Closely related to compactness but on a larger scale is con- 

 tinuity or patchiness of fuels. It represents the degree of 

 change, horizontally and vertically, in the physical character- 

 istics of fuels existing over a given area and is a measure of 

 the uniformity of continuous combustion (for a constant set of 

 weather conditions). At present, a meaningful quantitative de- 

 scription of continuity is lacking. Development of an objective 

 description of continuity would be of benefit to continued fire 

 research as well as fuel appraisal and fire control planning. 



A discontinuous fuel array exhibits abrupt changes and is a unique example of 

 nonuniformity . We employ a broad definition of nonuniformity that includes discon- 

 tinuities . 



A simple example of a nonuniform array is one in which only the depth varies. 

 The fire cannot achieve a constant rate of spread throughout the entire fuel complex, 

 but may achieve it for some uniform subunit of the larger nonuniform array. Conse- 

 quently, fire will accelerate and decelerate as it moves through the array. An ob- 

 server usually calculates the average rate of spread from the time it takes the fire 

 front to travel some given distance, but the result may not be related to either the 

 individual depths or to the average depth. 



Nonuniform fire behavior implies more than one result for the rate of spread 

 and intensity. Results should be expressed as a frequency distribution allowing the 

 user more complete information on which to base a decision. The breadth of the 

 distribution indicates the range of options to be considered in the management or 

 control of a fire. Rothermel (1974) gives general accuracy requirements for the 

 application of fire behavior models to fire management and control ranging from 

 training aids to real-time fire predictions. As a training aid the new information 

 will help emphasize the probabilistic nature of fire behavior. The impact on planning 

 and management can be profound allowing a realistic assessment of the range of effects 

 for alternative treatments of the land. The highest requirement for accuracy is pre- 

 dicting real-time fire behavior. A knowledge of fire nonuniformity at this stage is 

 essential . 



A method of collecting information from the field that represents fuel nonuni- 

 formity is not common to present, fuel inventory systems. Use of average fuel para- 

 meters in the unifonn model as an alternative produces less reliable results as the 

 fuel array becomes more nonuniform. An averaging of the fuel parameters prior to 

 processing by a fire behavior model ignores the variable nature of fire as it moves 

 through a nonuniform fuel array. A change in the fuel does not imply a proportional 

 change in the fire behavior. An improved estimate of fire behavior can be derived 

 from an analysis of the distribution of fire characteristics produced by the model 

 as the fire passes through the array. 



Both rate of spread and intensity are implied in the frequency distributions of 

 fire behavior results. Distributions of the rate of spread allow for a realistic 

 assessment of the actual range of spread rates and area growth rates essential to 

 prescribed burning and the control of wildfire. Distributions of the fire intensity 

 provide insights for an assessment of the distributions of flame lengths and crown 

 scorch heights to be expected on a site. As research into quantification of the 

 heat pulse impinging on the site progresses and is related to the intensity distri- 

 bution, assessment of the proportion of the burned area that will regenerate vege- 

 tation after a fire should be possible. Variation in regeneration is seen in the 

 response of seedlings in the vicinity of pile burning. Davis (1959) writes: 



3 



