Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Methods for evaluating the effects of forest fire management in Alberta Murphy, Peter John 1985

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-UBC_1985_A1 M86_5.pdf [ 10.83MB ]
JSON: 831-1.0075312.json
JSON-LD: 831-1.0075312-ld.json
RDF/XML (Pretty): 831-1.0075312-rdf.xml
RDF/JSON: 831-1.0075312-rdf.json
Turtle: 831-1.0075312-turtle.txt
N-Triples: 831-1.0075312-rdf-ntriples.txt
Original Record: 831-1.0075312-source.json
Full Text

Full Text

METHODS FOR EVALUATING THE EFFECTS OF FOREST FIRE MANAGEMENT IN ALBERTA by PETER JOHN MURPHY B.Sc.F. The University of New Brunswick, 1953 M.Sc.F. The University of Montana, 1963 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF FORESTRY We accept this thesis as conforming to the required standard  THE/U^lVERSITY^^RT*I^eOLUMBIA June 1985 ©Peter John Murphy, 1985  In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the requirements f o r an advanced degree a t the U n i v e r s i t y o f B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make it  freely  a v a i l a b l e f o r r e f e r e n c e and study.  I further  agree t h a t p e r m i s s i o n f o r e x t e n s i v e copying o f t h i s  thesis  f o r s c h o l a r l y purposes may be granted by t h e head o f my department o r by h i s o r her r e p r e s e n t a t i v e s . understood t h a t copying o r p u b l i c a t i o n o f t h i s for financial  gain  FORESTRY  The U n i v e r s i t y o f B r i t i s h 1956 Main Mall Vancouver, Canada V6T 1Y3  Date  thesis  s h a l l not be allowed without my  permission.  Department o f  It is  SEPTEMBER 30, 1985  Columbia  written  ABSTRACT Programs for the prevention and control of forest fires have evolved in response to a need to protect lives and property in forested settings, and to protect the perceived values of the forest itself. However, costs of these fire management activities have always been a concern to those who provide the funds, and considerable effort has been directed towards attempts to determine optimal levels of management effort. The question of costs has become more acute in recent years as forest services have developed increasingly sophisticated yet expensive methods for controlling fires. Compounding the problem has been an increase in frequency of fires. Determination of appropriate levels of fire control has been hampered by a lack of knowledge about the relationship between expenditure on fire control activities and the resulting area burned and losses incurred, and by an inability to describe the effect on this relationship of variations in fire season severity. This dissertation addresses these questions using the conditions in Alberta as a case study. Five hypotheses were tested and substantiated. 1.  Descriptive historical accounts of fire policy and fire seasons can be verified by analysis of actual annual expenditures on fire. Annual reports were reviewed to describe the evolution of fire management policies.  2.  There has been a decrease in area burned which is related to increased fire management effort. Analyses of age-class distribution obtained from the provincial forest inventory were used as a basis for reconstructing an estimate of historical rates of burn for the past 80 years. Cost data were obtained to try to quantify the relationship between level of expenditure and rate of burn.  3.  Variations in fire season severity can be described better than by existing methods by considering both the potential for fire spread and the actual number of fires. A new index of fire load which combined fire rate of spread with number of fires was developed which achieved this result.  4.  Potential area burned in the absence of fire control may be estimated by means of a fire growth model. A fire growth model was developed to provide a mathematical basis on which to estimate area burned in the absence of any fire management activity.  5.  There is a relationship among fire season severity, fire management effort, and area saved from  ii  burning. The combined relationships among fire season severity, fire management effort, and area burned or value were applied to illustrate some of the analyses which may be conducted with these data through evaluation of the Alberta situation.  iii  Table of Contents Chapter  Page  1.  INTRODUCTION  1  2.  REVIEW OF LITERATURE  5  2.1 Economics of forest fire control  5  3.  4.  2.2 Age-class analysis  15  2.3 Fire behaviour and fire growth modelling  17  ALBERTA - THE SETTING  20  3.1 Introduction  20  3.2 Physical description  20  3.3 Weather and climate  21  3.4 Forests and fuels  23  3.5 Forest fires  25  3.6 Forest administration  26  HISTORY OF FOREST FIRE POLICY IN ALBERTA  28  4.1 Introduction  28  4.2 Pre-European times  28  4.3 Rupert's Land - 1670 to 1870  29  4.4 Dominion administration of the North-West Territories - 1870 to 1905  30  4.4.1 Introduction  30  4.4.2 Before the railways - 1870 to 1885  31  4.4.3 The early railway and settlement years - 1885 to 1899  32  4.4.4 Advent of the Dominion Forestry Branch - 1895 to 1905  33  4.5 Alberta - Dominion administration - 1905 to 1930  34  4.5.1 Introduction  34  4.5.2 Establishing the Forest Reserves - 1905 to 1911  35  4.5.3 Building the organization - 1911 to 1918  37  4.5.4 Consolidation and transition - 1918 to 1930 iv  39  5.  6.  4.6 ALBERTA - 1930-1981  41  4.6.1 Introduction  41  4.6.2 The depression and wartime period - 1930 to 1948  41  4.6.3 Postwar development - 1948 to 1958  42  4.6.4 Growth and technical change - 1958 to 1980  44  4.7 Summary  45  METHODS  47  5.1 Introduction  47  5.2 Cost data  47  5.3 Age class data  49  5.4 Fire season severity and fire growth  51  5.4.1 Selection of the study area  51  5.4.2 Selection of the study period  55  5.4.3 Forest fire records  55  5.4.4 Weather stations  55  5.4.5 Weather data  56  5.4.6 Calculation of Fire Weather Index and Season Severity Rating  59  5.4.7 Development of the WHATIF.FIRE growth model  60  5.4.8 Model output  72  RESULTS AND DISCUSSION  74  6.1 Costs  74  6.2 Age-class analysis  79  6.2.1 Age-class  79  6.2.2 Rates of burn and fire management effort  86  6.2.3 Variation in rates of burn  88  6.2.4 Rates of burn and site  90  6.3 Fire season severity and fire growth  94 v  6.3.1 Relationship of indices to areas burned  94  6.3.2 Adjustment of potential area burned  95  6.3.3 Indicators of fire control effectiveness  103  6.3.4 Area saved from burning  106  6.3.5 Production function related to fire season severity (FLI)  Ill  6.3.6 Least-cost-plus-loss analysis  113  6.3.7 Application of a risk and uncertainty matrix  121  6.3.8 Geographic distribution of mean fire season severity  126  6.3.9 Intimations of resilience and surprise, and policy implications  128  7.  SUMMARY AND CONCLUSIONS  133  8.  FURTHER STUDIES SUGGESTED  137  9.  REFERENCES CITED  141  10.  APPENDIX  150  vi  LIST OF FIGURES Figure  Page  2.1  Total liability function (Sparhawk 1925).  illustrating point of  2.2  Transforming the cost and damage functions (Simard 1976).  3.1  Forest fire weather zones of Alberta (Simard 1973).  22  3.2  Forest regions in Alberta provincial forests (Rowe 1972).  24  3.3  Forest administration units.  27  5.1  Northern study area and Forests.  53  5.2  Northern study area showing Forest Regions and Sections (Rowe 1972).  54  5.3  Weather stations and polygons.  57  6.1  Constant-dollar costs of Forest Service expenditures not including fire suppression. Base year 1981 = 100.  75  6.2  Constant-dollar fire-related costs suppression. Base year 1981 = 100.  fire  76  6.3  Provincial age-class distribution as of 1969. Negative exponential, coefficient of determination 0.564.  82  6.4  Provincial age-class distribution as of 1909. Negative exponential, coefficient of determination 0.997.  83  6.5  Age-class distributions for Alberta for the base years 1909 to 1969 transposed to a linear format.  84  6.6  Age-class distribution for the northern forest for the base years 1909 to 1969.  85  6.7  Age-class distributions for the Forest Reserve for the base years 1909-1969.  85  6.8  Relationship between average annual burn and FME representing total presuppression and fire suppression costs.  87  6.9  Distribution of annual rates of burn for the fivemost recent 20-year age-classes on 10 Forests in Alberta .Rates of burn determined by  89  vii  of  least-cost-plus-loss  presuppression  7  9  and  percent of area in youngestage-classes. Hatched rates occurred during the period 1950-1969.  6.10  Annual variation of the Season Severity Index, and actual and potential area burned.  96  6.11  Annual variation of the Daily Spread Potential, and actual and potential area burned.  97  6.12  Annual variation of the Fireload Index, and actual and potential area burned.  98  6.13  Relationship between adjusted potential area burned and actual area burned by year.  104  6.14  Scatter plot showing relationship between adjusted potential area burned and actual area burned among individual fire years.  107  6.15  Relationship of area burned to F M E and FL1 plotted regression-derived values.'  from  112  6.16  Cost-plus-loss for the economic contribution of timber of $72 per ha, showing line of optimal solution.  118  6.17  Cost-plus-loss for the timber replacement value of $175 per ha, showing line of optimal solution.  119  6.18  Estimation of optimum F M E for various forest resource values.  120  6.19  I so-lines of 15-year mean values of Fireload Index.  127  viii  LIST OF TABLES Table  Page  5.1  Weather stations and polygon areas.  58  5.2  Daily Spread Potential (DSP) and Fireload Index (FLI) - a sample calculation.  61  5.3  Expected fire behaviour as related to season and fuel type.  69  6.1  Rates of burn derived from age-class data.  80  6.2  Age-class distribution as related to site.  91  6.3  Summary of index values and areas burned.  92  6.4  Coefficients of determination (r ) for regressions of index values with areas of burn.  93  6.5  Summary of fires by cause.  99  6.6  Fifteen-year burn area summaries and adjusted potential burn area figures.  102  6.7  Actual and adjusted potential burn areas by year, and ratio of potential to actual area.  105  6.8  Total area saved from burning in 15 years, by fire cause and size-class categories.  108  6.9  Study area fire costs in constant 1981 dollars.  110  6.10  Cost-plus-loss associated with various levels of FME andfire season severity (FLI) to illustrate minimax, minimin, and Laplace decisions at values of $72 per ha.  124  6.11  Cost-plus-loss associated with various levels of FME and fire season severity (FLI) to illustrate minimax regret at values of $72 per ha.  124  6.12  Cost-plus-loss expected with various levels of FME and fire season severity (FLI) at values of $72 per ha to illustrate application of probabilities.  125  2  ix  ACKNOWLEDGEMENTS A study of this scope could not have been completed without the assistance of a number of individuals and agencies. It was warming and gratifying to find so many willing to extend a cooperative hand whenever needed. The guidance and encouragement of my graduate committee was much appreciated, and particularly the patience of my supervisor J.H.G. Smith, assisted by J . Thirgood, D. Haley, R. Strang, and I. Fox. The study was supported by grants from the Alberta Forest Service to whom I am also indebted for the extra effort extended in gathering data and providing other services. In particular I would like to acknowledge C.B. Smith, director of the forest protection branch, and his staff for their support. Supplemental funding was provided through the Canadian Forestry Service for some of the related studies. A year of study leave from the University of Alberta made it possible to undertake this program of graduate studies in the first place. I am particularly grateful to P. Fuglem of the Forest Protection Branch of the British Columbia Ministry of Forests. Peter's early work in adapting fire growth models showed the possibility for using that technique in estimating fire potential. His subsequent consultations always proved stimulating, and his suggestions inspirational. To address the need for establishing parameters in developing the fire growth model a group of five notable and knowledgable individuals made themselves willingly available for a workshop, the results of which laid the groundwork for the WHATIF.FIRE model. The group comprised M . Alexander of the Canadian Forestry Service, H . Gray and T. Van Nest of the Alberta Forest Service, D. Quintilio of the Forest Technology School, and R. Lanoville of the Northern Affairs Program in the Northwest Territories. Their contributions were important, as were opportunities for subsequent consultation with them. Technical help was essential. B. Feddersen did an excellent job in programming the model and assembling the fire and weather data. Those were major tasks, competently handled and much appreciated. G . Armstrong and G . Merkel assisted with other computer applications and related studies. B. Wong and J . Emmanuel at U.B.C. were invaluable in the earliest stages in demystifying the computer and illustrating its capabilities.  x  Typing and revision of the manuscript was tirelessly done by L . Ehrler and J . Jacobs whose good nature and efficiency were both much admired virtues. Finally, yet foremost, the support of my family through the years of this time-consuming project was a fundamental necessity.  xi  1. INTRODUCTION Uncontrolled forest fire can be a devastating event. Large wildfires have resulted in loss of human life, destruction of communities, and mortality of timber and forest regeneration. Canadian fires such as the Miramichi in New Brunswick in 1825 (Brown and Davis 1959), the Lac-Sainte-Jean in Quebec in 1870 (Lortie 1979), in Ontario the Cochrane fire of 1911, Matheson of 1916 and Haileybury of 1922 (McClement 1969), and more recently in Alberta the Slave Lake fire in 1968 (McLean and Coulcher 1968), and the disastrous 1981 fire season throughout the province (Harvey and Janz 1982) serve as reminders of the power of high-intensity fast-moving wildfires. Some degree of fire control is essential where human values are at risk. Historically, fire control programs have developed in response to a need to protect lives and property in forested settings, and to protect the perceived values of the forest itself (Murphy 1982a, 1985). Fire policy objectives have run successively through eras of preventing fires, to stopping large fires, to keeping fire losses to below stipulated total areas based on allowable percentages, and are evolving now into integration of fire policies with land management objectives. However, costs of fire control activities have always been a concern to those who provide the funds, and considerable effort has been directed towards attempts to determine optimal levels of fire control. Forest services have developed more effective and increasingly sophisticated methods for controlling fires. Through incremental increases in budgets, improvements have been effected in the system from prevention through detection, initial attack, suppression, mop-up and rehabilitation. But the improvements have entailed a substantial financial cost, and total costs in an active season can be very high. Fire suppression costs in the Northwest Territories in 1979 were $4 million, while in 1980 costs doubled to $8 million. Fire suppression costs in Alberta were $12 million in 1980, while for the three fire seasons of 1980, 1981 and 1982 the total suppression cost was in excess of $172 million. These costs are a concern to governments and the public they represent, for there has been virtually no ceiling 1  on these expenditures, and money thus spent cannot be used for other purposes. Zaozirny voiced his concern but left the question open in a recent public comment, explaining that during those last three x  Hon. John Zaozirny, Minister of Energy and Natural Resources, Alberta. Personal communication, 29 April 1983. 1  2  years forest fires in Alberta had burned over 6 million acres (2.4 million ha) and had cost almost $200 million to suppress. Despite these high costs, there has been an understandable reluctance to scale back appreciably on expenditures for fire control since the possible consequences of such action cannot be readily assessed. Although fire is commonly seen as a destructive force it can have beneficial effects, adding further complexity to the fire problem. The degree of benefit, if any, depends on the particular ecosystem, its location and land management objectives for it, and the frequency and intensity of fires. For example, savannah-type forests benefit from periodic fires which keep them open, reduce build-up of fuels and enhance forage and browse for large mammals. In areas of low forest productivity, in remote regions, or in designated park or wilderness areas where timber harvesting is unlikely to occur, fire may also be desirable. It can serve to maintain a mosaic of vegetative types by destroying one and setting the stage for emergence of a new one. On the other hand, wildfire in timber-growing sites would not be considered desirable at all. Therefore, a balance in fire control effort must be found, with costs related to economic benefit but tempered by the characteristics of each particular ecosystem, its location, management objectives for it, and political considerations. These levels must also be tempered by the risk of fires escaping from zones of lower protection level to areas of high value. However, requisite to determination of appropriate levels of expenditure on fire control is an understanding of the relationship between such expenditure and benefits gained or damages averted, and how this relationship varies with the severity or potential of individual fire seasons. Much has been written on the concept of least-cost-plus-damage analysis as a means for determining appropriate fire control expenditure levels. However, the theory has been difficult to apply in the absence of quantifiable relationships. In particular, no adequate method for describing differences in fire season severity has been available, nor has there been an effective means for estimating area not burned or damage averted as a result of fire control activity. Several techniques were tested in this study, including analysis of cost data and age-classes, and the effectiveness of the Season Severity Index of the Canadian Fire Weather Index System. Three new  3 techniques were developed and tested - an index of potential for fire spread, a fire load index to quantify differences in fire season severity better by combining potential for burning with number of fires, and a fire growth model for estimating areas that might be burned in the absence of fire control. The following five hypothesis were examined: 1.  Descriptive historical accounts of fire policy and fire seasons can be verified by analysis of actual annual expenditures on fire. 2  2.  There has been a decrease in area burned which is related to increased fire management effort.  3.  Variations in fire season severity can be described better than by existing methods by considering both the potential for fire spread and the actual number of fires.  4.  Potential area burned in the absence of fire control may be estimated by means of a fire growth model.  5.  There is a relationship among fire season severity, fire management effort, and area saved from burning. This study was conducted through three sub-studies: 1) compilation and analysis of  fire-related costs, 2) analysis of age-class data to determine changes in annual rates of burn over time, and 3) application of fire behaviour models to describe fire season severity and estimate potential areas burned. Economic analyses were applied where fire area and cost data permitted. The study area was restricted to Alberta. The general cost and age-class analysis was applied to the entire provincial forest area for the period 1890-1969. The application of fire behaviour and fire growth models was confined to northern forest areas and analyzed for the 15 fire seasons of 1968-1982. The assessment of fire damage was done on a macro scale based on area burned, irrespective of strategic forest management implications. The thesis presents a review of literature, and describes the Alberta setting as related to forest fire concerns. A review of the history of forest fire policy presents a perspective by which to judge contemporary policy. The Methods and Results sections deal separately with the fire cost, age-class The term Fire Management Effort (FME) is used to describe the total costs involved in preventing and controlling forest fires, and includes both presuppression and suppression costs.  4 analysis, and modelling of fire behaviour and fire growth sub-studies.  2. REVIEW OF LITERATURE  2.1 Economics of forest fire control There are several criteria by which forest resource policies may be evaluated. Clawson (1975) outlined five criteria which provide a useful framework for policy analysis: 1.  Physical and biological feasibility and consequences.  2.  Economic efficiency.  3.  Economic welfare or equity.  4.  Social or cultural acceptability.  5.  Operational or administrative practicality. More specifically for fire, the planning guidelines for the USDA Forest Service (Lotan 1977)  stipulated:  "A fire management plan should be based on prescriptions that consider biological, physical, historical, climatic, and economic factors. It must reflect sound management objectives, be ecologically and economically realistic, and be professionally prescribed and carried out." Fire management policies have emerged which, on balance, appear in a general way to satisfy most of these criteria. However, the most urgent question at present is related to Clawson's second point, that of economic efficiency. This is reflected in changes in USDA Forest Service policy changes since 1975 (Gonzalez-Caban et al. 1984), in which a qualified economic efficiency criterion for evaluating fire suppression activities was included: "---suppression actions which result in the lowest cost plus net value change, having a reasonable probability of success, and providing for personal safety should be selected" (USDA Forest Service 1981). Much has been written on this subject. Martell (1978) reviewed the topic succinctly. Comprehensive review papers were prepared by Simard (1976) and Gorte and Gorte (1979), Baumgartner and Simard (1982), and Murphy (1982c) elaborated upon several aspects as well. Details may be found in those papers.  