1993 Proceedings of the Midwest Oak Savanna Conferences

AN INTRODUCTION TO THE USE OF COMPUTER GENERATED FIRE BEHAVIOR MODELS IN DETERMINING APPROPRIATE PRESCRIBED BURNING CONDITIONS

The purpose of a fire prescription is to define those conditions that will result in a fire of sufficient yet controlled intensity and to make the ecosystem changes desired. Techniques range from crumpling leaves and throwing them into the air to fuel loading transects to custom behavior models. Roethermal, Byrom, Andrews and others have developed systems to predict fire behavior. These can be applied through the BEHAVE and RXWINDOW computer programs to define a prescription in terms of fire behavior rather than a series of parameters.  

INTRODUCTION 

A prescribed fire is - A fire set in a defined geographic area, under exacting weather conditions to achieve stated resource management objectives. (Wade, no date)  

The purpose of the prescription is to define those conditions that will generate a fire hot enough to accomplish the ecosystem changes desired yet not so hot as to cause unwanted side effects or fires that cannot be safely controlled. Two methods are used to control fire intensity. First, the pattern of ignition can be varied from the use of backing to head fires. Second, proper weather conditions, prescriptive burning, can be chosen. 

DISCUSSION

The following general approaches are used when choosing prescriptive burning: 

GESTALT - Pick up some leaves, crumple them, smell the results, throw them up in the air and see how far and in what direction they fall. This has been done by Kansas ranchers for decades (Wright 1982). When used by experienced burn managers this is an effective procedure. However, the generations of experience needed for that particular site are often not available to modern restoration managers. 

SQUARE BOX (Andrews 1990) - Define a set a parameters such as temperature, relative humidity, wind speed, and number of days since last rain. As a workshop exercise we set a range for these parameters based on the experiences of the participants in burning a flat oak woodlands (model 9) area in northeastern Illinois. For each particular parameter we determined a minimum value where someone had conducted a burn and something other than drip torch mix carried fire. The maximum value was a burn where no litigation or escape fire resulted. 

Define your prescription parameters broadly so that some combinations yield an unacceptably intense fire, and some generate a fire hot enough to be able to accomplish management objectives. Or tightly define parameters to insure all combinations are appropriate. In this case the prescription often becomes so restrictive that few burning days occur. 

Consider the situation where you may have one condition out of prescription on the "hot" side (wind in the 25-30 mph range) and another out on the "cool" side (Temp, RH, and days since last rain yielding higher fuel moisture). These may not be an indication of unsuitable burn conditions. They may cancel each other out yielding acceptable fire behavior. One method to conceptualize this would be to design a 4 dimensional graph. Each dimension would be a separate prescription parameter. Within this graph would be a multidimensional oblong spheroid where the coordinates within it defined acceptable burning conditions. This would account for the interplay of all parameters affecting fire behavior. 

MODEL FIRE BEHAVIOR

The wildfire control community has sponsored research to develop methods to predict the spread and intensity of fires in wildland fuels. Roethermal, Andrews, Byrom, Albini and many others have developed a body of knowledge in predicting fire behavior. In many areas this knowledge is generally accepted as a prerequisite to functioning as a burn boss. 

There are two general approaches to fire behavior predictions. The National Fire Danger Rating System (NFDRS) and the BEHAVE computer system or Roethermal nomographs. The NFDRS was developed to predict fire suppression load over large forested tracts. NFDRS has been used successfully for prescription fires (Evenson 1986) yet it uses assumptions of an average worst case scenario. Also its end product is a manning class for suppression purposes (Burgan 1978). 

BEHAVE, or the Roethermal (Roethermal 1983) nomographs are designed to predict fire behavior on specific sites. From the results of both field studies and wind tunnel test bed fires, mathematical models were developed. The nomographs can be used to manually calculate behavior if desired. They are an excellent teaching tool to understand the BEHAVE system. BEHAVE takes the models and makes them of use in a computer environment. 

LIMITATIONS OF PREDICTION SYSTEMS

Both NFDRS and BEHAVE are of great use to prescription fire managers. BEHAVE predictions, for example integrate the interaction of the many variables which affect fire behavior. Yet, the prediction systems have the following limitations. They 

• describe a flame front advancing steadily in uniform surface fuels, with uniform site and weather conditions. This is often not the case. 
• cannot predict severe fire behavior such as crowning, fire whirls, or spotting. Nor will they predict fire spread caused by these phenomenon even though the onset of these can be inferred from fire intensity.
• assume fire spread has progressed far enough to not be directly affected by the method of ignition. For example, in drafts from adjacent strips. They predict what a point source fire would do if left alone.
• do not predict large fuel involvement or burn out.
• cannot predict depth or residence time of the flaming front although residence time can be inferred from rate of spread (Pyne 1984). 
• assume your site has the exact same fuel loading and burning characteristics as one of the 13 standard models.

There are methods to get around some of these limitations. You can design a custom fuel model using fuel loading transects to describe your site. For example, the Everglades National Park has developed a custom model to describe the jack strawed pine stands resulting from hurricane ANDREW (Passek, pers. communication). Another method is to apply a "fudge factor" based on past experience on that site or adjacent fuels (Roethermal 1983). 

