m2 plots would give reliable results.
A 10 X 10-m grid was superimposed on each unit, and 10 plots were randomly selected in each stratum.
When adjacent plots were selected, one was randomly rejected and an additional plot selected. This was necessary to create a 5-m wide buffer zone around each plot. The total area burned for each plot was 400 m2, including the 100 m2 plot itself and the buffer zone. For the low intensity fires used, a 5-m strip was sufficient for the fire to reach an energy output equilibrium before reaching the plot, thereby eliminating edge effects and preventing unnatural burning conditions as the fire stopped. Additional rejection criteria were established for plots falling in two fuel types, or with excessive rock outcrops or heavy fuel.
VEGETATION. Overstory basal area by species was determined by measuring every tree over 3 m high standing within the plot. Every second tree standing exactly half way in a plot was not included in the determination. Overstory basal area was not remeasured after the fires since there was no change.
Understory vegetation was measured on four 1 m2 subplots distributed randomly within each 100 m2 plot. Subplots containing large overstory trees or excessive heavy fuel were rejected. Basal area was recorded for each species for all the trees in the 1-m to 3-m height class. Understory density by species was measured on each subplot by counting stems in three height classes: 0-0.3 m, 0.3-1 m, and 1-3 m. The same subplots were remeasured 1 month after burning to determine mortality.
FUEL. Heavy fuels, defined as having a diameter greater than 2.5 cm, were sampled for an entire plot by the line intercept method of Van Wagner (1968). Absolute and percent reduction in heavy fuel was calculated from the before and after burning measurements.
Sampling for fine fuel weight was done on randomly distributed, paired 20 X 50 cm subplots. After the first of each pair was located, the second one was established in conditions as nearly identical as possible to the first. For each pair, one subplot was randomly selected for prefire measurements, the other for postfire measurements. Paired subplots reduce variance by decreasing the difference between before and after burning measurements due to variations within the larger plot. Prefire measurements were made immediately prior to burning.
Since sampling for fine fuel weight is destructive, the "before" subplots were reconstituted so as not to affect the fire. Each fuel sample was divided into fresh, weathered, and decomposed needle layers which were then placed in plastic bags and sealed. Figure 7 shows a 20 X 50-cm subplot with the surrounding fuel removed. In part (a) the fresh and weathered layers were removed to show the decomposed layer. The fresh layer was removed in part (b) to show the weathered layer. Part (c) shows all layers intact with the fresh layer on top. Vegetative fuels were sampled similarly above each 20 X 50-cm subplot to a height of 3 m.
The accuracy desired, the variance of the variables, and the design of the experiment determined the number of fuel subplots. Type I and type II errors were chosen to be 0.05 and 0.10, respectively. Data from a pilot study in 1970 gave an estimate of the variance of fuel weight reduction. Using these values, the total number of subplots was determined to be 320 or eight pairs of subplots per plot. Figure 8 shows a typical plot with its subplots.
PLOT PREPARATION. After each plot and its subplots had been marked by steel stakes and sampled for heavy fuel and vegetation, a fire line was constructed around the plot and its buffer zone. The fire line consisted of a strip 0.5 m wide dug to mineral soil. Understory vegetation was cut away 1 m on each side of the strip and any logs which crossed the line were cut.
As an additional precautionary measure, a trail was constructed to each plot, which enabled a standby truck with a slip-on tanker to reach it.
FIRING DETERMINATION. The selection of the plot to be burned, the method of burning, and the order of their burning were random. As soon as the areas were clear of snow in the spring, fuel sticks were placed 12 inches above ground at each plot. These were monitored for at least 48 hours to determine when a plot would be burned. The four fuel moisture levels were the following intervals: 9-11%, 12-14%, 15-17%, and 18-20%. An effort was made to burn at the interval midpoints.
Once the fuel moisture was at the proper level and a decision was made to burn, the other weather variables were measured to determine if they were within the prescribed ranges as established by Schimke and Green (1970). Wind speed was measured with a hand-held anemometer. If the wind speed exceeded 10 mph, burning was postponed. Wind speeds were adjusted for differences in slope. Air temperature was required to be within the range from 20 to 84°F. The acceptable range for relative humidity as measured by a sling psychrometer was from 29 to 65%. If all these values were acceptable, the plot was ignited with a single pass of a backfiring torch using a 4:1 diesel to gasoline mixture. Backfires were ignited at the side away from the wind or at the upper side of the plot. Headfires were ignited from the windward or bottom side.
FIRE VARIABLES OF INTEREST. If a single pass did not ignite the fuels or the fire went out before it left the buffer zone, ignition was said not to have occurred.
Rate of spread was determined by measuring the time necessary for the fire to burn between two pairs of plot corner stakes in line with the fire direction. If a fire went out within a plot, the rate of spread was said to be 0. These values were averaged and converted to a rate of spread in meters per second.
The available fuel energy is the amount of energy released by the fuel volume which is burned. The total available energy for each fire was calculated by multiplying the heat yield in kilocalories per gram for each fuel layer by the weight in grams lost in that layer and adding the products for the four layers. Heat yield is determined by adjusting the various heats of combustion for losses due to radiation, heat of vaporization of the initial moisture and the water of reactions, and incomplete combustion (Byram 1959).
Fire intensity I is calculated from the equation:
I = Er
where E is the available fuel energy and r, the rate of spread. This is a measure of the energy release per unit of fire front and time and is expressed in kilocalories per second per meter. Scorch height was determined one month after a plot was burned by measuring the height to which randomly selected trees on the plots were scorched and then averaging these values. Scorch could be seen within a month by a color change of the needles which were killed. The temperature necessary to cause needle death is approximately 60°C (Davis 1959).
All "before" fuel samples were taken to the laboratory in sealed plastic bags and weighed before being oven-dried. Oven temperature was 65°C and a sample remained in the oven until there was no additional weight loss. The moisture content for the before fuel samples was obtained by subtracting the before and after oven-drying measurements. Percent and absolute reduction in fuel weight were determined by comparing the oven-dried weights for each sample pair.
After drying, the before fuels were saved for caloric determinations. Fuels from the same fuel type and fuel layer were shredded and mixed into 16 type-layer combinations. Random grab samples were taken for grinding through a 20-mesh Wiley Mill. Six pellets were made from each fuel using 2000 psi pressure and were then oven-dried. Standard procedures were followed using a Paar model 1242 bomb calorimeter (Paar 1971). Adjustments were made for the heat of combustion of the fuse wire and for the energy remaining in the free acids formed by combustion of the fuels.
The heat of combustion for each fuel was then determined with and without inorganic ash. The inorganic ash content was determined from sampled fine fuels. The samples were weighed before and after placement in a muffle furnace at 600°C for 24 hours.
Computer programs were used to compile raw data for statistical analysis. The DANIEL multiple regression program (Daniel and Wood 1971) was then used for the analysis of variance and covariance runs. Orthogonal contrasts were also run on the program by altering the X matrix to include the contrast coefficients.
The results from the F-tests for each dependent variable and the results from the orthogonal contrasts for each dependent variable were given in van Wagtendonk (1972). All tests for significance were at the 0.95 level.
Last Updated: 01-Mar-2005