5  6 William Pearce seems to have been the first Canadian to refer to the economics of fire control. In his report to the Department of Interior in 1896 (Canada 1897) dealing with prairie fires in southern Alberta, he stated:  "An attempt is being made by the undersigned to obtain as far as possible accurate data regarding the probable loss during the past season by prairie fire. I had hoped to have had it in time for this report, but it is not available. I trust it may be shortly. The immediate object in view is this: assuming prairie fires could be largely or wholly prevented, to what extent could the cost of doing so be justified?" Unfortunately, that was the only reference to his study. It was never mentioned in any of the succeeding reports. The subject apparently was almost as intractable then as it is at present. However, the concept of that benefit-cost study was remarkably close to that of least-cost-plus-damage. Sparhawk (1925) is generally recognized as the first to describe the least-cost-plus-damage model. The basic principle was to minimize the sum of what he called total liability (suppression cost plus resource losses) and the primary protection cost of presuppression. His concept is illustrated in Figure 2.1. Gorte and Gorte (1979) also reviewed subsequent elaborations on the concept. Flint (1928) prefered to put expenditures and losses on a per-acre basis. Hornby (1936) used a modified version based on acreage burned rather than presuppression expenditure as the independent variable. He took that approach since damages were difficult to measure, a difficulty which remains. Although not stated as such by Gorte and Gorte (1979), total amount alone spent on presuppression is not an adequate measure of preparedness. Presuppression comprises a number of elements such as fuel treatment, public education and prevention, detection, and preparations to enhance initial attack and suppression such as staff training, equipment purchases, construction of access roads, purchase of vehicles and aircraft contracts. Various of these points were elaborated upon by Arnold (1949), Kun (1958), Mactavish (1965), Parks (1964) and Mills (1979) among others. Their models dealt with the macro-economic aspects of fire control for whole areas, but also lent themselves more particularly to assessing the efficiency of fire suppression activities. Annual budget statements do not at present make it possible to distinguish among presuppression activities in Alberta. The only major distinction which may be reasonably made, although not entirely, is the direct suppression cost  Fire Management Effort ( F M E )  Figure 2.1  Total liability function illustrating point of least-cost-plus-loss (Sparhawk 1925).  8  which may be estimated from among various budget codes. Thus, detailed efficiency studies are not possible with available data. This problem is not unique to Alberta, Gonzilez-Caban (1984) referred to similar difficulties in recent USDA Forest Service studies. Mactavish (1965) also recognized the importance of considering the various presuppression elements. More important, he added weighted averages of two stochastic variables - a frequency distribution of fire intensity, and the probability of concurrent fires. He suggested using an index of fire intensity which would give an indication of probable rate of spread, and which could be extended to an index of fire load by multiplying the index for fire intensity by the number of fires. He drew on experience in flood control operations by drawing an analogy between floods and forest fires. His conceptual approach appeared to be reasonably realistic, but application was hindered by lack of data. A technique incorporating fire load is developed and tested in this study, although it is related to rate of fire spread rather than to fire intensity. As Martell (1978) pointed out, the result of fire management effort is a reduction in area burned. Williams (1969) stated, "Perhaps what is really needed is a measure of what was saved not what was lost. After all, the whole fire control effort is designed to save forests". Davis (1971) in similar vein compared fire protection to insect and disease control, national defence, and city police and fire departments as well as flood control, in the sense that all had similar objectives  - to prevent something  from happening. He recognized the limitations of the  least-cost-plus-loss model for the first time, according to Gorte and Gorte (1979). Three sorts of problems were identified - estimating costs, estimating damages, and associating changes in costs with changes in damages. In his comprehensive paper dealing with the economics of policy alternatives in wildland fire management, Simard (1976) extended the discussion of least-cost-plus-loss and suggested the application of production economics to the problem. Simard elaborated upon the basic model presented by Sparhawk (1925), but developed his curves in individual quadrants, as illustrated in Figure 2.2. Quadrant II represented the production function which served to link the cost curve in quandrant III to the damage function to produce a total cost-plus-damage curve in quadrant I. The example (Figure  9  Figure 2.2  Transforming the cost and damage functions (Simard 1976).  10  2.2) was explained as follows:  "In Fig. (2.2), the optimal effort level is E*. This requires an expenditure of C*, and results in an area burned B*, and damage D*. This level of effort results in the lowest total cost T*. Increasing the level of effort above E* results in proportionately greater increases in costs than would be compensated for by decreases in damage. Conversely, decreasing the effort level below E* results in proportionately greater increases and damages than would be compensated for by decreases in cost. Thus any shift in the levels of fire management from E* will result in a nonoptimal solution." The transformation in quadrant II is the production function to which reference is made later. Simard recognized the limitations of this model stating that the arguments assumed a continuous and deterministic world with perfect information which, of course, did not exist. Another limitation was assumption of constancy in fire season severity in the major model. Simard later illustrated a two-dimensional graph reflecting fire occurrence and fire control effort with a postulated series of discrete curves, or isoquants, representing production levels for various degrees of fire occurrence. A three-dimensional graph showing the relationship of area burned (damage), fire management effort, and fire season severity would better illustrate the continuum of the relationships, and is produced in this study. Further, if the probability or risk of given levels of fire season severity could later be established, the fire planning process may be further enhanced through application of the business principles of risk and uncertainty, an example of which is also developed. A refinement of the least-cost-plus-loss model is one which recognizes that fire may also have beneficial effects. Simard (1976) discussed this in his application of marginal analysis. Mills and Bratten (1982) referred to the model as least-cost-plus net value change (C + NVC) in their development of the FEES project. The C + NVC approach was not attempted here given the more specific focus on the production function in this study and the paucity of net valuation data for Alberta. Benefit-cost analysis is closely related to the least-cost-plus-loss approach. As outlined by Gorte and Gorte (1979), the similarity lies in the concept of value protected: the two components of value protected are actual resource damages and damages averted. Benefit-cost analysis uses damages averted as its measure of benefit, to be compared to protection cost. The least-cost-plus-loss approach  11 uses actual damages as a component of the costs to be minimized. Either method, using various presuppression levels, for example, and measuring marginal changes, would give the same result. However, least-cost-plus-loss is used much more frequently in fire economics literature than benefit-cost because of the perceived difficulty of defining value protected and measuring damage averted. Given that fire has some probability of destroying or damaging resources, value protected is a viable concept. However, until its probability can be explicitly defined, value protected (and in turn damage averted) is not measurable. Actual damages are probably easier to assess than damages averted, according to Gorte and Gorte (1979), though that does not imply that measurement of actual damages is simple to do. The techniques employed in this study should assist in benefit-cost analysis by making it possible to estimate better the benefit component. Gorte and Gorte (1979) pointed out that throughout the history of fire control and fire management planning the problem of evaluating fire effects both physically and monetarily comes up repeatedly. They reviewed the work of a number of authors who suggested a variety of techniques, all with some merit, but also with inherent weaknesses. Noste and Davis (1975) stated, among many others, that there was a need for a standard appraisal system. That ideal has not yet been attained. The problem is complex because many of the resources protected have non-market values, there are both short- and long-term effects, and because it is difficult to say which resources will be threatened by fire and how much damage or benefit those resources may sustain as a result of fire. Martell (1978) reviewed appraisal of benefits of fire management in four categories; public safety, reduction of property damage, protection of timber, and protection of the esthetic benefits. Although he acknowledged that public safety was perhaps the most important benefit of forest fire management, modern fire management agencies were seen as sufficiently effective that "non-fire" personnel were seldom injured or killed by forest fires. He noted the virtual impossibility of determining how many lives would be saved as a consequence of fire action. With respect to reduction of property damage, once the saved items had been identified, replacement costs could be used to estimate the monetary value of that benefit. Estimating the value of protected timber entailed a 2-step process - first identifying stands of timber that were not burned as a consequence of fire management  12 effort, and then assigning a monetary value to the timber damage that was averted. Despite Martell's comment that most Canadian fire management agencies used what they believed were suitable damage rates to estimate annual fire losses, there is little general agreement as to their suitability. A fundamental problem remains as to where in the production flow from forest to manufactured or refined product should the value be placed, and how to handle the problems of substitution of timber supply and discounting values of immature timber. The fourth factor, the protection of the esthetic benefits, was judged to be incapable of expression in monetary terms. Marty and Barney (1981) enlarged on these points but their suggested procedures required too much detail about fire effects for practical application. Van Wagner (1981) had some fundamental concerns about the entire question of valuation. In his summary of fire economics, he stated:  "A fresh look at the economic implications of forest fire is past due. Basic concepts as well as actual data are involved. Some points that bear examining are: 1. The dubious nature of direct values assigned to burned timber. 2. The economic impact of fire on the whole forest under management. 3. The key role of substitution. 4. The prior need for evaluating the forest harvest itself before fire's economic impact can be measured. 5. The possible uselessness of attempting to assign monetary values to non-marketable or environmental forest uses or benefits, at least a purely economic analysis." The limitations of using area burned as a criterion of damage have been described, yet area remains the one variable available from all agencies and amenable to comparison. Although value figures would be more applicable to economic analysis, estimates of timber volumes may be derived from area figures, and other relationships may be established in future studies. In this study area burned is used as a surrogate measure of fire damage, while the figure of area saved from burning serves to indicate the benefits of fire management activity. This approach is taken in light of the difficulties of estimating dollar values, and in view of the focus of this paper  13 which addresses the questions of effects of fire management activity and variations in fire season severity. Van Wagner (1983) expressed reservations about the the use of area, but these are addressed in the discussion of results. Smith (1971) suggested that cost-effectiveness of fire protection expenditures needed to be evaluated. Smith reviewed several indices of efficiency of fire control in British Columbia for six periods between 1912 and 1968. Although some indicators were evident, he suggested a great deal of refinement was needed in analysis techniques, including mathematical modelling and simulation. Of note, however, was his comment that a particularly difficult aspect of any such studies was the need to live with extremely bad years, and he listed seven years in which extremely large areas were burned. He stated that in extreme years, averages may appear irrelevant. These problems of variation in severity of years and the frequency of large fires are important considerations, and are addressed in this study. Lessons might be learned from analysis of historic data compiled by forest protection agencies if we had the means to analyze them adequately. One analysis which would help immeasurably would be to know what would have happened in any fire season if no fire management effort had been applied at all. Davis (1971) recognized that problem when he stated:  "To make any comparisons at all, we need good time series data on expenditure levels and damage levels for the protected area in question. More than this, however, we need a good crystal ball The big problem is that we do not really have a good way of estimating (damage with zero organized presuppression effort)." Mills and Flowers (1983b) also referred to "---our inability to estimate the resource impact of having no fire management programs at all." Gorte and Gorte (1979) concluded by observing that it was difficult to apply any of the models they had discussed. One problem was that all of the models assumed that there is a known relationship between fire management and resource losses. In actuality, they stated, the relationship is not known; in fact, there has been some question as to the nature of the relationship. It is even possible that area burned and fire management effort are not correlated, in their view, and they referred the lack of evidence of a relationship in the literature. Literature references emphasize the problem that the results of fire management effort cannot be measured without knowing the consequences in the absence of that action. To this end a fire growth  14 model has been developed to assist with that estimate. Earlier attempts to apply the least-cost-plus-damage model were weakened by the simplistic expectation that dollars spent would result in measurable benefits, without consideration of the significance of variations in fire season severity. It is necessary to quantify these differences in order to refine the estimates of potential consequences without fire action. For this purpose an index of fire load was developed and tested. For these two latter studies detailed information on weather, fire occurrence and costs is needed. These were available for only the most recent fifteen years in Alberta. However, an indication of the general production function relationship was obtained for earlier years through age-class analysis techniques. A detailed and complex program for assessing the economic efficiency of fire management program options by the USDA Forest Service, but which contains some of the concerns of this study, has been described by Mills and Bratten (1982). Specific aspects of its development were reported by Althaus and Mills (1982), Bratten (1982), Schweitzer et al. (1982), Mills and Flowers (1983a), Flowers et al. (1983), and Bellinger et al . (1983). The Fire Economics Evaluation System (FEES) is being developed to generate at least three types of analytical information for evaluating fire management program options (Mills and Bratten 1982): dollar estimates of economic efficiency; quantitative estimates of effects on resource outputs, and assessments of the risk associated with these values. It is a simulation model, but one which will incorporate empirical data in sub-programs whenever they are available. The results will quantify many of the variables to assist the decision maker. The estimates of economic efficiency are based on the C + NVC model. Estimates of effects on resource outputs are based on the difference between the resource output without and with fires. Such assessment is based on the calculated probabilities of variations in fire season severity. There are two major differences in approach to estimating burned area in the FEES program and this study. The FEES program estimates NVC by comparing damage based on actual fires to the no-fire situation. Estimate of areas of fires which escape initial attack in the model is based on an expert gaming process through which fire behaviour experts provide probabilistic estimates of fire size, although Flowers et al. (1983) describe a  15 proposed model to simulate large fire growth. In contrast, in this study loss is based on area burned in the absence of fire control as generated by a fire growth model which indicates burn potential for all fires, not just those which escaped initial attack. Potential areas of large fires are also based on the fire growth model. The FEES program requires detailed input data including fuels and costs of the various fire management program elements. It is a very promising model which should enhance further, more specific, analyses here when it becomes fully operational, but more comprehensive records will be needed. The techniques outlined in this study lend themselves to determining the basic relationships of area burned, money spent and fire season severity, with the more general data available, and set the stage for later focus on specific aspects for more detailed study.  2.2 Age-class analysis  Forest Services prepare annual reports which list basic information about annual fire seasons including total areas burned, number of fires, and money spent on direct fire suppression costs. On first review it would appear that these would indicate the results of fire management effort through simple comparison of dollars spent to areas burned. However, the published data are lacking in three critical aspects: there is no measure of fire season severity, they do not show the fixed presuppression costs, nor can they indicate the size of the area saved from fire as indicators of effectiveness. The question of effectiveness is more than academic. As Noste and Davis (1975) pointed out, from a practical standpoint allocation of additional funds cannot be justified no matter how high the values at stake, unless the protection agency is capable of reducing damages through the increased expenditures. Van Wagner (1978) described how the age-class structure of a forest dependent on random periodic fire for disturbance and renewal assumes the form of the negative exponential distribution. An important feature of the concept is that the present age-class structure of such a forest is the key to its past fire history. In the calculation a value "p" estimates the annual probability of fire in any one stand, a figure which represents the proportion of the whole forest that burns every year. The reciprocal of the probability is the fire cycle, a figure which represents the number of years required to  16 burn over an area equal to the area of the forest. If the age-class distribution of the forest is the negative exponential the reciprocal of the fire cycle is equal to the mean age of all stands. The closeness of those two figures can also be used as an indication of the closeness of fit of the data to the negative exponential distribution. By way of illustration, Van Wagner (1978) analyzed several representative age class distributions. In boreal forest stands the age class distribution was found to follow the negative exponential distribution. In studying the data for the St. Regis Forest Management Agreement area around Hinton, Alberta, he noted the "abnormally low value in the youngest age class (1915-1960), possibly the result of increasingly efficient fire control during that period". He plotted curves for the entire age class distribution, and for a recalculated distribution as of 1915 to show the difference. His figures suggested a reduction in average annual rate of burn from 2.0 percent in 1915 to 1.5 percent in 1960. However,the average rate of burn suggested by the percentage in the youngest age class itself was about 0.25 percent per year - a significant drop. If that were sustainable, it would be possible to assess the significance from the standpoint of area saved from burning through the costs of fire management effort. It would be necessary to assume that the variation among years would average out within each of the age classes. In the St. Regis example Van Wagner (1978) estimated the age-class distribution as of 1915 by subtracting the area in the youngest age-class (from 1915-1960) and redistributing it proportionately among the remaining age classes. This roll-back procedure is useful in that it makes it possible to estimate age-class distributions at earlier times and thus assess changes in rates of burn which may have occurred. A weakness of the procedure is that as age-classes are rolled successively back in time, the redistributed ages accumulate in fewer and fewer classes. As a result, there is a tendency for the apparent age-class percentages to increase, so it is a procedure which should not be extended too far back in time. Johnson and Van Wagner (1985) reviewed the negative exponential fire model and emphasized the importance of the two stability criteria. First was that all parts of the study area must have the same fire regime, second was that each element in the study must have on average a constant fire  17 regime during the time span of the study. The first criterion was addressed in this study by distinguishing the boreal north from the subalpine southern Forest Reserve, the second by identifying changes in fire control capability during the various age classes.  