USE OF THE SYSTEMS

If you determine fuel model (Anderson, no date), fuel moisture (% moisture content by weight) mid-flame wind speed (as opposed to 20 ft), slope (in degrees), wind vector (direction in relation to slope), and spread direction (back, flank or head), you can predict: 

• rate of spread or forward movement of the flaming front in chains ( 66 ft) per hour,
• fire line intensity ( BTU/FT/SEC), 
• flame length (length from ground to tip of bent flame,
• scorch (the height of heat lethal to vegetation),
• mortality in some western tree species,
• size, perimeter, and shape of the fire, and
• spot fire potential & distance. 

SETTING PRESCRIPTIONS

The RXWINDOW module of the BEHAVE program reverses the prediction process. In RXWINDOW you determine the ranges of acceptable fire behavior (constraints) and input site parameters (fuel model, slope, etc). The constraints would be developed from: 

• minimum flame length to accomplish your management objectives,
• minimum rate of spread to be able to accomplish the burn in the time allowed,
• maximum flame length based on the width of your most critical holding point, or 
• maximum flame length allowed to not damage desired species or size classes.

Your site parameters would be based on: 

• actual site conditions - slope, fuel model etc.
• minimum wind speed - In Illinois 5 mph per Illinois Environmental Protection Agency permit restrictions
• maximum wind speed - often set by agency or local standards, (The National Weather Service measures wind speed at 20 ft. above ground. Methods to adjust this to mid-flame wind speeds are found in the cited references.)
• fuel moisture - use a realistic range based on past experience. (This can be directly measured or it can be estimated by means of the SITE module of BEHAVE, nomograph from Roethermal, or charts in the S-390 text.)

RXWINDOW then determines all combinations of site parameters and fire spread direction that yield acceptable fire behavior given your constraints. Defining a prescription in terms of a range of acceptable fire behavior is often called "backing" into a prescription. Printouts will include tables showing combinations of site parameters and fire spread directions that yield acceptable fire behavior. These charts become your prescription. 

CONCLUSION

If fire managers are successful in using either gestalt or square box approaches to setting prescriptions the question arises - Why do this additional work of modeling fire behavior? 

Use of the fire behavior prediction systems can expand your burning window. It rationally explains the use of what at first might appear to be extreme conditions if viewing only one parameter. Replication of successful burns using information from different sites and times is possible and improvement can occur. The appropriate use of behavior predictions will assist restorationists toward a scientific application of fire for stated resource management purposes. We can begin to tie predicted fire behaviors to desired ecosystem changes. 

Lastly it provides a more easily articulated method for justifying and defending burn day decisions. The minimum width of fire lines can be determined using predicted fire behavior and fire control standards (National Wildfire Coordinating Group 1981). Decisions on crew size, equipment needed, and the amount of time to complete a burn can now be based on standards for control (hauling chart) and rates of spread. These methods are easier to defend than trying to articulate the gestalt methods of the past. 

REFERENCES

Anderson, Hel E. no date. Aids to determining fuel models for estimating fire behavior. U.S.D.A. Forest Service, Gen. Tech. Rep. INT-22, Intermountain Forest and Range Experiment Station, Ogden, Utah. 

Andrews, Patricia and Larry S. Bradshaw. no date. RXWINDOW: defining windows of acceptable burning conditions based on desired fire behavior, Gen. Tech, Rep. INT-273, U.S.D.A. Forest Service, Intermountain Research Station, Ogden, Utah. 

Andrews, Patricia and Larry S. Bradshaw. 1990. RXWINDOW: fire behavior program for prescribed fire planning. Fire Management Notes 51: 3, U.S. Department of Agriculture. 

Burgen, Robert E. and John D. Cohen, John E. Seeming. 1978. Manually calculating the fire danger ratings - 1978 national fire danger rating system. U.S.D.A. Forest Service General Technical Rep. INT-40, Intermountain Forest and Range Experiment Station, Ogden, Utah. 

Evenson, David. 1986. Maintenance of brush prairie in Burnette County, Wisconsin. Prescribed Burning in the Midwest Symposium, Stevens Point, Wisconsin. 

National Wildfire Coordinating Group. 1981. S-390 Intermediate fire behavior. Student Text, NFES. 

Passek, Janet. no date. Everglades National Park. 

Pyne, Steven J. 1984. Introduction to wildland fire. Wiley-Interscience Publication. 

Roethermal, Richard C. 1983. How to predict the spread and intensity of forest and range fires. General Technical Rep. INT-43, U.S.D.A. Forest Service, Intermountain Forest and Range Experiment Station, Ogden, Utah. 

Roethermal, Richard C. and George C. Rinehart. 1983. Field procedures for verification and adjustment of fire behavior predictions. General Technical Rep., INT-142, U.S.D.A. Forest Service, Intermountain Forest and Range Experiment Station, Ogden, Utah. 

Simard, Albert J. and James Eenignburg, Richard Blank. 1986. Predicting injury and mortality to trees from prescribed burning. Symposium on Prescribed Burning in the Midwest: State of the Art, University of Wisconsin-Stevens Point, Wisconsin. 

Wade, Dale D. and James D. Lundsford. A guide for prescribed fire in the southern forests. Revised Technical Publication R8-TP11. Atlanta, Georgia: U.S.D.A. Forest Service, Southern Region. 

Wright, Henry A. and Arthur W. Bailey. 1982. Fire ecology: United States and Southern Canada. Wiley-Interscience Publication. 

(RXWINDOW software is available from Forest Resources Systems Institute, 122 Helton Court, Florence, AL 35630).