2.3 Fire behaviour and fire growth modelling One of the major problems in applying analysis techniques to fire economics is that of describing the variation in fire season severity. Agency records may show numbers of fires and area burned each year, but the burned area figures have been affected by fire management effort so do not truly indicate the potential. The Canadian Forestry Service introduced Forest Fire Danger Tables for use in Alberta in the late 1950's (Canada 1956, 1959). These were modifications of the national system described by Beall (1948) and adapted to the Alberta East Slope and northern forests through studies of rates of spread in Alberta fuels. As reviewed by Van Wagner (1974) the system was refined and revised over the years and supplanted in 1970 by the Canadian Forest Fire Weather Index (CFFWI) System. This System has provided a uniform and consistent scale for rating fire danger since. Its value lies in a combination of simple daily weather readings, adaptation to local fuel types through on-site tests and observations, and versatility of application through a number of sub-codes. The system and its applicability have been well described (Lawson 1972, Van Wagner 1974, Kiil et al. 1977). Alexander (1983a) provided a succinct review. The CFFWI system is composed of six modular components. The first three of these are moisture codes that follow, on a daily basis, changes in moisture content of three classes of fuel with different rates of drying. The three moisture codes with the addition of wind speed are linked in pairs to form intermediate subindexes, the Initial Spread Index (ISI) and the Buildup Index (BUI), which represent rate of spread and fuel available for combustion, respectively. Finally, is the Fire Weather Index (FWI) representing the flame front intensity of a spreading fire. Four weather variables are recorded each day at 1300 h DST (equivalent to noon local standard time), comprising temperature, relative humidity, windspeed, and amount of rain during the previous 24 hours. The calendar date is introduced into the calculation of the Drought Moisture Code (DMC)  18  and the Drought Code (DC) to allow for variations in day length throughout the fire season. The moisture content of fuels represented by the Fine Fuel Moisture Code (FFMC) is less dependent on day length. The use of these codes makes it possible to compare burning conditions among individual days and, through the use of forecasts, to predict probable fire behaviour. In its initial application the starting Drought Code was assigned a value of 15, representing a best-estimate of DC values assuming normal overwinter precipitation and spring soil moisture recharge (Turner and Lawson 1978). However, Alexander (1982, 1983b) developed a system for adjusting starting Drought Code values based on previous fall DC values adjusted for overwinter precipitation. This was done in response to several spring fires which burned with greater intensity than anticipated. Williams (1959) developed an index of fire season severity based on the 1956 Forest Fire Danger Tables to reflect a combination of potential rate of spread and fire intensity. Williams assigned an index figure to various classes of fire severity days. The daily severity index ranged in value from 0 to 14. Fire season severity was described by simple arithmetic addition of the daily indices, divided by the number of days in the season. It was intended to provide a means by which relative severity of fire seasons in different areas and from year to year could be compared. The weakness in the system was that it did not recognize the exponential increase in fire rates of spread at the top of the 'extreme' class; furthermore it did not consider number of fires. Van Wagner and Pickett (1975) described a computer program for calculating components of the CFFWI System along with the Fire Season Severity Index (SSI) developed by Williams (1959). The SSI Index had been modified by Van Wagner (1970) to eliminate the classes, enabling the index figures to better respond to the more flammable conditions at the upper end of the scale. Continuing work by staff of the Canadian Forestry Service (Quintilio et al. 1977, Simard et al. 1911, Van Wagner 1973) has resulted in formulas for calculating rates of spread of fires in a number of fuel types based on the CFFWI indices, particularly the Initial Spread Index. The availability of those made it possible to consider development of fire-growth models based on the available weather readings and typical fuels. Smith and Gilbert (1976) had estimated rates of fire spread in British Columbia from analysis of fire reports. However, the data in Alberta fire reports were not amenable to that form of  19 analysis, nor were spread rates linked to the necessary CFFWI system components. An alternate approach was necessary. Murphy (1981) prepared a summary of computer applications to forest fire management, including models for fire growth, and Martell (1982) reviewed models more specifically related to operational research. Van Wagner (1969) illustrated how the ellipse formula could be used to approximate the area of a fire given the linear rate of spread at the head and the time since ignition. The applicability of the ellipse to fires in continuous fuels was also supported by studies in Australia (Green 1983, Green et al. 1983). The Van Wagner model lends itself well to simplified calculations. Kourtz et al. (1977) described an area-specific model for predicting perimeter location of forest fires which will be of immense value in the future. Unfortunately, for purposes of this study, input data included hourly temperatures, relative humidities, precipitation, windspeed and direction, and general fuel types entered in cells of approximately 0.5 or 2 ha. Both the data requirements and the computer costs were too great to consider for use in this study. However, the fire spread sub-routine of the AIRPRO program (Simard et al. 1977) indicated potential for use in this study, and application by 3  Fuglem demonstrated its ability. The adaptation of this program is described in the Methods section. A major research emphasis within the USDA Forest Service has been directed towards modelling fire growth for use with the National Fire Danger Rating System (Deeming et al. 1972), and with the comprehensive computer simulation model FOCUS (Fire Operational Characteristics Using Simulation) for fire management planning (Storey 1972). The substantial progress made has been described by such researchers as Anderson (1983), Rothermel (1972,1983) and Albini (1976). The need for detailed information on fuel characteristics limited the ease of application of those models to field situations. Those problems are being overcome, and application of the models and computer analysis techniques lend themselves well to basic studies of research on fire behaviour. Use of these models, or components of them, was contemplated, but in light of the differences in fuel types, differences in required weather readings, and the overriding relevance and validity of the Canadian work, the CFFWI system and its related programs were employed in this study. 3  Peter L. Fuglem, Forest Protection Division, British Columbia Ministry of Forests, Victoria. Personal communication.  3. A L B E R T A - T H E SETTING  3.1 Introduction The  study is based on conditions in the province of Alberta in western Canada. Alberta is  bounded on the south and north by latitudes 49° and 60° north, and on the east by longitude 110° west. The  western boundary comprises the continental divide along the Rocky Mountains in the southwest, 2  and longitude 120° west in the northwest. Total area is 644 400 km .  3.2 Physical description Alberta is divisible into four major physical regions (Green and Laycock 1967). The southwestern boundary lies along the Rocky Mountains with elevations up to 3650 m. Immediately to the east, and also running in a northwesterly direction is the Foothills Region consisting characteristically of long ridges with steep-sided valleys. The northeast corner, east of the Slave River and  north of lake Athabasca, comprises the edge of the Canadian Shield, characterized by a hard  rugged, rolling surface and The  numerous lakes.  majority of the province, 75 percent, lies in the Western Plains region, characterized by  level to rolling country interspersed with several hill ranges which may rise to as much as 750 m above the surrounding plains. The region is dissected by the major river valleys, some of which may be as deep as 280 m below the surrounding country. This region contains most of the arable land in the south, and the major forested areas in the north. Elevations range from about 270 m in the northeast to about 1 370 m in the high plains. The Swan Hills rise to over 1 830 m south of Lesser Slave Lake. The  northern forests are drained by four major rivers all of which are a part of the MacKenzie River  system. The largest area of these is the Peace River basin followed by the Athabasca, Hay and Slave River basins. Most of the Rocky Mountains and Foothills regions are drained by the north-easterly flowing Saskatchewan River system, with parts of the northern foothills area draining into the Athabasca and Peace Rivers.  20  21  3.3 Weather and climate Referring to Koppen's classification, Longley (1967) described the Continental climate of the forested area, lying virtually entirely within the short-cool-summer sub-zone of the cool temperate zone. Mean July temperatures range from 10°C in the north to 16°C in southern forests. Mean January temperatures, in contrast, range from -24 to -lTC in the south. The number of summer days with temperatures above 26.7C (80T) range from 15 to 25 (Longley 1968). Mean total precipitation ranges from 400 mm to over 610 mm in the foothills. Mean summer precipitation from 1 April to 30 September varies from less than 230 mm to over 400 mm at higher elevations (Longley 1968). The major fire spreads are often associated with short but catastrophic burning periods (McLean and Coulcher 1968, Murphy 1978). The major airmasses during the fire season are continental polar and marine polar, the interaction of which can lead to rapid changes and considerable variation, particularly in precipitation patterns. A generalized summary of forest fire potential was developed by Simard (1973) whose forest fire weather zones are depicted in Figure 3.1. The zones range from low in central Alberta to very high in the southwestern foothills forest. The zone boundaries were determined on the basis of average Fire Weather Index, the index representing a relative measure of expected fire intensity. The climate is influenced by latitude and altitude as well as airmasses. The Rocky Mountains 4  have a major effect on fire incidence and fire behaviour through lee cyclogenesis and dry lightning . As outlined by Gray and Janz (1983), westerly winds blowing over the mountain ranges create a favored area for pressure falls and consequently the development of cyclonic centres (cyclogenesis) in the down-wind region. These low-pressure centres tend to cause an increase in the westerly surface winds blowing in the mountainous areas. Such winds are frequently dry, increasing the fire danger. In addition, the induced low-pressure area may further generate brisk southerly winds in the eastern part of the Province. If these low-pressure areas move eastward or northeastward the eastern portions of the province may experience a wind shift from brisk southerly to brisk southwesterly. 'Lightning unaccompanied by significant precipitation  22  Figure 3.1  Forest fire weather zones of Alberta (Simard 1973).  23  Increased convective activity is also associated with lee cyclogenesis (Gray and Janz 1983). The lightning associated with convective activities often is accompanied by precipitation, but under dry conditions and in initial stages the lee cyclogenesis is accompanied by dry lightning substantially increasing the risk of new fires. Nimchuk (1983) described the impact of breakdown of established upper-level ridge circulation patterns which were associated with explosive runs taken by on-going fires in 1981. These accounted for 77 percent of total fire losses in 1981 during two periods totalling 8 days. Gray and Janz (1983) elaborated on the significance of this phenomenon, particularly when associated with periods of light precipitation. The so-called Omega Block, a strong, stationary high-pressure ridge, also exerts a profound influence on flammability and fire spread when located to the northeast of Alberta (McLean and Coulcher 1968). Such systems induce warm, southeasterly winds, a condition which contributed to the seriousness of the 1968 fire season. These examples illustrate how climatic functions can create a latent potential for small uncontrolled fires to burn large areas when conditions were right.  3.4 Forests and fuels The two major forest regions in Alberta are the boreal and subalpine (Rowe 1972). Small areas of the Montane forest region occur along the mountain valleys of the Athabasca, Upper Saskatchewan, Bow and Crowsnest Rivers and some of the most southerly foothills. The Boreal forest region comprises the greatest proportion (Figure 3.2). As outlined by Rowe (1972), white spruce (Picea glauca var. albertiana S. Brown, Sarg) and black spruce (P. mariana (Mill.), B. S. P.) are characteristic species. Other conifers are tamarack (Larix laricina (Du roi) K. Koch) balsam fir (Abies balsamea(L.) Mill.) and jack pine (Pinus banksiana Lamb.) prominent in the central and eastern portions, and alpine fir (Abies laslocarpa (Hook.) Nutt.) and lodgepole pine (Pinus contorta var. latifolia Engelm.) in the west-central and western parts. Although the forests are predominantly coniferous there is a substantial mixture of broadleaf trees, primarily trembling aspen (Populus  24  Figure 3.2  Forest regions in Alberta provincial forests (Rowe 1972).  25  tremuloides Michx.) and others including balsam poplar (P. balsamifera L.) and white birch {Bet papyrifera Marsh.). The subalpine forest region is a primarily coniferous forest found on the mountain foothills and western uplands. The characteristic species are Engelmann spruce (Picea engelmanni Parry.), alpine fir and lodgepole pine. In the small areas of Montane forest region some interior Douglas-fir  (Pseudotsuga menziesii (Mirb.) Franco) appears in addition to the species listed for the subalpine region, along with some trembling aspen and balsam poplar. The conifer forests are flammable at virtually any time of year when fuels are dry and the soil surface is free of snow. The broad-leaved forests are particularly flammable in the leafless stage from the time of snow melt to significant leaf development of herbaceous plants, and again after leaf-fall and before snowfall (from "break-up to green-up" and "leaf-fall to snow-fall"). The fuel types are characterized by horizontal continuity with relatively few natural breaks other than major mountain ranges in the southwest, the Slave River, the lower regions of the Peace and Athabasca Rivers, and a few major lakes. This continuity of fuels facilitates the development of large fires. The rate of occurrence of fires and areas burned in any one year can only be described as highly variable.  3.5 Forest fires A table of fire statistics for Alberta for the years 1918-1979 (Stocks and Barney 1981) presented figures which illustrate this characteristic of variability. A summary of the table is included in the Appendix. The total number of fires has ranged from a low of 84 in 1951 to a high of 1 758 in 1922. Lightning-caused fires have ranged from zero in 1928 to 518 in 1979, and man-caused from 77 in 1951 to 1 752 in 1922. Areas burned varied from a low of 1 824 ha in 1962 to a high of 711 603 ha in 1938. Although these figures serve to show the variation, they are not suitable for assessment of trends since methods of reporting fires, size of the protected area, and the efficiency of detection have all varied considerably since 1918. Gray and Janz (1983) used more recent statistical data to highlight the 1979-82 fire seasons, and particularly the season of 1981. During the decade of 1969-78 Alberta experienced an average of  26  675 fires per season, with a peak incidence of 906 fires in 1971, a 20-year record. During the seasons 1979-82 there was an average occurrence of 1 280 fires per season with a peak of 1 522 in 1981. Area burned during the 1969-78 decade averaged 2 500 ha per season with a peak of 60 000 ha in 1971. During 1979-82 the average was 718 000 ha burned per season with a peak of 1 365 581 ha in 1981. Their conclusion was that the early part of the 1980s' decade was the most severe fire situation since the introduction of modern forest protection in Alberta. Additionally, a new recorded overnight ignition record was set. In one evening during June 1982 there were 125 dry lightning starts spread over five Forests in northern Alberta. As they explained, the situation gave a new perspective to the multi-start situation in the Boreal forest. In addition to record starts, the fire behaviour was very severe due to drought-induced low fuel moisture content (Smith 1983), crowning occuring within three minutes of ignition in several cases. This example serves to illustrate both the need to be able to quantify fire season severity to enable comparisons between and among years, and also emphasizes the importance of considering the effect of numbers of fires as well as potential for fire spread when assessing fire season severity.  3.6 Forest administration Virtually all forested lands in Alberta are in public ownership. The National Parks are administered and protected by Parks Canada, a federal government agency. The subject of this study is the provincial forested lands, protection for which is the responsibility of the Alberta Forest Service through the Department of Energy and Natural Resources. The responsibility for forest fire management has been assigned to the Forest Protection Branch, one of six staff branches of the Forest Service. The forested area has been divided into ten Forests with staff to whom line authority for forest protection and management is delegated (Figure 3.3).  27  Figure 3.3  Forest administration units.  4. HISTORY OF FOREST FIRE POLICY IN ALBERTA  4.1 Introduction Levels of expenditure on forest fire control and degree of success in fire management effort are a reflection of fire control policy. Policy has evolved from one of essentially no control at the turn of the century to one of vigorous and sustained attack by 1980. An historical perspective is required by which to assess estimated rates of burn and expenditures in earlier years. Detailed reviews of the evolution and development of policy for control of forest fires in Alberta were prepared by Murphy (1982a, 1985). This chapter presents a synthesis of the major policies and events. Fire is a natural phenomenon on the land. It has been pervasive in Alberta since vegetation returned after the retreat of the continental glaciers. The advent of man brought about changes in the frequency of fires, increasing fire occurrence in earlier times and, more recently, decreasing fire frequency through fire control activities in most locations. At the same time, man has been affected by forest and prairie fires - either directly by fire itself, or by the subsequent ecological consequences of post-fire succession. Man's response to fire at any one time is represented by "fire policy". Policy in earlier times must generally be inferred, since there are few recorded statements of policy as such. The response of man to fire through eleven periods of history is reviewed, and a "fire policy" has been postulated for each.  4.2 Pre-European times The setting in these times was one of native Indian use of the land. That use was characterized by relatively low-intensity land-use pressures. The lifestyle was largely nomadic with relatively small groups of people travelling in response to game movements, occurrence of food plants, and seasonal considerations. Wildfires were a vicissitude, a phenomenon to be avoided by escape from its path, movement to safe areas or, when necessary, defense through backfiring. 28  29  Several anthropologists have made a convincing case that Indians, and Metis in later times, used fires in certain places and at chosen times, to create conditions more favourable to maintaining their ways of life. Policy, if it were to be stated as such, would have been to use fire seasonally as it appeared appropriate, and to avoid or fight against wildfires as necessary. On the prairie grasslands and in the parklands, the effects of deliberately-set fires could have been extensive through the continuity of  fuel in dry periods. In the northern boreal forests the effects were probably more commonly  confined to specific meadowland or parkland types, with occasional escape fires into adjacent forests.  4.3 Rupert's Land • 1670 to 1870 This 200-year span was generally a time of stability except for increasing European settlement during the last 60 years. Indian use of the land continued essentially as described previously. The economic opportunities provided by the fur trade resulted in a shift of attention to trapping of fur-bearing animals, and may have resulted in increased attention to maintaining their particular habitats such as burning to sustain early successional stages as habitat for beaver and muskrat, and care to exclude fire from mature forests which supported squirrel and marten. The  Hudson's Bay Company and the Northwest Company were very much exploitation  oriented. They operated with a profit motive and practiced virtually no management either of game or of fire, depending on exploration to provide new areas to produce fur. There was a general perception of vastness of resources. Few people had concerns about resource depletion except on a local scale, and when the buffalo herds became sparse before the mid-1880s'. The  accounts of early European explorers and travellers invariably referred to the fury and  devastation of fires, and to the dreary and melancholy waste left behind. A few individuals commented on the ecological effects of fire and post-fire succession of plants and animals, but the references depicted fire as an undesirable phenomenon. These observations reflected the prevailing European view and contributed to the single-minded policy of fire control or exclusion which prevailed for many years. A major change in attitude toward fire developed about the time of Selkirk's settlement along the Red and Assiniboia Rivers in 181.1. The threat from fire became more acute with permanent  30 settlement and the associated values-at-risk such as human habitation, structures, livestock and their required hay crops. Concerns led to the first forest and prairie legislation in western Canada, passed by the Council of Assiniboia in 1832. Emphasis in the early ordinances was clearly on prevention of man-caused fires through penalties imposed for careless ignition. Subsequent amendments reflected the persistent difficulty of enforcement. Later changes in the law also introduced requirements for fire-guarding of buildings and hay stacks - the first action-oriented approach to fuel management to prevent losses. However, there was no evidence of organized fire control at that time. Political events culminated in the Rupert's Land purchase, formation of the North-West Territories, and the beginning of organized governance by a government rather than control by a company.  4.4 Dominion administration of the North-West Territories • 1870 to 1905  4.4.1 Introduction This 35-year period was characterized by Sir John A. Macdonald's compelling thrust to extend dominion over all of British North America, including construction of a trans-Canada railway and 5  populating the new Northwest Territories (NWT). The aim of government was on development, and was reflected in the generation of Canadian nationalism, extensive land survey programs, and promotion of settlement. Administrative bureaucracies were gradually developed, first within the federal Department of the Interior, and later within a territorial administration as well. Although Macdonald did not remain in power for the entire period, the policy thrust remained essentially the same. The problems and responses in Assiniboia were characteristic of what has generally followed to the present. With the arrival of more permanently-located European settlers, the number of fires increased, and the settlement structures and crops increased the values at risk from fire. Initial policy efforts were focused on prevention, and on prosecution of offenders. Development of control effort 5  Variously termed North West Territories and North-West Territories in early references, the contemporary term is used here.  31 gradually emerged as the limitations of prevention became evident, and as people, funds, and infrastructure made control effort possible. Subsequent increases in fire control effort were usually made as recurring years of extreme fire problems demonstrated inadequacies in control capability.  4.4.2 Before the railways • 1870 to 1885 Extensive prairie fires were frequently reported even before the railways arrived, the rate of burning accelerated by increased human activities. The North-West Territories Act of 1875 provided for a new governing council of the Northwest Territories, and an ordinance for the prevention of forest and prairie fires was passed at their first session in 1877. The North West Mounted Police (NWMP) force had been formed in 1873 and soon became involved in fire control, especially in areas of settlement, but their sparse manpower greatly limited their effectiveness. The response to prairie fire problems was strong in legislation but weak in enforcement capability. However, the situation was receiving attention. A growing Canadian concern about timber supply was emerging during this period. There was also a recognized need for wood on the prairies to aid in settlement and development which led to two major policy initiatives - prevention of fires and tree planting. The Dominion Lands Act of 1872 enabled the federal government to set apart timber lands reserved from sale and settlement, authority later used to establish the first forest reserves. This act also required timber operators to prevent the ignition and spread of fires. The Department of the Interior was formed in 1876 with primary responsibility for development and management of the Northwest Territories. A Crown timber agent was located in Winnipeg in 1879 within the Timber, Mines and Grazing Branch. His responsibilities were primarily to collect timber royalties, and only later was he involved with fires. A Crown timber agent was established in Edmonton in 1882, and the following year the first two forest rangers were appointed in Alberta, one each in Edmonton and Calgary. These people were also largely concerned with collection of timber dues.  32  The report of the first Forestry Commissioner, J . H . Morgan, in 1884 recommended more attention to protection and tree planting. An amendment to the Dominion Lands Act that year permitted the Governor in Council to reserve lands on, adjacent to, or in the vicinity of the Rocky Mountains for the protection of forest trees and maintenance of water sources. These were formative years in which problems were recognized and some of the easier legislative and administrative solutions were tried.  4.4.3 The early railway and settlement years • 1885 to 1899 A persistent dichotomy in point of view emerged during this period. Opposition members of parliament repeatedly expressed concerns about the cost of administration of the NWT and the paucity of financial returns. In contrast, there were increasing pleas by Department of the Interior staff in their annual reports for additional resources to develop an administrative organization for enforcement, control, and the other fire and resource management tasks. The government had not yet developed a central agency to give specific attention to fires and problems of forest management. The Crown timber agents became involved with fires, but largely through encouragement of prevention activities and enforcement of fire laws. Dues on Crown timber were reduced for fire-killed timber to encourage its utilization before cutting green timber. The Dominion Timber Regulations were amended in 1898 to include requirements for disposal of logging debris and to require timber operators to share the costs of forest protection. Some timber reserves were established by order-in-council, although there were some political concerns about tying up lands which might otherwise be available for exploitation through patronage grants. The objective of forest reserves was to remove the best forest lands from settlement and to concentrate protection efforts on them. Prairie fires were a persistent and growing problem. The Department of the Interior first discussed the need for firebreaks, fire guardians and organized volunteer fire brigades in its annual report in 1886. That same year the Council of the NWT passed an ordinance establishing fire districts and appointing fire guardians. The North West Mounted Police were spread very thinly throughout the country, making it difficult for them to effectively enforce the fire ordinances. They reported that the  33 Justices of the Peace were usually reluctant to prosecute. Settlers evidently had a fatalistic approach to prairie fires, confounding efforts to encourage ploughing of fuelbreaks. Construction of the railway led to greatly increased difficulties through additional fire starts during the construction period and during railway operation. Court records revealed a continuing search for the proper wording of ordinances through which prosecutions and convictions of the railways could be accomplished. Irrigation in southern Alberta using water from the mountains and foothills was first mentioned in 1893. This gave impetus to the establishment of a forest reserve along the east slopes to protect the water supplies.  4.4.4 Advent of the Dominion Forestry Branch • 1895 to 1905 The persistence of fire problems aggravated by an influx of settlers, increasingly evident need for more effective federal administration of forest reserves, and possibly a change in philosophy with a change in federal government in 1896, led to establishment of the Dominion Forestry Branch in 1899 and appointment of Elihu Stewart as Chief Inspector of Timber and Forestry. Stewart was a patronage appointee, but well chosen. He was concerned about fire and began organizing a forest service. He also recognized the two persistent major programs - protection and tree planting. The philosophical dichotomy prevailed - House of Commons debates showed sustained concerns about costs and proliferation of staff, while the annual reports written by Department staff reflected the evidently inadequate resources to protect and manage the area. This applied also to the NWMP. The situation was such that Stewart included a discourse in one report to explain why forestry was a legitimate function of government. The Minister of the Interior was committed to act on the protection and tree planting problems. He indicated in 1900 that seven to eight additional people would be needed for Manitoba and the Northwest Territories to serve as fire guardians and to encourage tree planting - a rather optimistic view. By 1903 there were four fire rangers in Alberta, two each under the forest rangers in Edmonton and Calgary. Their work, as can be imagined, was largely prevention oriented.  34 There were repeated references to European forestry and forest administration as examples of what should take place in the NWT. The need for a Canadian forestry school was first discussed in 1901. Stewart also recognized the need for a forest inventory to determine the nature and extent of the forest resource. Stewart initiated formation of the Canadian Forestry Association in 1900 to enhance public support for action on forestry matters and support of forestry programs. Fire rangers were employed entirely on a casual basis, hired whenever the fire hazard increased, and released within the season whenever it abated. That practice, along with pay of $3 a day with the ranger providing his own horse, resulted in difficulties in recruitment and holding of staff. The prairie fire problem continued. Prosecutions were made under the Territorial ordinance but the pace of settlement seems to have overwhelmed the prevention and enforcement resources. With the extension of agriculture, the NWT council recognized that the new steam-powered threshing engines were a new hazard, and included regulations about them in their 1903 ordinance. Railway fires continued to be a major problem and several court appeals by railway companies successfully challenged the authority of the NWT ordinances. Increasing populations and resultant political activities led to a request by the NWT council for provincial status through one new province of Assiniboia with control of its natural resources. However, in 1905 Parliament instead established three provinces and retained control of natural resources within the federal government. This led to a form of dual administration for the next 25 years.  4.5 Alberta - Dominion administration - 1905 to 1930  4.5.1 Introduction During this time the federal government administered the natural resources including forests and minerals. The federal thrust was to focus attention on forest reserves, which represented the major high-value timber areas on lands not suited for agriculture, and to provide some degree of fire control in the northern forests. The province assumed responsibility for prairie fires in settled areas. However,  35  responsibility for fire problems in areas of active settlement fell between the federal and provincial agencies creating control problems which were slow to be resolved. Toward the end of this period efforts were designed to try to get ahead of anticipated problems. During this evolutionary time technology was increasingly applied, usually increasing both cost and effectiveness.  4.5.2 Establishing the Forest Reserves - 1905 to 1911 Greatly increased immigration and settlement along with the economic activity which it stimulated brought new pressures on the forest resources through increased fires and demands for timber and land. The still-new Dominion Forestry Branch responded to the fire and timber concern by sustained planning to try to prevent and control fires. In general, the strategy was to focus attention on the forest reserves while extending fire control activities into northern forests through patrols along the major rivers. Two major events in 1906 reflected government and public concerns. The first was the National Forestry Conference convened by Sir Wilfred Laurier. It was probably the most important gathering of its kind in Canada to that time, focusing public attention on national problems in forestry and forest protection. The second event was passage of the Forest Reserves Act which consolidated reserves previously established under Orders-in-Council, and created many new ones. It reserved the areas from settlement and provided for their administration by the Dominion Forestry Branch. The intent was to preserve the forests and conserve water supplies through forest protection. The act also provided for transfer of the timber and royalty collection functions from the Timber, Mines and Grazing Branch to the Dominion Forestry Branch on the reserves, but this was evidently not done and created a dichotomy which led to fire hazard problems through inadequate treatment of logging debris. Despite increased attention, House of Commons debates still reflected concerns about increases in staff and cost. Given this focus, the emphasis in Dominion Forestry Branch activities was on identifying and surveying new reserves, and administration of the existing ones. The surveys revealed the extent of  36 previous fire losses which lent further impetus to placing additional land under reserve status. Some interesting suggestions and developments emerged during this period. In 1906 the Forestry Branch advocated surveys of land before settlement was permitted, in order to distinguish lands of agricultural capability from those which should be left in forest reserves - both to preserve forested lands and to enhance the success of settlers. The first professional forestry program in Canada was established at the University of Toronto in 1907, followed in 1908 by one at the University of New Brunswick. The Forestry Branch began to strongly advocate the appointment of permanent staff for ranger positions and provision of technical training facilities for them. Preventive spring burning practices were described in 1909 in order to reduce fire hazards around the forest reserves. One of the Forestry Branch staff gave the first talk on forestry to school teachers in 1910, hoping thereby to reach school children with a conservation message. Grazing on the forest reserves was advocated as a fire prevention measure. Cattle reduced dry grass fuel accumulation, created trails which could serve as fuelbreaks, and the leases encouraged ranchers to become more concerned about fire control to maintain the grazing resource. Administrative problems began to emerge as the organization developed. Recruitment of rangers with qualifications appropriate to the job became difficult. The need also became evident for supporting facilities - such as lookouts, trails, communication lines, tool caches, and cabins. Forest rangers were still seasonal, although it was recognized that they should be permanent, assigned to specific districts, live nearby to the reserve and should make frequent patrols of their areas. The Forest Reserves and Parks Act of 1911 consolidated the forest reserves on the Rocky Mountain foothills, and marked the beginning of a period of buildup of fire control capability on the forest reserves. The major causes of fire were railways, settlers, campers and travellers, in that order. The prairie fire problem continued, but began to diminish in settled areas as development of roads and fuelbreaks broke up fuel continuity, and establishment of fire districts and enforcement aided in prevention and suppression. However, in areas of active settlement, frequent fire escapes into forests were a concern, but Justices of the Peace were still lenient in prosecuting fire offences.  37 Fiscal resources for fire prevention and control were provided in an incremental manner in response to demonstrated inadequacies. This was illustrated in 1909 when the number of fire rangers in Alberta was increased to 34 from 12 the previous year. The year of 1908 had been described as a "bad year". The year 1910 was also a difficult year which resulted in a further increase to 45 fire rangers in Alberta in 1911. Formation of the Commission on Conservation in 1909 was another reflection of public interest in and concern about natural resources. Chaired by the former Minister of the Interior, Clifford Sifton, the Commission conducted many studies which focused attention on problems in forest protection and forest conservation. These studies and their attendant representations assisted in increasing levels of support for forestry. In the Department of the Interior annual report for 1910, H. R. MacMillan made the following comment: "The measures adopted to protect the forests from fire are now generally understood. They are the removal by education or legislation adequately enforced of the causes of fires, the organization of a patrol to find and extinguish such fires which will inevitably start, and the improvement and organization of the forest areas so as to render most efficient the efforts of firefighters and to minimize the chances of any fires getting beyond control." MacMillan thereby captured the essence of fire control planning. Actions from that time to the present have been taken to try to achieve and refine those measures with the object of controlling fires.  4.5.3 Building the organization -1911 to 1918 Once it has been determined that a system of forest reserves represented the best course of action, activity was directed towards building the organization needed for their proper protection and administration. As before, there may have been a dichotomy in perception between the House of Commons which was still concerned about costs and proliferation of staff in government service, and the Dominion Forestry Branch which by now had selected the U.S. Forest Service as its model and pursued the "National Forest" concept with considerable zeal. Surveys for new forest reserves continued in Alberta, culminating in an extension to the Rocky Mountains Forest Reserve and establishment of the Lesser Slave Lake Forest Reserve in 1913.  38 The work of the Forestry Branch evolved into three major areas of administration. The first of these was the forest reserves. Developments there included organizing defined districts and permanent rangers on them, with supervisors in charge of the forests. Trail standards were set and stopover patrol cabins were constructed along the trails at intervals representing reasonable daily travel by horse. Training for forest rangers was still advocated, with the first formal course finally undertaken in Vancouver in 1917-18 for returning veterans. Fire prevention was still emphasized, but prevention activities in the national parks during this time showed the greatest innovative approach. Recreation in both forest reserves and national parks was tempered by fire prevention considerations. The approach was to aggregate people on safe camping areas where fuels were removed and trees pruned to prevent fires from spreading and escaping outside the designated areas, creating the first designated campgrounds. The second major area of activity was extending fire ranging in the northern forest areas outside the reserves. These comprised boat patrols on the Athabasca, Slave and Peace river systems using canoes and, later, steamers. The third work area was railway inspecting to prevent fires on operating lines, and to closely inspect lines under construction. The Grand Trunk and Canadian Northern Railways were closely monitored during their construction west from Edmonton with the result that few fires occurred. On the other hand, reports indicated that construction of the Edmonton, Dunvegan and Peace River Railway presented great problems. At a national level the Forestry Branch entered the field of research in 1917 with establishment of the Forest Products Laboratory, followed immediately after with forestry-oriented studies at the Petawawa Forest Experiment Station, studies which included work on fire behaviour. Forest research was to become a major activity of the Forestry Branch following the 1930 transfer of resources. The provincial administration during this period of duality evidently had not developed its own prairie and forest fire control capabilities to the same extent. Dominion Forestry Branch staff complained that brush disposal on provincial roads created fire hazards and that it was difficult to get cooperation to improve the situation. Dominion staff also commented that settlers in forested areas  39 berated them for not fighting fires in those areas, while in fact the province had the responsibility for settlement areas and would not appoint Forestry Branch staff as fire guardians under provincial statute. There may have been lingering resentment among provincial authorities about federal retention of natural resources. The railway fire problem persisted. New legislation was finally passed by Parliament as an amendment to the Railways Act in 1912 which empowered the Railway Commission to require companies to employ fire rangers, require maintenance of patrols on the railway lines and make the railway companies liable for damage caused by fires started by any locomotives. That legislation had also been advocated and supported by the Commission of Conservation. The new legislation at last made it possible to obtain convictions against the railways.  4.5.4 Consolidation and transition • 1918 to 1930 A plateau of sorts appeared to have been reached at this time within the Dominion Forestry Branch. The work consisted essentially of administration of the forest reserves, fire ranging outside the reserves, and railway fires prevention as outlined before. Refinements and improvements continued to be made to the essential components of the forest protection activities - including construction of roads, trails, communications lines and buildings. About 1925, when the consolidation of the forestry program appeared to be well in hand, discussions about transfer of natural resources to the provinces began in earnest. Within a few years it became evident that a transfer would take place and the Dominion Forestry Branch would lose its forest management responsibilities in the western provinces. Despite many serious reservations among Forestry Branch staff, the transition ultimately went smoothly. Major causes of fire during this time were campers, settlers and railways. The large number of fires of unknown origin reflected the relatively small staff numbers available to conduct investigations. Staffing levels in 1918 resulted in an average district size of 84 600 ha on the forest reserves, while on the Dominion lands outside the reserves the average was 486  000 ha, almost six times greater.  40  The 1919 fire year was termed "disastrous" with extensive burns reported in the forest reserves and in eastern Alberta and western Saskatchewan. The fires highlighted the need for detection, faster communication, particularly phone line construction, and for quicker access to fire areas. The use of aircraft for fire patrol - detection and scouting - began in 1920 through cooperation of the Air Board of Canada. It was deemed so successful that it was expanded and maintained through to 1930. The first lookout cabin was constructed in 1921, the same year that the first formal fire control plan was developed. That year also saw the first seasonal ranger school for training of new staff for summer work. A joint federal and provincial conference on forest fire protection was convened in January 1924 by the Minister of the Interior during which the fire problem in Canada was "exhaustively discussed". The operating department of the Canadian National Railways also held a conference on forest fire protection at which many of the provincial and federal fire protection officers were present. The national exposure led to sustained and increasing attention to the problem. A major step in Alberta-federal cooperation took place in 1921 when Alberta amended the Forest and Prairie Fire Protection Act to give Dominion Forestry Branch staff and fire rangers ex officio authority to enforce provincial legislation. Included in the amendments were strengthening of the power of fire guardians and increased penalties for infractions. Forest fire research began about 1926 within the Dominion Forestry Branch. Initial focus was on how fires started, and the relationships of fire ignition and spread to weather. Additional weather stations were established on forest reserves. A fire permit system for burning of settlers slash was introduced in 1928. During this time the Dominion Forestry Branch continued its public education activities through speaking tours. A "Save the Forest Week" was introduced in 1925, precursor to the present National Forest Week. Provincial representations led to the Transfer of Resources Act in 1930. The transfer was made effective October 1, 1930 in Alberta. The Dominion Forestry Branch shifted its activities largely to research and information gathering.  41 4.6 ALBERTA - 1930-1981  4.6.1 Introduction The young province was evidently determined to continue the forestry program as developed by the federal administration. There appeared to be a smooth transition of staff and activities from federal to provincial operation. Unfortunately, the developing depression caused severe financial problems within the province and program cuts and staff layoffs resulted. Operation of the Alberta Forest Service and the fire control organization was a struggle until the late 1940's. At that time the petroleum and natural gas industry along with a generally increased level of economic activity brought the forest resources into greater demand, and generated increased provincial revenues which made more protection and management action financially possible. The increased access as a result of this activity and greatly extended agricultural development resulted in substantially increased fire problems. The basic approach to forest fire control outlined by MacMillan in 1910 remained, but a great deal of new technology and greater sophistication in planning and administration were brought to bear.  4.6.2 The depression and wartime period - 1930 to 1948 With the transfer of resources Alberta took over the responsibility for protection and 2  administration of 50 410 km of established forest reserves and what was then known as the "Edmonton 2  Fire Ranging District" comprising 370 000 km of the northern forest. During the 1932-33 fiscal year, financial difficulties made it necessary to cut the staff to about one-third its former strength. Five permanent staff and 32 seasonal rangers were left employed on the forest reserves, while eight permanent timber inspectors and 48 seasonal rangers handled the rest of the province. Basic emphasis was focused on prevention of fires, since the ability to respond effectively to going fires was limited. In 1932 the Alberta Forest Service assumed responsibility for the Prairie Fires Act which included authority for the fire permit system. Settler fires remained a serious problem throughout this period and the next. Fire permits were initially available from many sources including honorary fire guardians, an arrangement which did not result in much control. That was changed  42  shortly to require Forest Service staff to issue permits after an inspection, although the logistics of doing so often precluded effective action. The Forest Service also worked with the railways to ensure fuel removal through burning of rights-of-way, use of good grades of coal to prevent sparks, and use of fire prevention apparatus on locomotives. Encouragement of grazing on the forest reserves and aggregation of campers into fire-resistant campgrounds were sustained as fire prevention measures as outlined before. The Canadian Forestry Association forestry and tree-planting tours continued, actively supported by the Forest Service. The first forestry radios obtained in 1938 proved to have many advantages over telephones. A start was also made on tower construction to meet the needs for a greater detection capability, and the first use of bulldozers, tractors and ploughs was mentioned. Manpower shortages as a result of World War II made use of mechanical equipment particularly important. A report of the Alberta Post-War Reconstruction Committee in 1946 signaled the postwar changes which were to so significantly affect the province. Recommendations included forest inventories, expansion of fire prevention services, reforestation, training programs for staff recruitment, and additional tree nurseries. Blefgen, the Director of Forestry, summarized these early years in his comment that "  during the depression years we were definitely informed that no money  could be made available and during the war years the necessary labour could not be secured." In 1948 a federal-provincial agreement established the Eastern Rockies Forest Conservation Board to direct increased protection and management to the Rocky Mountains Forest Reserve. In that same year the order-in-council establishing the "Green Zone" made a distinction between forested lands and lands capable of supporting agriculture throughout the province as a first step in rationalizing the patterns of settlement and to minimize the serious burning problem which still accompanied settlement.  4.6.3 Postwar development - 1948 to 1958 This was a time of rapid development and change with extensive exploration and development in the petroleum industry, accelerated rates of land clearing for agriculture, and a rapidly growing provincial economy. Increased support was gradually provided to the Alberta Forest Service to help it  43 cope with both fire and forest management problems, but the pace of industrial and agricultural development outstripped the meagre Alberta Forest Service resources. Increased help was provided, but apparently never quite enough to enable the organization to cope with the demands which it encountered. A provincial forest inventory was begun in 1949 which, by the mid-1950's, was able to show the extent of the forest resource and the effects of past fires on it. The year 1949 was a serious one for fires, with settlers accounting for 42 percent of the area burned. The following year fire permits were issued only by Forest Service staff after inspection. At this time uniforms were first provided to Forest Service staff, the first formal ranger course was held jointly with park wardens at Banff, and a substantial group of foresters were hired from the graduating class of the University of British Columbia. All forest reserves except the Rocky Mountains Forest Reserve were abolished. Some of the old restraints on forest fire fighting were gradually removed to improve and extend forest protection. In 1951 ministerial approval was required in order to hire more than one bulldozer on a fire, but that authority was passed on to the Director of Forestry the following year. In the fall of 1952 the policy of suppressing fires only within 10 miles of roads and major rivers in the north was removed, setting the stage for a greatly expanded fire control effort. Helping to accelerate the rate of change was the Rocky Mountain Section, Canadian Institute of Forestry Fire Brief in 1953 which documented the inadequacies in levels of support for fire control as reflected in high rates of burn compared to other provinces. At that time the average annual rate of burn on the forest reserve was 0.05 percent, while in the northern forests the rate was 1.1 percent. A 1953 reorganization established a Forest Protection Branch with specific responsibility for fire control. The northern forest was divided into six forest divisions for more effective administration, and a buildup in fire fighting equipment was begun. In 1955 a forest management agreement between North Western Pulp and Power Limited and the province led to construction of the first pulpmill in Alberta. The province accepted the fire control responsibility on the lease area. Serious fires in 1956, some of which burned on the lease, pointed up some still serious inadequacies in equipment and staffing levels and led to initiation of many new  44 programs.  4.6.4 Growth and technical change - 1958 to 1980 This was a time of accelerated development in Alberta for agriculture, the forest industry, petroleum and natural gas, and coal. These great changes resulted in increased problems in forest management and fire control, and led to consequent growth of the Forest Service to cope with them. Increases in numbers of fires and the increase in values-at-risk in the forest continually taxed the Forest Service resources, resulting in successive incremental increases in support after serious fire years. There was a general and sustained response to try to attain what were perceived as adequate levels of control, especially after about 30 years of impoverishment. These years were characterized by major equipment and vehicle purchases, construction of caches and ranger stations, and increases in initial attack crews. The first Forest Service aircraft for use in fire was obtained in 1957. The first defensive action on fires outside Alberta was taken in 1958 with initial attack on fires in British Columbia and Saskatchewan by Alberta crews to stop the fires before potential spread into Alberta. Increased construction and development activities led to a 1966 reorganization in the Forest Protection Branch establishing six administrative sections dealing with fire control, weather, forest fire research, communications, equipment development, and construction. Staff training and training of firefighters were emphasized. A new training centre was constructed at Hinton in 1960 with programs developed to provide technician-level training to rangers, specialized training in various aspects of forest fire control coupled with a qualification and certification scheme, and training of firefighters and staff outside the Forest Service for overhead positions. Training and education of young people were initiated through Junior Forest Wardens and a Junior Forest Ranger work program. A two-year technology program was developed through the Northern Alberta Institute of Technology to generate technicians who could be recruited into the Forest Service.  45  A forest protection area was defined in 1968 within which greater control on burning was planned by the Forest Service. A serious fire season that year highlighted the persistent problem of settler fires and underscored weaknesses in weather forecasting, prevention and initial attack. A revised Prairie and Forest Fires Protection Act in 1971 addressed many of the weaknesses. In that year a fire control policy called for control of all fires within the first burning period as an objective. By 1976 the control objective stipulated that average annual area burned was not to exceed 1/10 of 1 percent of the forest land area. The organization continued to grow, extending protection into the north through a combination of towers and aircraft patrols, ranger stations and manpower. The aircraft fleet was augmented with helicopters and contracted air tanker units. Economics was considered when certain low-value northern areas were identified as limited-action zones where fires which escaped initial attack could be allowed to burn under observation. A system of priority zoning later facilitated allocation of resources during times of heavy fire load. The fire control objective in 1980 maintained that the average annual area burned should not exceed 1/10 of 1 percent. In order to achieve that, the general control target remained the 10 o'clock. 6  rule, with specific objectives including discovery size of 1/4 acre or less, reporting time 5 minutes or less, control size three acres or less, get-away time for initial attack 15 minutes or less, and action on all fires within one hour or less. These objectives provided guidelines by which to measure operational effectiveness of the various components of the fire control system. Correction of perceived weaknesses continued to guide further build-up of fire control resources.  4.7 Summary The Alberta age-class data are available in 20-year age groupings. These are presented as of the end of 1969. The resulting 20-year classes fit closely to the periods of major development for purposes of analysis.  6  10 o'clock rule - action to ensure that a fire is brought under control by 10:00 a.m. the next day, the presumed start of the next burning period.  46  1890-1909  Time of essentially no effective fire control.  1910-1929  Federal build-up of fire control on the forest reserves and extension of fire ranging service to other forest areas.  1930-1949  Alberta administration during the period of constraint imposed by the Depression and war-time years.  1950-1969  Period of major post-war buildup of fire control capability.  The most recent 15-year period from 1968-1982 represents a time of more intensive levels of fire control and application of many technological innovations. This period is the subject of the more detailed study involving differences among individual fire years.  5. METHODS  5.1 Introduction This study comprised three major components. First, fire cost data were obtained for the period 1905 to 1982 and compared to historical developments and changes in rates of burn. Second, an analysis of age-class data was employed to estimate rates of burn for four 20-year periods from 1889 to 1969. The third major component involved a detailed analysis of annual areas burned and fire costs which were then related to various indices of fire season severity. The methods used for each of these components are described separately in this section.  5.2 Cost data Cost data for Alberta for the period 1905-06 to 1982-83 were summarized and their derivation explained by Murphy (1984) for general reference and this study. It was not possible to compile administration or fire costs for Alberta from studying the published reports of the Canada Department of the Interior for the period 1905-06 to the transfer of resources in 1930. However, a Department of the Interior financial summary (Canada 1934) located in the Alberta Provincial Archives contained figures for Alberta which appeared reasonable when compared to such data as were available in annual reports. Costs to 1930 were derived from that statement. Cost data for 1930 to 1974-75 were obtained from annual reports of the Department of Lands and Mines (1930-1948), Department of Lands and Forests (1949-1973) and Deparment of Energy and Natural Resources (1974-75). Supplementary data for clarification of certain data and capital costs were obtained from annual reports of the Provincial Treasurer. A change in budget procedures for 1975-76 required a detailed review of internal Forest Service accounts records which were summarized for this study by Chornoby and Holwein (1984). It was intended that fire-related costs be distinguished from total administration costs, and further that fire costs be categorized into two categories: presuppression costs budgeted before the advent of the fire season, and fire suppression costs representing the extra costs incurred in fighting 47  48 fires. For none of the years was it possible to clearly distinguish the presuppression costs. From 1931-32 a separate budget code identified fire suppression cost, although it appears that this figure did not include all the extra costs incurred. Especially during the most recent period of 1975-76 to 1982-83, funds from Special Warrants for direct fire suppression costs were commonly distributed among the various operational budget codes, so it was not possible to identify clearly these extra fire fighting costs because some appeared in the presuppression category. The individual fire reports contained summaries of direct costs for each of the various fires. However, in each fire season additional direct costs were incurred at both Head Office and Forest levels through such activities as manpower recruitment, operation of holding camps, and stockpiling of fuel, fire retardants, food and other supplies for use on fires generally. These general costs were not directly allocated to individual fires. For this study these global costs are apportioned both to the study area and individual fires proportionally, based on the total costs appearing on the fire reports. The summary by Murphy (1984) detailed how presuppression cost estimates were derived from the available figures. In general, an estimated percentage of the various budget codes was assigned to the presuppression category based on historical review of the major Forest Service activities at the time, and the nature of the budget codes .These estimates were reviewed with senior staff within the Alberta Forest Service to reach a concensus on percentages assigned. The portion of this study involving the more detailed analysis involving the fire growth model was applied to the northern Alberta forest which comprises 85.6 percent of the Provincial forest protection area, based on total land area excluding water and barrens. Provincial fire-related costs which could not otherwise be allocated to the study area were assigned on that same proportional basis. All annual costs were adjusted to terms of constant dollars using the Consumer Price Index (Canada 1984) with a base year 1981. The Consumer Price Index was available from the year 1914. Estimated equivalent figures for 1905-13 were extrapolated using the General Wholesale Price Index (Canada 1983) as a reference.  49  5.3 Age class data 7  Age class data were obtained through the Timber Management Branch of the Alberta Forest Service, Department of Energy and Natural Resources. The data had been collected through the Phase-3 Forest Inventory conducted during the 1970s. The year of origin of forest stands had been determined through- a combination of field sampling and estimation from aerial photography. Ages were determined only for treed areas or potentially-productive areas recently disturbed. The most recent age-classes comprise both areas burned and clearcut harvested with no distinction between the two. The data represent ages on forested or potentially-productive lands only, so do not reflect history of disturbances on non-productive lands. The years of stand origin had been compiled variously in 10-year and 20-year age classes. Since the majority of the province was represented by 20-year classes, all data were obtained on that basis. Since the data had been collected over a period of several years based on aerial photography of different dates, the request for data was programmed to stipulate that the most recent age class include the years 1950-1969. This enabled comparisons among all regions on the same basis. The age classes also coincided with the major periods of fire control policy as outlined previously. It was not possible to distinguish the cause of stand origin from the inventory data, whether from fire, logging or other disturbance. The assumption in this study was that stands originated as a result of fire. The harvesting influence was minimized in two ways. First, the base year of 1969 for age-class data avoided the accelerated harvesting rates in most recent years. In the detailed analyses, data for the Edson forest were excluded to avoid the effect of the major early extensive harvesting in that area. A further mitigating factor was the predominant form of logging before 1969 which was typically shelterwood or selection based on a diameter limit designed to leave 60 percent of the original stand as residual. The ages of these stands were taken from the residual trees which commonly reflected the original stand age. The data were compiled individually for each of the ten Forests, Athabasca, Bow-Crow, Edson, Footner Lake, Grande Prairie, Lac La Biche, Peace River, Rocky-Clearwater, Slave Lake and  7  Courtesy of D . Fregren, G . Sunderland and J . Scheffer  50  Whitecourt. They were summarized for the southern Forests of the old Rocky Mountains Forest Reserve in the subalpine forest region (Bow-Crow and Rocky-Clearwater), the northern Forests in the boreal forest region corresponding to the area selected for the fire season severity study (Athabasca, Footner Lake, Grande Prairie, Lac La Biche, Peace River, Slave Lake, and Whitecourt), and the province as a whole. The distinction between boreal and subalpine was intended to distinguish different fire regimes resulting from both ecological and historic influences. The data sheets also indicated age class distribution by the most predominant species in each type, and by site productivity class. The area 2  2  of these productive forest lands was 192 138 km for the province as a whole, 155 456 km for the 2  northern forests and 20 773 km for the old Forest Reserve. Van Wagner employed a semi-logarithmic technique using a calculator and iterations to derive s  his negative exponential factors (Van Wagner 1982 and 1984°). Since a non-linear regression program was available at the University of Alberta (PHBL: NREG) which yielded the same factors and also plotted the data and curves, it was employed instead. A trial employing both techniques based on data which simulated perfect correlation yielded the same results. It was most convenient to use the non-linear technique since the theoretical model remained the same, and use of the non-linear technique also made it possible to include the older age classes since they did not carry a disproportionately large weight in the regression as a result of logarithmic transformation. The plotted points comprised the mid point of the age class, ie. 1960 for the age class 1950-1969, and the percent of area contained within it. The probability (p) values from Van Wagner's (1978) negative exponential technique represent average values derived from the curve of age class distributions. If the rate of burn has been reduced in recent years, the indicated rate of burn (p x 100) for the curve will be higher than that reflected in the youngest age class itself. For example, in the St. Regis (Alberta) case cited earlier the p-derived rate in 1960 was 1.5 percent per year, yet the rate for the most recent age class alone (1915-60) was about 0.25 percent. Burn rates were therefore calculated as well from data in the most recent age classes for "CE. Van Wagner, Pacific For. Res. Cent., Can. For. Serv. Letter to P.J. Murphy, 12 at  1982.  'C.E. Van Wagner, Petawawa Nat. For. Inst., Can. For. Serv. Letter to G. Armstrong, 25 April 1984, file PI-4-022.  51 comparison. This point was elaborated on by Johnson and Van Wagner (1985) in which the importance of distinguishing different fire regimes was illustrated. As outlined earlier, the 20-year age classes closely fit the major periods of development of fire control and therefore represent changes thus brought about. Rates of burn in the last 20 years (1950-69) were estimated based on the area of the youngest age classes. Using the technique described by Van Wagner (1978), the age classes were adjusted successively back to the base periods of 1930-49, 1910-29, 1890-1909 and 1870-1889 to approximate the age-class distribution which might have prevailed at those times. This was done to assess the changes in rates with time and to try to estimate natural rates of burn before suppression activities were effectively started. The estimate of probability (p) in the most recent age class was derived from the following formula since the curve within the age class is a segment of the negative exponential P = ln (l-py)/20 where: p = estimated probability py = proportion of area in youngest 20-year age class Since age data were available in 20-year classes, mid-class age figures were used in the regression calculation. The value for probability (p) represents the intercept values on the y axis at the lower end of the age class. It therefore yields a higher value than that derived from the most recent age class alone. The fire cycle and rates of burn were calculated for these periods using both Van Wagner's negative exponential technique, and the area of the youngest age-class.  5.4 Fire season severity and fire growth  5.4.1 Selection of the study area The northern Alberta forest area was selected for the fire season severity and fire growth study for several reasons. It comprises the entire areas of the six northern Forests of Athabasca, Footner Lake, Grande Prairie, Lac La Biche, Peace River and Slave Lake, along with most of the Whitecourt  52 Forest with the exception of District 2 at the south end which comprises lands of probable agricultural potential (Figure 5.1). The study area lies entirely within the Boreal forest, with the exception of a minor segment in the southwest which is Subalpine. Figure 5.2 outlines the Forest Regions and Forest Sections within the study area according to Rowe (1972). Forest cover is virtually continuous, comprising stands of conifers, hardwoods, mixed wood stands all interspersed with areas of treed and open muskeg. All cover types are physically capable of supporting fire, at least at some time during the fire season. The land is essentially an extension of the Great Plains with level to rolling topography. There are few natural barriers to the spread of crown fires. 2  The gross size of the protected area is 344 734 km including water and barrens. This figure was used in calculating burned area percentages for the fire growth model in which physical barriers to fire 2  spread could not be considered. The net land area supporting vegetation is 324 180 km . The entire surficial area within the perimeter of the study area, including agricultural and other non-protected 2  lands is 365 600 km , the area containing the weather stations polygons, as described later. A smaller study area was contemplated, but the Director of Forest Protection for Alberta  10  requested that the entire northern area be covered if possible since there had been several major fire problems in that area in recent years and felt it could be more illuminating if all were encompassed. The final boundaries approximate those used in an earlier study by Harvey and Janz (1982) in which they developed a comparison of the 1980 and 1981 fire seasons. Boundaries were adjusted in a few locations to follow township and range lines to facilitate computer distinction of fires within the study area boundary. The total area figure was derived from an earlier study (Alberta 1971) and adjusted for the changed boundary using current forest inventory data (Alberta 1984).  1  "Clifford B. Smith  Figure 5.1  Northern study area and Forests.  Figure 5.2  Northern study (Rowe 1972).  area showing  Forest Regions  and Sections  55 5.4.2 Selection of the study period A 15-year period from 1968 to 1982 was selected for analysis. Although it would have been interesting to have extended the study to earlier years, there was concern about both reliability and availability of weather and fire records before 1968.  5.4.3 Forest fire records Data about each fire were obtained through the Forest Protection Branch  11  of the Alberta  Forest Service through copies of their computer tapes of fire reports. The specific data required for this study were extracted and copied to a new tape for this program. It was not possible to check all data from the tape back to original fire reports. However, the reliability of data for each fire was tested to ensure that it showed a positive growth during the time it burned. Abberations were checked and corrected. Original fire reports were checked where significant data omission was evident. A total of 7 492 fires was analysed during this 15-year period. Fires are categorized by size-class (Alberta 1985) conforming to a national convention for ease of reference and comparison. Class A  0.1 ha or less  Class B  0.2 - 4.0 ha  Class C  4.1 - 40.0 ha  Class D  40.1 - 200.0 ha  Class E  greater than 200.0 ha  5.4.4 Weather stations Harvey and Janz (1982) selected 16 Alberta Forest Service lookouts to represent weather conditions in their study. Discussion with them confirmed that, on the basis of their experience, these "Courtesy of C.B. Smith and R.S. Miyagawa  56  stations were reasonably representative of weather conditions in the study area. These weather stations are all located at Forest Service towers or lookouts and all operators had been given training in weather observing in pre-employment courses. Although located at higher elevations, they were believed to be more representative of weather conditions in forested areas than readings which might have been obtained elsewhere. Weather stations at Ranger Headquarters were considered to be of poorer quality with observations frequently made by casual observers without specific training. The weather stations operated by the Ministry of Transport provided reliable data, but the stations were located at lower elevations in more exposed sites which, they believed, resulted in over-estimates of fire hazard in the forested areas. That same criticism also applied to the data from Ranger Headquarters. Substitutions were made for two of the tower weather stations in this study, in consultation 12  13  with Harvey and Janz . The Johnson Tower used by Harvey and Janz typically closed early in the season so missed some important readings during critical years. Keane Tower immediately to the north was used instead. Yates Tower was considered to be in a more representative location than Whitesands, so it was substituted except for 1968 when Yates was not yet in operation. The study area was divided into polygons, the boundaries of which represented the mid-points between weather stations. The working polygon boundaries for the study were run on township and range lines so that entire townships were included within one of the polygons. That was done to make it possible to assign each individual fire to a specific polygon based on its legal description on the data tape. The location of the weather stations and the polygons is illustrated in Figure 5.3, and the areas of influence of each of the stations summarized in Table 5.1.  5.4.5 Weather data  The period of 169 days from 15 April to 30 September was agreed upon as a standard fire season, as discussed later. Most weather data were obtained from the Forest Protection Branch of the Alberta Forest  12  13  Dahl A. Harvey, Meteorologist, Alberta Forest Service Ben Janz, Supervisor of Weather, Alberta Forest Service  Figure 5.3  Weather stations and polygons.  58  Table 5.1 Weather stations and polygon areas  Station  Code  Adair  Area of influence (1 000 ha)  Percent of study area  AD  3 273  9.0  Bitumont  BM  2 221  6.1  Buffalo  BF  2 943  8.0  Cadotte  CU  1 852  5.1  Cowpar  CP  2 286  6.2  Edra  ED  4 943  13.5  Heart Lake  HL  1 772  4.9  Keane  K.N  2 124  5.8  Meridian  MN  3 082  8.4  Notikewin  NO  3 205  8.8  Pinto  PT  1 021  2.8  Puskwaskau  PU  1 246  3.4  Red Earth  RE  2 088  5.7  Swan Dive  SD  1 683  4.6  Tony  TY  766  2.1  Yates  YA  2 051  5.6  TOTAL  16  36 560  100.0  59  14  Service  in the form of computer tapes. The specific weather data required for this study were  extracted and copied onto a separate tape used in the programs which were developed. Data missing or in question were checked against the original telex weather messages on file. The data missing at the beginning and the end of the fire season were obtained from the nearest Canada Ministry of Transport Weather Station. Those data were purchased on computer tapes through the Ministry of Transport in Toronto. Data thus substituted were so identified in the record. In most cases only one to three weeks of record were needed at the beginning of the season. Where data were missing during the season, data were substituted from the nearest Forest Service tower weather station, and identified as replacement values. Earlier Forest Service readings were recorded in Imperial units. These were converted to metric units in the analysis program which was developed.  5.4.6 Calculation of Fire Weather Index and Season Severity Rating Van Wagner and Pickett (1975) described a computer program for calculating components of the Canadian Forest Fire Weather Index (FWI) System along with the Fire Season Severity Index developed by Williams (1959) and modified by Van Wagner (1970). A copy of that program was 15  obtained through the courtesy of M . E . Alexander . The program provided by Alexander was modified for the purposes of this study to include adjustment of the Drought Code for overwinter precipitation, to meet the data requirements of this program, and to adapt it to the University of Alberta computer system. Calculations for 1969 to 1982 were started with a Drought Code adjusted for over-winter precipitation. Weather records for 1967 and over-winter to 1968 were not available. However, starting Dought Code values for 1968 were obtained from Alexander who had compiled data for several stations in the southern part of the study area for his analysis of the 1968 fire season (Alexander 1983a). Starting values for the other weather stations were extrapolated from these. 1  "Courtesy of D . Harvey Martin E . Alexander, Fire Research Officer, Northern Forest Research Centre, Canadian Forestry Service, Edmonton. Program No. F-27. 15  60 5.4.7 Development of the WHATIF.FIRE growth model An important element which had not been addressed is that of fire load or fire season severity. If a fire season had a high flammability potential without fires, the impact would be substantially less than if many fires occurred during that time. In addition, a fire starting early during a period of high potential could have a far greater impact if uncontrolled than one starting towards the end of it. It is important that this load factor be assessed as well. To attempt to do this two new indices were developed: an alternative index of fire season severity based entirely on daily potential for fire spread; and a fire load index which incorporated number of fires. Williams' Season Severity Index (SSI) (Van Wagner and Pickett 1975) was based on daily calculations of a Daily Spread Rating (DSR) from the Fire Weather Index (DSR = 0.0272 (FWI)  1-77  )  which represents flame front intensity of a spreading fire. However, since the common measure of fire season severity is area burned, it was postulated that an index based instead on potential for linear spread would yield values more reflective of area burned than the SSI. An index of Daily Spread Potential (DSP) was developed based on the daily linear spread in metres calculated from a rate-of-spread formula for a standardized fuel type. The  16  Fireload Index  (FLI) is compiled on a daily basis. It reflects the total linear spread  potential of all fires burning on any one day, and is calculated by multiplying the number of fires by the potential daily spread. Calculation of spread of new fires considers time of ignition and hourly changes in the I SI. The load figure is additive day-by-day until weather or climate conditions would have caused extinguishment. A new compilation is begun with new fires as they occur. An example of how the DSP  and FLI are calculated is presented in Table 5.2.  In the computer calculations, all fires are recorded and treated individually. The estimate of Potential Area Burned (PAB) is calculated using the Van Wagner (1969) ellipse formula for each fire individually. The long axis comprises the total of the daily potential spread distances for the total calculated number of days of possible spread before extinguishment.  16  The FLI in this study is not to be confused with the Fireload Index described by Turner (1973) which was based on FWI and Adjusted Duff Moisture Content (ADMC) and related to records of fire occurrence and final fire size in British Columbia.  61 Table 5.2 Daily Spread Potential (DSP) and Fireload Index (FLI) - a sample calculation. Day number  Daily Spread Potential (m)  Number of new fires  Cumulative total fires  Fireload Index (spread x no. fires) 1  2  1  500  1  1  500  2  700  1  2  1 400  3  900  --  2  1 800  4  1 500  2  4  6  5  1 200  --  4  4 800  6  2 000  1  5  10 000  7  5 000  4  9  45 000  8  7 000  1  10  70 000  9  0  --  0  --  10  10  1  1  10  18 800  11  11  139 510  10-day total  Mean 3  Index  1 880  13 451  1.88  13.95  lightning and man-caused fires are distinguished in the actual compilation 2  Spread of new fires based on time of ignition and spread for first day.  3  Index = mean/1 000  000  62  The basis of the spread and area calculations was determined as follows. Van Wagner (1973) developed a rate of spread formula based on two indices of the CFFWI System. His equation was  R = a(ISI)  b  x (BUI/BUI „ f  where: R = rate of forward spread, I SI = Initial Spread Index, BUIo = a standard value of Buildup Index, and a, b and c are constants for a given fuel type. Van Wagner described the main part of the equation as comprising the I SI power term with its constants a and b. He commented that this is a very flexible formula, yielding a straight Une on ordinary graph paper when b equals 1, an up-sweeping curve when b>l, and a flattening curve when b<l. The constant "a" sets the relative level of the whole equation. He intended the BUI term, in contrast, to have a rather gentle effect. The value for BUI was set at 40, which was a roughly normal 0  value of that code for days on which fires occured in eastern Canada. For any given fuel type, he stated, the validity of the equation depended on reasonably good correlation between rate of spread and I SI. But, once the basic relation R = a(ISI)  was assumed, the only requirements were a few reliable  spread values at opposite ends of the practical I SI range. This equation was adapted by Simard et al. (1977) in the fire spread subroutine of the AIRPRO 17  program, and by Fuglem in models to predict linear spread with the I SI value adjusted by a diurnal cycle. Fuglem used the spread equation described by Van Wagner to which were applied separate formulas specific to seven different fuel models. His program simulated linear spread on an hourly basis in which the ISI was adjusted using the Simard et al. (1978) diurnal adjustments. During initial tests of Fuglem's program it became evident that there were many factors affecting rate of fire spread, parameters for which would have to be established through considered judgement, and that the  "Peter L . Fuglem, Forest Protection Division, British Columbia Ministry of Forests, Victoria. Fuglem kindly provided a computer tape of his 1983 program which was made operational on the University of Alberta computer.  63 computing cost of running the program would have to be reduced. As a result, Fuglem's program was essentially abandoned and rewritten to meet the specific needs of this study and conditions for the study area. A modified Delphi or group consensus approach was taken through a workshop comprising five 18  individuals knowledgeable about forest fire behaviour and the relationships among components of the FWI System and potential fire behaviour in the Alberta region. Ten questions were distributed to these individuals then discussed at a workshop in September 1983. Some further modifications were made as a result of subsequent tests. The major points and other assumptions used in the development of the model are described as follows. It was recognized that because of the many assumptions employed and the many variables involved in quantifying fire spread, the model could only be expected to yield a gross estimate. 1.  Basic assumptions The basic assumptions underlying this study were stipulated by Murphy (1983b). a.  Fuels in the study area are continuous so that fire spread is determined only through weather factors.  b. There are no major topographic effects such as steep slopes or physical barriers such as water or bare rock. c.  Fires burning out of the study area will be compensated for by fires burning into it.  d.  Individual fires will not overlap in area.  e.  The same areas may burn in successive years - each year being considered individually.  f.  The wind direction is constant, and the total area may be estimated through the long linear axis based on the ellipse formula.  g.  The area within each fire perimeter is completely burned.  h. The 16 weather stations are representative of the study area. 2. 18  Use of the Buildup Index  M. Alexander, H. Gray, R. Lanoville, D. Quintilio, T. Van Nest. These individuals are identified with appreciation on the Acknowledgements page and in the summary of the workshop (Murphy 1983b).  64 Van Wagner (1973) included an adjustment for the Buildup Index (BUI) in his rate of spread equation. It was agreed that this adjustment was based on arbitrary criteria and that use of this adjustment would not add substantially to the level of precision, particularly considering the other assumptions being made in this study. 3.  Application of the diurnal cycle It was agreed that it was not necessary to adjust the daily burning period according to day length as affected by latitude and season. Day length was considered of greater importance in control action than for linear spread which was the main object of this study, and would not add significantly to the precision considering the many other assumptions being made. However, fire spread during any one day had to be related to diurnal changes. The diurnal cycle values from the study by Quintilio and Anderson (1976) were recommended rather than those from the Simard et al. (1977 and 1978) AIRPRO study. The Quintilio and Anderson values were believed more applicable to northern Alberta, the geographic location of this study. In the original models used by Simard et al. (1978) and Fuglem (1983) fire spread was calculated through hourly iterations based on the diurnal cycle. Initial tests resulted in very high computing costs which would have made it financially impossible to complete the study as proposed. Instead, an adjusted mean daily value was used based on an average of the diurnal values presented by Quintilio and Anderson (1976) as related to the 1300h (MST) daily weather reading. The only exception to this was for the initial spread on the day of ignition which was calculated hourly on the basis of the diurnal cycle from the time of ignition during the first burning period. To further simplify calculation of spread, it was suggested that daily spread be assumed to take place during the 12-hour period from lOOOh to 2200h. This period represents the most common potential burning period. Extension beyond these times was believed to result only in the very marginal rates experienced at the beginning and end of the burning period. This criterion and others were tested in the model using spread data from documented fires, and results are noted  65  later.  4.  Selection of fuel types It was agreed that the model should be based essentially on the two major Alberta fuel types: hardwood (leafless aspen) and conifer (jack pine). Logging slash was recognized as a third significant fuel type, but one limited in extent. It was agreed that fires starting in slash types should be spread at the jack pine slash rate (Van Wagner 1973) during the first burning period and thereafter at the conifer rate, assuming that the fire would have run out of slash during the first burning period.  5.  Designation of fuel type Consideration was given to utilizing a generalized fuel type map prepared by the Alberta Forest Service as a basis for assigning fuel types to specific fires. However, it was agreed that the map was too general, and not adequate for predicting potential fire behaviour. Instead, the cover type indicated on the fire report would be used to stipulate the fuel type for each specific fire. It was assumed that fire would spread in the same fuel type throughout its period of potential spread. It was also assumed that the incidence of cover types on the fire reports would be representative of the availability of those fuels within each burning area. The Alberta Forest Service fire reports identified nine cover types. These are listed in the following summary along with the fuel type which was used. Mixedwood stands were to be categorized as either conifer or deciduous depending on the first species listed in the cover type symbol.  Cover type on fire report  Spruce Pine Deciduous Muskeg Bog Brush Grass Recent burn Clear cut  Designated fuel-type for fire-growth model  Jack pine Jack pine Leafless aspen Jack pine Jack pine Leafless aspen Leafless aspen Jack pine Jack pine slash (first day) then jack pine  66 6.  Fuel type rate-of-spread equations In all three equations the basic formula described by Van Wagner (1973) was used except that the BUI adjustment was not included. The basic element is the initial spread index (ISI). a.  For the aspen fuel type the spread formula described by Van Wagner (1973) for leafless aspen was selected. This is the most predominant fuel type which supports spring fires. 1 01  Rate of spread (m/min.) = 0.249(ISI) ' b.  For the conifer fuel types the spread formula for jack pine described by Quintilio et al. (1977) derived in the Darwin Lake study was selected. This formula was the most recent available and related specifically to Alberta. 1  78  Rate of spread (m/min.) = 0.0454(ISI)" c.  Rate of spread in logging slash was based on the jack pine slash formula described by Van Wagner (1973). 1 00  Rate of spread (m/min.) = 1.34(ISI) '  Mixed wood stands were categorized as either conifer or aspen on the basis of the species listed first in the forest cover type symbol. 19  Revised formulas have been printed in an interim form by Alexander et al. (1984) are not yet available for use since they are still under review by the authors. 7.  Fire season length As outlined earlier, it was decided that the fire season would be started 15 April and ended 30 September, a period of 169 days. Those dates include the annual period of major fire activity. Use of a period of consistent length facilitated comparison among fire seasons.  8.  Threshold values for spread and extinguishment The threshold values for fire growth and extinguishment are dependent on fuel type. For aspen fuel types, fires would be expected to spread at Fine Fuel Moisture Content (FFMC) values above 78, to smolder between 77 and 65, and to extinguish at values below 65. Expected fire  "Alexander, M.E., B.D. Lawson, B.J. Stocks and C.E. Van Wagner. 1984. User guide to the Canadian forest fire behaviour prediction system: rate of spread realtionships. Interim edition July 1984. Can. For. Serv. 76 p.  67 behaviour in aspen fuel types is summarized in the following tabulation. Expected fire behaviour FFMC  Leafless aspen  78 +  Fire spread per formula.  77 - 65  Smoldering fire - no spread. Subsequent spread or extinguishment dependent on indices on the following days.  64-  Fire extinguished.  Fires in conifer fuel types were judged dependent on the Duff Moisture Code (DMC) and the Drought Code (DC). Values of DC 100 or DMC 20 reflect threshold values between extinguishment and spread. Expected fire behaviour is summarized below. DC  DMC  Expected fire behaviour Conifer  100+  20+  Fire spread per formula  99- or  19-  Fire extinguished. No spread.  In the workshop discussions it was agreed that if only Drought Code or Duff Moisture Code were below the threshold value with the other still above that the fire would smoulder without spreading. However, in the initial test runs it was found that the Drought Code essentially remained above 100 after May so all conifer fires would have remained alive and potentially active for the fire season. As a consequence, it was decided that fires would be extinguished if either of the two factors dropped below the threshold values. Conifer fires were also extinguished if the ISI dropped to zero. For the leafless aspen fuel model the WHATIF.FIRE was programmed so that when FFMC fell in the smouldering range of 65-77, all ISI's except those with 0 values were assigned a value of 1 to ensure the fires remained in the smouldering state but did not spread until FFMC and  68 ISI conditions permitted further spread or caused extinguishment. 9.  Seasonal changes in fuel type There are significant changes in fuel conditions between spring and summer and between summer and fall. Specific dates for change-over between seasons were used. June 15 was selected as the date marking the beginning of summer conditions, a time substantiated by Kiil et al. (1977). September 15 was selected for the start of fall conditions. Although there is some variation with latitude, elevation and season, the difficulty of determining the appropriate date at any one point eliminated that option. The workshop participants agreed these stipulated change-over dates were reasonably representative of actual conditions. Fires commonly spread in leafless aspen at the beginning of the fire season until green-up. It was decided that fires in aspen fuel-types would be spread by the leafless aspen formula until 14 June when it would be assumed that green-up had occured, and fires would be self-extinguishing. Fires reported in aspen fuel types from 15 June to 14 September were assumed incapable of growing, and were not included in fire growth calculations. Following the presumed date of leaf-fall on 15 September fires were again assumed capable of spreading in aspen fuel types. Fires in all conifer types were spread by the jack pine formula irrespective of season. The relationship between season, fuel, and expected fire behaviour is summarized in Table 5.3.  10. Calculation of linear spread and area Estimates of areas of individual fires were calculated on the basis of the ellipse formula described by Van Wagner (1969). The total potential linear spread distance over the successive days of potential fire growth was taken as the long axis. The Van Wagner simplified area formula for an ellipse where the backing rate of spread is negligible is: 2  2  Area = v t x [ /(4(l/w))] w  where: v = linear spreadat the head t = time since ignition, l/w=length-to-width ratio.  69 Table 5.3 Expected fire behaviour as related to season and fuel type.  Season  Fuel Type  Expected Fire Behaviour  Spring  Aspen  Fire spread by leafless aspen formula until 14 June when fire self-extinguished  (15 April - 14 June)  through green-up. Conifer  Fire spread by jack pine formula irrespective of season change.  Summer  Aspen  Fire may ignite but not spread.  Conifer  Fire spread by jack pine formula  (15 June — 14 September)  irrespective of season change. Fall  Aspen  (15 - 30 September)  Fire spread by leafless aspen formula until extinguished 30 September.  Conifer  Fire spread by jack pine formula until extinguished 30 September.  70 Estimation of potential burned area of each fire was to be based on the total cumulative potential linear spread distance over the estimated duration of the fire, with areas calculated by the simple ellipse formula (Van Wagner 1969). For this estimation of potential linear spread and area burned the workshop discussions were unable to resolve the questions of how to specifically apply the criteria of diurnal variation in I SI, length of the daily burning period, and how to determine an appropriate length-to-width ratio for fires. To search for solutions to these problems, combinations of data for six well-documented fires were simulated in the model. The fires, or portions of fires, employed were all free-burning and unaffected by suppression action. Combinations of three variables were tested: I SI, length-to-width ratio, and length of burning period. Four conditions of I SI were tested: 1) unadjusted noon I SI reading, 2) I SI adjusted by the average diurnal value of 0.7, and 3) I SI adjusted by the 12-hour 1000-2200 h (time) average value of 0.8, the latter two values from Quintilio and Anderson (1976), and 4) ISI adjusted by an arbitrarily selected figure of 0.9. The test results indicated the 12-hour value of 0.8 yielded the closest results. The workshop discussions suggested that a frequently-occurring length-to-width ratio for free-burning fires was 2:1. However, it was recommended that other combinations be tested. Anderson (1983) illustrated how the length-to-width ratio of the ellipse varied with wind speed. A difficulty with fires burning more than one day was determining a method of selecting an appropriate wind speed. An earlier test using average daily maximum wind speed did not appear satisfactory. The results suggested that it may be more appropriate to use the ratio indicated for the maximum wind speed experienced during the potential life of the fire, assuming that the maximum wind event would have had the major influencing effect on fire shape. Anderson's formula for this calculation is as follows. 1/w = 0.936EXP(0.1147U) + 0.461EXP(-0.0692U) where: 1/w = length-to-width ratio U = windspeed at 1.5 ft. or mid-flame miles per hour.  71 Two drawbacks to Anderson's formula were manifested in early tests. The windspeed data available from weather records were recorded at the standard 10 m height whereas the Anderson formula necessitated an estimated adjustment to mid-flame height. Further, the results yielded 20  length-to-width ratio figures as great as 11:1. Alexander observed that few fires have numerical values exceeding 6:1, and it was agreed that the long-narrow shapes predicted in Anderson's formula were uncharacteristic. An alternative length-to-width formula in Imperial units had been developed by Simard and Young (1978) for the AIRPRO study. 1 2  W/L = exp(-0.0287 W ' )+'(0.00312 sW)W where: W/L = width to length ratio (converted to L/W by reciprocal). W = wind speed at 10 m height, mi/hr. sW = standard deviation of wind direction, degrees. The direction variability factor, estimated by the standard deviation of the wind direction 21  (sW), used a constant factor of 10° without a great deal of justification. Alexander agreed that the use of this factor was open to questions since it undoubtedly varied with wind speed, yet did not add substantially to the estimate. It was therefore left out of the calculations. The formula, converted to the length-to-width ratio in metric units, became: 1 2  L/W = l/exp(-0.0787*0.62137 W ' ) The Simard and Young (1978) formula eliminated both drawbacks with respect to weather readings and ratios. Both formulas were tried in the test runs of area calculations. The Simard and Young formula yielded the closer results. The third variable tested for linear spread and area calculation was length of burning period. The 12-hour period suggested by the workshop and a 24-hour period were selected. The 24-hour results were closer in the test runs.  20 21  M.E. Alexander - Personal communication. M.E. Alexander - Personal communication.  72  The following criteria were therefore employed in the model for calculation of area. a.  ISI adjusted by the 12-hour (1000-2200h) average of 0.8 from the diurnal cycle, presented by Quintilio and Anderson (1976),  b.  length-to-width ratio based on maximum windspeed and the Simard and Young (1978) formula,  c.  spread over a 24-hour potential burning period. The total estimated area of the six fires tested in the model was just 6% higher than the actual total of 427 000 ha. The closeness of this estimate was encouraging, given the many assumptions and generalizations, and in light of the unknowns of fire behaviour as outlined by Albini (1984).  11. Fires naturally extinguished The fire reports listed many fires showing a zero area and with no direct fire suppression costs 22  attached to them. It was agreed that these represented fires which would likely have gone out by themselves but which were only inspected and confirmed as fires by staff whose expenses would have been covered by the fixed presuppression budget. These fires were not included in the Fireload Index or growth model calculations since the fires would not likely have developed further. The annotated computer program for the model of potential fire growth WHATIF.FIRE is included in the Appendix.  5.4.8 Model output The model calculates three indicators of fire potential for each fire season: 1) an index of spread potential (DSP), 2) a fire load index (FLI), and 3) an estimate of potential area burned in the absence of fire control activities. The index of Daily Spread Potential (DSP) is based on a daily calculation of potential linear fire spread during each day of the fire season. The leafless aspen fuel type was used until 15 June, and the jack pine fuel type from 16 June to the end of the season. Daily spread was calculated using the 22  In consultation with H.W. Gray, Head of Operations, Forest Protection Branch, Alberta Forest Service.  73  model factors of 0.8 I SI and 24-hour spread. A index of Seasonal Spread Potential was calculated for each station, representing the total of the estimates of daily linear spread throughout the season. The seasonal spread potential for the study area represents the mean of all station values weighted by their proportional areas. The program for Fireload Index (FLI) calculates the spread potential for each fire for the number of fires possibly burning on each day. For each station-day the number of fires as determined from the fire reports which may still be active is multiplied by the daily spread potential for the station, and added to the potential spread for each new fire estimated on the basis of time of ignition. These daily totals (daily FLI) are added to compile a seasonal index for each station. The seasonal fire load potential for the study area is represented by the weighted average of the seasonal fire indices for each of the stations. The Potential Area Burned represents an estimate of total area burned at each station calculated according to the total cumulative linear spread of each individual fire at the time of probable extinguishment, with individual areas calculated by the ellipse as outlined earlier. The seasonal figure represents the total Potential Area Burned figure for each station. The program also summarizes for each station year the number of days in the longest duration during which fires could spread, and compiles a frequency distribution of the length of potential burning periods.  6. RESULTS AND DISCUSSION  6.1 Costs A summary of all forestry administration and protection costs for the fiscal years from 1905-06 to 1982-83 is included in the Appendix. Bar charts illustrating Forest Service costs, and combined fire presuppression and fire suppression costs are presented in Figures 6.1 and 6.2. All costs are in terms of constant 1981 dollars adjusted by the CP I. Cost figures for the Alberta Forest Service represent only the budgeted operational codes, not including the variable fire fighting costs. The exception is the period from 1905-06 to 1929-30 during which time it was not possible to isolate fire suppression costs within the cost summaries. The figures reflect the slow build-up of the Dominion Forest Service during its infancy, the major growth appearing in 1911-12 after the Forest Reserves and Parks Act of 1911. The small fluctuations in the budgets undoubtedly reflect variations in fire seasons, as outlined later, but a fairly consistent level of support was evident during the 1920's. The first major change occurred between the years 1931-32 and 1932-33. During the last full year of federal administration in 1929-30, the adjusted Forest Service budget was $2.12 million. The $1.78 million spent during the first full year of provincial administration in 1931-32 contrasted quite favorably to this, reflecting the stated Alberta provincial government determination to maintain the federal level of forest management. However, the level declined 46 percent over the following two years. Annual expenditures were reduced to $1.07 and $0.95 million respectively, reflecting the severe cuts made in response to the Depression. In fact, it was 16 years later, in 1947-48, before the constant dollar expenditure exceeded that of 1931-32, and it was a full 20 years, in 1949-50, before expenditures exceeded the previous peak of $2.12 million. The fairly steady level of funding to 1948-49, although it shows some increase, clearly reflects Blefgen's comment (Alberta 1947) that during the depression years there was not money available to do the work, and during the war years there was not the manpower available.  74  7 5  11010090-  •p> 80oo o»  5  70-  C  .2  •g  60-  S  50H  "g  40-  «>  Q.  X  UJ  302010 H  1905  1915  1925  1935  1945  1955  1965  1975  Fiscal (fire) year  Figure 6.1  Constant-dollar (1981) costs of Forest Service expenditures not including suppression .Base year 1981 = 100.  1985  76  Figure 6.2  Constant-dollar (1981) fire-related costs of presuppression and suppression .Base year 1981=100.  77 As outlined by Murphy (1982a), 1948 marked a turning point brought about by post-war reconstruction, need to protect non-forest values, veterans returning to settle new areas for agricultural development, and the beginning of the first Alberta forest inventory. What made this increase and all subsequent budget increases possible was the over-riding impact of increased provincial revenues generated by petroleum and natural gas activity. These meant that for the first time, the government of Alberta could afford to do some of the things which had been so long advocated. The initial surge of expenditures to 1955-56 largely reflected the building of the Forest Service to extend the administrative structure and services to northern areas, and to build strength in locales where activities generated by mineral exploration and development .settlement, and timber harvesting made it necessary. 23  Northwestern Pulp and Power Limited signed the first forest management agreement and began building its mill in 1955. Under the impetus of increased focus on timber values combined with the severe fire season of 1956 another surge in development became evident from 1957-58 to 1962-63. Increased concern with timber management, more focus on forest renewal, and concerns over integrated land use during the time of heightened environmental awareness during the 1970's brought about the next surge in expenditures from 1973-74 to 1975-76. Expenditures in the last three years largely reflect responses to those problem fire years coupled with expanding programs in forest management and renewal. These expenditure levels and trends closely reflect the written descriptions in the annual reports, providing an interesting perspective against which they may be reviewed. The chart illustrating adjusted total fire costs (Figure 6.2) distinguishes estimated presuppression costs from actual fire suppression costs from 1931-32. That distinction was not possible during the earlier years, but some observations may be made regardless. Fire-related expenditures began to increase in 1910-11, coincident with protecting the new Forest Reserves. The year 1910 was a serious one throughout the west which undoubtedly helped to support increased expenditures over the next few years. The fire years 1913 and 1914 are an interesting contrast. The annual report referred to 1913 as a very satisfactory year with respect to fires, with favourable weather keeping losses to a minimum. The following year 1914 was described as the worst since 1910. However, the adjusted expenditure levels for 23  Now St. Regis (Alberta) Ltd.  78those two years show very little difference. This suggests that the Forest Service at that time did not have a great capacity to incur extra expenses in fire suppression since most action involved manpower and there was not a great deal of it readily available. Similarly, the year 1917 was described as the most serious since 1914, and the years 1919 and 1920 both experienced extensive burns - yet those expenditure years do not show significant increases either. Some of the peak years begin to show up more prominently after 1930. The fire year of 1931 was described as possibly the most difficult one experienced by any forest protective organization in Alberta. During its first full year of administration the government undoubtedly tried to do the right thing including hiring from the increasing number of unemployed. The years 1935 and 1936 present another contrast - 1935 described as 'outstanding' through favourable distribution of rainfall, whereas 1936 was a serious one again. However, the relatively low suppression expenditures in 1936 reflect the restraint which evidently prevailed during those times of reduced funding, as did the succeeding fire years of 1938, 1941,1944, 1949, and 1950. The level of presuppression funding began to increase during the late 1940's, as explained earlier, and appeared to level off in 1961, despite the major fire suppression expenditures in that year. The peak fire years of 1956, 1958, 1959, and 1961 can be clearly identified. The year 1958 was notable since fire fighting costs exceeded $1 million (unadjusted) for the first time. Costs exceeded $2 million in 1961. The advent of high-cost aircraft use was a major contributing factor, while increasing provincial revenues made their use possible. No outstanding increases in level of presuppression funding were evident again until 1981 following the major problems encountered in the 1980 fire year. Other significant fire years were 1968 when expenditures exceeded $5 million (unadjusted) for the first time, 1974 when the $6 million level was reached, and 1979 when fire fighting costs rose to over $12 million. The quantum increases in fire fighting costs during 1980, 1981, and 1982 were in response to a combination of increased numbers of fires and severe burning conditions. The presuppression increase in 1981-82 is clearly evident, designed primarily to enhance initial attack capability. Gray and Janz (1983) described how unique climatic factors conspired to set the scene for those three years, representing an anomaly not generally encountered during the 1960's and 1970's, the period which they  79  had analyzed.  6.2 Age-class analysis  6.2.1 Age-class Results of the calculations based on age-classes are summarized in Table 6.1. Separate summaries are presented for the province as a whole, the northern forests excluding the Edson forest, and the old forest reserve area. The probability (p) represents an estimate of the probability of any one point burning in any one year and, multiplied by 100, is the estimated average annual rate of burn expressed as percent of area. The fire cycle is the reciprocal of p, representing the time in years during which an area equivalent to the total area would be expected to burn. The mean age was determined from the original age-class data. The two rates of burn were determined respectively from p x 100 derived from the negative exponential, and from the proportion of the area in the youngest age-class. Van Wagner (1978) pointed out that if the areas of the youngest age-classes are reduced through fire control, the plotted points for the age-classes thus affected will appear as low anomalies compared to the negative exponential curve. By adjusting age class distributions successively back in time those anomalies are removed and a better fit of the curve to the actual values is reflected in two 2  ways. The first is an increased coefficient of determination (r ). The second is a closer relationship between mean age and the fire cycle (1/p) which should be equal with perfect correlation. This is 2  illustrated in the results for the province as a whole which show a coefficient of determination (r ) of 0.564 for the 1969 age-class distribution, with a fire cycle of 95 compared to a mean age of 63 years. However, with ages adjusted to the year 1909, the coefficient of determination increased to 0.977 while the fire cycle and mean ages were 35 and 32 years respectively, suggesting substantially less human intervention through fire control at that time. For the age class base year 1909 the factor p is an estimate of average annual rate of burn in that forest complex up to that period of 1890-1909. However, in more recent years where human intervention begins to reduce the rate of burn a more indicative rate of average annual burn is derived  80 Table 6.1 Rates of burn derived from age-class data Age-class base year Region and variables  1969  1949  1929  1909  0.0106 0.564 95 63 1.06 0.26  0.0176 0.936 57 46 1.76 1.66  0.0200 0.916 50 40 2.00 1.70  0.0285 0.977 35 32 2.85 2.66  0.0111 0.560 90 60 1.11 0.26  0.0196 0.966 51 43 1.96 1.94  0.0207 0.938 48 38 2.07 1.88  0.0263 0.945 38 31 2.63 2.47  0.0083 0.370 121 79 0.83 0.13  0.0121 0.618 82 61 1.21 0.70  0.0175 0.762 57 49 1.75 1.17  0.0334 0.985 30 39 3.34 3.23  PROVINCE (19.2 million ha) probability (p) correlation (r ) fire cycle (1/p) mean age burn % (p x 100) burn % (age class basis) NORTHERN FOREST 2  (15.5 million ha) probability (p) correlation (r ) fire cycle (1/p) mean age burn % (pxlOO) burn % (age class basis) FOREST RESERVE 2  (2.1 million ha) probability (p) correlation (r ) fire cycle (1/p) mean age burn % (pxlOO) burn % (age class basis) 2  81 from the percentage of area in the youngest age class alone as described earlier and as cautioned by Johnson and Van Wagner (1985). For example, for the province as a whole in 1969, 4.99 percent of the area is found in the youngest 20 year age-class - representing an average annual burn of 0.26 percent. The average annual rate of burn suggested by the negative exponential distribution is 1.06 percent, but that curve-derived figure is influenced by the higher rates of burn under more natural conditions during earlier times. If the rate of burn could be maintained at 0.26 percent through human intervention, the p value from the negative exponential would gradually approach the 0.26 percent level. However, the percentage derived from p will still be higher than the figure derived from the most recent age class, as explained previously, since the p value represents the intercept at the lower limit of the age class, while the figure derived from the age-class alone represents the mid-point of the age class. This is reflected in the provincial data for 1909 where the p intercept value is 2.85 percent at the lower class limit of 1890, while the mid-point age class value is 2.66 percent. Plotted relationships for the provincial data for 1969 and 1909 are presented in Figures 6.3 and 6.4. The effect of the substantially smaller youngest age-class in 1950-69 is clearly evident in the 1969 distribution. The closer fit of values to the negative exponential in 1909 is quite apparent. The change in age-class regressions over time is illustrated in Figure 6.5 for the provincial data. These straight-line regressions were transposed from the non-linear form on a semi-logrithmic plot. Indicated rates of burn show a progressive decline from 1909 to 1969. It is interesting to compare the transposed linear regressions for the northern forest areas and the Forest Reserve (Figures 6.6 and 6.7). The regressions for these two areas also reflect historic fire policy. The greatest change in rate of burn on the Forest Reserve area occurred between 1909 and 1929 during which time the Dominion Forest Service focused attention there. A substantial change also took place in the northern forest during that same interval, but not of the same magnitude. Two partially offsetting forces were probably at work - a combination of extension of 'fire ranging' into the north offset by increased settler-caused fires. The major change in the northern forest took place during the 1949-60 interval, a time coincident with Alberta Forest Service efforts to extend and intensify its fire control effort in that area as reviewed earlier. The slope of the 1969 regression for the Forest Reserve is  82  4S-,  40 H  Age (years)  Figure 6.3  Provincial age-class distribution as of 1969. Negative exponential coefficient of determination 0.564.  Age (years)  Figure 6.4  Provincial age-class distribution as of 1909. Negative exponential coefficient of determination 0.977.  Figure 6.5  Provincial age-class distributions for the four 20-year base years 1909-69 transposed to a linear format.  85  Figure 6.6  Age-class distributions for the northern forest for the base years 1909 to 1969.  100-3 H 0  , 100  1  200  1  ^  300  400  Age (years)  Figure 6.7  ft  Age-class distributions for the Forest Reserve for the base years 1909 to 1969.  86  flatter than that for the northern forests reflecting its more favourable position with respect to more effective fire control action. These relationships are also reflected in the figures presented in Table 6.1 in which the coefficient of determination shows the greatest change since 1949 in the Northern forest, but shows an earlier and more extended period of change on the Forest Reserve.  6.2.2 Rates of burn and fire management effort By combining the cost data with figures for rates of burn obtained through analysis of age-class data it is possible to plot a relationship between the two. The data in summary form for the province as a whole are as follows (the term F M E refers to the combined fire management effort of pre-suppression and fire suppression expenditures).  Year  Average Annual Burn (percent)  Pre-1909 1910-1929 1930-1949 1950-1969  2.85 1.70 1.66 0.26  Average Annual FME Smillion (constant)  0.001 1.217 1.646 14.359  The cost data are in constant dollar values. The pre-1909 value was given a virtual zero value since cost data from 1899 were not available, and annual reports indicate relatively little effective fire control was undertaken in Alberta before 1909. These figures represent the estimated total spent on both presuppression and suppression activities. The relationship is illustrated in Figure 6.8, coefficient of determination 0.975. The curve was compiled and plotted using the same non-linear regression program (PHBL:NREG) which was employed for the analysis of age-class data. This curve is similar to the 'total liability' curve described by Sparhawk (Figure 2.1) although his comprised suppression cost plus losses. It is more closely described by the 'production function' outlined by Simard (1976) (Figure 2.2) although the shape of this one is a negative exponential while Simard postulated a sigmoid relationship. However, Simard's curve suggests a negative exponential  87  3-i  A  •\ 2.5  -\ :\  \ 2-  c o <D  \  CL  AU  c  1.5-  00  \  "6 c c  <  1-  \  \  0.5-  A  0.0  5.0  10.0  Fire Management Effort ($ million)  Figure 6.8  Relationship between average annual burn and F M E representing total pre-suppression and suppression costs.  15.0  88  relationship at the lower end. Given the great immediate potential for reduction of area burned through simple initial attack on accessible fires, which is a relatively low-cost activity, it would appear that the negative exponential reasonably describes the production function in the fire management field. The predicted values at the lower end of Figure 6.8 suggest that the average annual rate of burn should have been more in the nature of 0.017 percent, whereas the actual value was 0.26 percent. A basic assumption in presenting this curve is that the fire load during the various periods in question has remained relatively constant within each of the 20-year periods. This apparent anomaly in the curve is probably a reflection of an increasing fire load during the most recent period from 1950-1969. Had the fire load remained at the levels previously experienced, the actual value may have been closer to the predicted value. It is also possible that a curve of some configuration other than the negative exponential may be more appropriate. Assuming for sake of analysis that the curve is reasonably correct, it is possible to derive a cost per hectare saved from burning. The pre-1909 rate of burn of 2.85 percent per year indicates an annual burn area of 547 200 ha per year. The reduced rate of 0.26 percent per year shows losses reduced to 49 920 ha per year, suggesting that for an average annual cost of $14,359 million, 497 280 ha was saved from burning. The cost per hectare saved from burning was $28.88 in 1981 terms. The implication of this value is discussed in a later section.  6.2.3 Variation in rates of burn Average annual rates of burn were compiled for five 20-year periods in ten Alberta forests, not including the Edson forest..The distribution of these percentages among the five periods and the ten forests is illustrated in Figure 6.9. The rates vary from 0.25 percent to 3 percent per year with most falling between 1.25 and 2.25 percent. The median rate is 1.50. All of the rates in the 0.25 category fall within the most recent period from 1950 to 1969. If these are disregarded, the distribution resembles a normal curve with a mode of 2 percent and median of 1.75 percent.  0.25  0.50  0.75  t  1.25  1.50  1.75  2  2.25  2.50  2.75  Average Annual Burn (%)  Figure 6.9  Distribution of -annual rates of burn for the five most recent 20-year age-classes on 10 Forests in Alberta.Rates of burn determined by percent of area in youngest age-classes. Hatched rates occurred during the period 1950-1969.  3  90  This variation substantiates the point made by Van Wagner (1978) about the importance of considering the size of area in drawing conclusions from age-class analysis, given the potential for large fires and therefore catastrophic influence on small areas. The size of the areas analyzed here should be large enough to avoid this problem. These figures represent 20-year means, so do not reflect the even higher variation in rates which could be derived through an analysis of annual fire losses.  6.2.4 Rates of burn and site Smith (1981) applied Van Wagner's (1978) negative exponential technique to age-class data for British Columbia to try to determine whether there was an influence of site on rates of burn. His analysis for both spruce and lodgepole pine in Interior bioclimatic zones showed greater rates of burn on the better sites as reflected in mean age. Since the age-class data for Alberta were also available by site a similar analysis was run. The results are summarized in Table 6.2. No such trend was evident for the Alberta provincial data as of the 1969 base year. When the age-classes were adjusted to 1909, where correlation was greater, it appeared that the good sites may be experiencing a higher rate of burn than on the fair sites. No such trend was evident in the northern forests which are primarily boreal. However, data for the Forest Reserve, which is predominantly lodgepole pine in a subalpine region, and therefore more related to the Smith data, indicate that the rate of burn on good sites is substantially greater than on fair sites, and that the trend is progressive through medium sites. That result agrees in trend with the data for the Interior of British Columbia presented by Smith (1981). The result suggests a relationship worthy of further study. If the good sites in Alberta were primarily valley-bottom locations the reason for higher rates of burn may be exposure to greater risk of ignition through human activity. However there may also be bioclimatic factors at work affecting flammability.  Table 6.2 Age-class distribution as related to site Area  Site  Variable  Base year 1969  1909  ALBERTA Province  Northern Forest  Forest Reserve  Good  Mean age Fire cycle  68 107  31 37  Medium  Mean age Fire cycle  60 91  28 33  Fair  Mean age Fire cycle  70 100  43 43  Good  Mean age Fire cycle  71 112  34 44  Medium  Mean age Fire cycle  59 88  29 36  Fair  Mean age Fire cycle  59 86  36 41  Good  Mean age Fire cycle  58 . 93  19 17  Medium  Mean age Fire cycle  73 119  26 26  Fair  Mean age Fire cycle  100 152  62 58  BRITISH COLUMBIA  Spruce  Lodgepole  Interior  Good  Mean age  151  91  (Smith 1981)  Medium  Mean age  168  105  Fair  Mean age  169  113  92 Table 6.3 Summary of index values and areas burned Year  Season Severity Rating  1968  Daily Spread Potential  Fireload Index (xl 000)  Potential Area ha (xl 000)  Actual Area ha (xl 000)  1.80  884  343  4 203  321  1969  2.53  1 013  274  4 503  26  1970  2.60  1 434  463  4 184  52  1971  3.24  1 236  749  19 565  61  1972  2.80  1 223  657  12 097  48  1973  2.10  1 022  213  3 345  10  1974  1.45  715  108  1 286  15  1975  1.37  703  136  1 097  5  1976  0.04  245  350  11 709  20  1977  1.71  696  215  3 341  9  1978  1.51  629  171  1 637  7  1979  1.77  795  650  12 184  190  1980  3.88  1 153  2 093  91 805  664  1981  3.12  1 105  1 072  19 237  1 349  1982  2.08  728  1 044  20 537  671  93 J  Table 6.4 Coefficients of determination (r ) for regressions of index values with areas of burn.  Potential Area Burned  Actual Area Burned  Index  Area  Log (n) area  Area  Log (n) area  Season Severity Index Season Severity (log n) Daily Spread Potential Daily Spread (log n)  0.356  0.330 0.015  0.197  0.276 0.095  0.070  0.097 0.043  0.018  0.111 0.105  Fireload Index Fire Load (log n)  0.876  0.812 0.922  0.693  0.621 0.746  Potential Area Burned Potental Area (log n)  0.518 0.502  0.623  94 6.3 Fire season severity and fire growth  6.3.1 Relationship of indices to areas burned A total of 9 390 fires was reported within the study area during the 15-year period. Of these, 1 898 were not spread in the model because they did not fall within the criteria previously described. Fire spread and estimates of area burned were calculated for the remaining 7 492 fires. The results of these and the other index calculations are summarized in Table 6.3. Coefficient of determination between various combinations of the index values and area figures were calculated using the *SPSSX statistical package for regression analysis available at the University of Alberta. A summary of those results (Table 6.4) illustrates the coefficients of determination for the various indices to the area-burned figures. The two variables to be predicted are actual area burned and potential area burned. A distinguishing feature of both is that neither represents an absolute base for comparison. The actual area burned is affected by fire management effort, the degree and effectiveness of which vary from fire to fire and from year to year. The potential area burned is an estimate, with no precise means yet available to determine its actual magnitude. However, despite those limitations, it is important to obtain as close a prediction as possible. The Season Severity Index (SSI) is the one associated with the Canadian Forest Fire Weather 2  Index System. Correlation (coefficient of determination r ) with actual area burned was very low at only 0.197 with the arithmetic value. It showed only weak correlation to potential area burned with a correlation of 0.356. Harrington et al. (1983) found stronger correlations with this index at four stations in Alberta based on analysis of monthly areas burned and monthly index averages for the period May - August 1953-80. Their mean values ranged from 0.36 and 0.37 at Slave Lake and Whitecourt respectively, to 0.45 at Fort McMurray in northern Alberta. The highest correlation was at Rocky Mountain House, 0.61. As Harrington et al. pointed out the correlations would probably have been higher if the analyses could be done on a more area-specific and time-specific basis. It is probable that the SSI correlations in this study would also have been higher than indicated if done on more than  95 a seasonal basis. However, all indices tested in this study were based on the same seasonal basis and may be so compared. The index of daily spread (DSP) proved to be of no value as a predictor of area burned. It is evident that potential for spread alone is not a sufficient indicator. The SSI which combines potential for spread with fire intensity did better. The Fireload index (FLI) showed a strong correlation with actual area burned, with a coefficient of determination of 0.693. The correlation was considerably higher for potential area burned with 0.876. This strong relationship of FLI to potential area burned could be expected since both were calculated on a similar basis. However, of the three indices tested, the FLI also shows the strongest relationship to Actual Area Burned. The figure for potential area burned shows a coefficient of determination with actual area burned of 0.518 for arithmetic values. It suggests that potential area burned could also be used as an index for predicting actual area burned, although the FLI appears stronger. The plotted relationships of the two area figures with each of the three indices are presented in Figures 6.10, 6.11 and 6.12. Of the eight assumptions listed earlier on which the fire growth model was based the major weakness is probably the one of overlap. Maximum individual fire sizes generated in the model did not seem unreasonable compared to two large historic free-burning fires. The 1950 Chinchaga fire (Murphy 1978) burned 1.16 million ha with a long axis of 260 km. The 1919 fire in Alberta and Saskatchewan (Murphy 1983a) may have covered as much as 5 million ha with axes of 460 and 220 km. The largest fire generated in the model had predicted values of 2.9 million ha with linear spread of 591 km.  6.3.2 Adjustment of potential area burned Although the model-generated figure for total potential area burned provides an index to actual area burned, it is obvious that the figure as a measure of potential area itself is too high. The summary of fires by cause (Table 6.5) illustrates this. The potential average annual rate of burn is 40.7 percent. If this were a result primarily of increased human activity it might be explained, but the rate for fires  96  100000?  Legend ;A S « a » o n S « y « r i t y lnd«x x WOO  lOOOOd Actual A r a Bunwd  1000 0)  lOOd.  tod  11965  —I  1970  1  1975  1980  1985  Year  Figure 6.10  Annual variation of the Season Severity Index, and actual and potential area burned.  97  lOOOOOq  Legend 10000-  A  Dally Sprtod PottnHd  •  Actual Area Burned  / /  \ \  lOOOd  > lOOd  10J  "1 1965  1  1  1  1  1970  1975  1980  1985  Year  Figure 6.11  Annual variation of the Daily Spread Potential and actual and potential area burned.  Figure 6.12  Annual variation of the Fireload Index and actual and potential area burned.  Table  6.5  Summary  of f i r e s  by c a u s e  Area  Number  of F i r e s  Burned  Actual  (1 0 0 0 h a )  Area  Potential  Area  Year  Lightning  Man  Total  1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982  58 179 233 340 279 65 167 162 155 92 223 356 512 756 600  262 155 205 220 193 145 92 197 301 156 154 124 439 353 319  320 334 438 560 472 210 259 359 456 248 377 480 951 1 109 919  0 .2 24 .9 44 .4 56 .3 46 .5 0 .5 14 .0 4 .0 0 .6 2 .6 5 .8 181 .6 296 . 1 1 344 .3 667 .8  321 . 1 1. 1 7 .8 5. 1 1 .9 9 .6 0. 7 1 .3 19 .4 6 .6 1 .5 8. 7 367 .5 4. 5 3 .0  321 .3 26 .0 52 .2 61 .4 48 .4 10 . 1 14 .7 5. 3 20 .0 9 .2 7 .3 190 .3 663 .6 1 348 .8 670 .8  1 1 1 1 17 12 12  67 . 3 959 . 9 024 . 7 895 .7 778 . 9 712 .8 668 .O 467 .6 533 . 7 160 . 2 215 . 3 587 . 3 122 ..0 850. . 3 445 ..8  177  3 315  7 492  2 689 .6  759 .8  3 449 .4  79  489 , 5  1 312 958 .. 2  500  179 .3  50 .6  229 .9  5 299 . 3  8 753 . 2  14 052 . 5  0.55  0. 16  0.71  15 . 4  25 . 4  40 . 8  4  Total  279  Mean Average % Burn'  221  LIghtning  Annual  Man  Total  r Base  Study  A r e a : Net=32 4 18 0 0 0 h a  G r o s s = 3 4 473 OOO  ' A c t u a l a r e a p e r c e n t b a s e d on n e t a r e a P o t e n t i a l a r e a p e r c e n t b a s e d on g r o s s a r e a  ha  Man  Lightning  2 2 10 4  4 1 2 8 7 2  1 1 2  74 6 8  135 .. 3 543 . 3 216 .0 669 . 3 318, . 2 632 . 2 6 18 . 4 629 .8 175 . 3 180 . 4 422 . 1 597 ..0 682 .. 9 386 .9 091 . 1  Total 4 4 4 19 12 3 1 1 1 1 3 1 12 91 19 20 210  202 . 6 503 2 240. . 7 465 .0 097 .. 1 345 .0 286 . 4 097 .. 4 709 ..0 340. 6 637 .. 4 184 . 3 804 ,.9 237 .. 2 536 . 9  787 .. 7  100 of lightning origin alone is 15.3 percent, far in excess of historic natural rates. However, this ability to distinguish between fires of lightning or natural origin from those of man-related causes provides a basis for estimating more probable figures for potential area burned. Reasonable figures for potential area burned would enable calculation of the area saved from burning and would lend themselves to economic analyses. Some of these analyses could be done using the unadjusted figures for illustration, but it would be preferable to use figures which appeared more realistic. Before the arrival of European explorers and settlers, lightning would have been the major ignition agent. There were undoubtedly Indian-caused fires as described by Lewis (1977) and others, but their effects were reportedly confined primarily to relatively small areas of camping and trapping activity, and fires were ignited with some element of containment in mind. If the pre-settlement rate of burn was assumed to be largely of lightning origin, if that rate could be determined, and if it was assumed that the lightning rate during the 15-year study, was representative of historic ignition rates, then the lightning rate for the study could be scaled down accordingly and the man-caused rate adjusted on the same basis. This was done in the following manner. An estimated pre-settlement annual rate of burn of 2 percent was derived subjectively. This rate estimates all burns that would have been recorded in contemporary Alberta Forest Service reports, including both stand-destroying fires and surface fires which might induce scarring and damage without destroying the stand itself. The age-class analyses described previously indicated a rate of 2.85 percent for the province as of 1909, and rates of 3.34 and 2.63 percent for the southern Forest Reserve area and northern forest respectively. Those figures would be higher if the incidence of non-destroying surface fires had not been excluded, but even in 1909 these figures would also reflect increases induced through human activity. For example, the 3.34 percent rate for the Forest Reserve includes the years of accelerated activity in that region associated with increasing exploration from 1860, to construction of the Canadian Pacific Railway completed in 1895 (Byrne 1968). The extent to which these human activities contributed to increased rates of burn could not be determined nor could the possible influence of  101 climate change be assessed. The lower rate of 2.63 percent in the northern forest probably reflects a lesser level of human activity, but that figure would certainly have been somewhat influenced by man, as outlined in the historical section. Rolling the age-class data back to 1850 or 1870 was not attempted. The roll-back method accumulates age-class area into fewer and fewer classes, resulting in inordinately high indicated rates of burn. In an earlier age-class analysis of data from northeastern British Columbia (Murphy 1982b) the average annual rate of burn for the period 70-110 years ago was 1.77 percent for stand-destroying fires only. The rate in that boreal forest area was probably little affected by man at that time. Edgecombe and Caverhill (1911) reported rates of burn in the northern part of the Rocky Mountains Forest Reserve of 1.6 percent in the last 50 years (1860-1910) and 1.9 percent in the last 25 years based on evidence of destroyed stands. Van Wagner (1978) in his study of the St. Regis (Alberta) age-class data around Hinton described an annual burn rate of about 2 percent up to 1915, and suggested that this represented the time before which fire control became effective. These comparative figures suggest that a base-level pre-settlement rate of 2 percent area burned per year including all fires is reasonable. The adjustment of the potential area burned figures was done by scaling back the area for the lightning rate of 15.4 percent to the estimated base-level rate of 2 percent, a reduction factor of 7.7. That factor was also applied to the area for man-caused fires and total (Table 6.6). The figure for total potential area burned generated by the model had also been summarized in three fire-size categories: Actual Class E fires, Potential Class E fires, and fires which fell into Classes A-D. The actual Class E fires are ones which actually exceeded 200 ha and which were therefore recorded as Class E fires on the fire reports. The Potential Class E fires were fires whose actual area was 200 ha or less, yet which exceeded 200 ha in the model if allowed to grow in the absence of control. The third category comprised fires whose actual area was 200 ha or less and whose size remained below that limit even when allowed to grow in the model. The model-generated potential area burned for actual class E fires was 4.2 times as great as the actual area burned. Although this was judged to be high considering the limited effectiveness of fire  102 Table 6.6 Fifteen year burn area summaries and adjusted potential burn area figures.  Number of Fires  Actual Area (1 000 ha)  Model Potential Area (1 000 ha)  Adjusted Potential Area (1 000 ha)  Mean Actual Area (per fire)  Mean Adj. Potential Area (per fire)  Lightning  4 177  2 689.6  79 489.5  10 323.3  640  2 470  Man  3 315  759.8  131 298.2  17 051.7  230  5 140  Total  7 492  3 449.4  210 787.7  27 375.0  460  3 650  293  3 402.4  14 409.5  5 103.5  11 612  17 418  Potential Class E  4 471  36.3  195 457.2  22 167.0  8  4 958  Actual Class A - D  2 728  10.7  921.0  104.5  4  38  Total  7 492  3 449.4  210 787.7  27 375.0  Mean  500  229.9  14 052.5  1 825.0  460  3 654  Category A. Fire cause  B. Fire size-class Actual Class E  103  fighting activities on large fires, the possibilities of unaccounted overlap of modelled fire areas, and low potential for early re-burn in conifer stands, the figure is sufficiently low to lend credence to the performance of the fire growth model. The indicated reduction factor of 7.7 would have reduced the area for actual class E fires to a figure somewhat less than that which actually burned. It was estimated that the potential for actual class E fires was 1.5 times greater than actual, considering that many fires in that class were controlled before changed weather conditions stopped or extinguished them, and that virtually all were patrolled and mopped-up to ensure there would be no further spread. The remaining adjustment in area was made to the totals in the other two categories. The results of the adjustment are also listed in Table 6.6. Figure 6.13 illustrates the relationship between the adjusted potential area burned and actual area burned by year.  6.3.3 Indicators of fire control effectiveness The  potential for man-caused fires is evident in Table 6.6. Although total number of  man-caused fires is less than lightning, and the actual area burned less than a third that due to lightning fires, the model indicates a potential burn area 1.7 times that of lightning-caused fires. This is probably due in large measure to the timing of man-caused ignitions which occur commonly early in the fire season. The model indicates that, should these not be controlled, they may continue spreading during the fire season. This suggests the possibilities for intensified and targeted fire prevention measures. The for  adjusted potential area for Potential Class E fires is over 4 times as great as the potential  the actual Class E fires. Although the average size of the potential Class E fires is less than that of  the actual Class E potential, the number of fires is so great that the total potential area becomes substantially larger. The  actual and adjusted potential fire areas are summarized by year in Table 6.7. The ratio of  potential to actual area can help to identify fire years deserving of particular scrutiny. The year 1976 has the highest ratio indicating that the actual area burned was 76 times smaller than the adjusted potential area suggesting particularly effective control. Other years with high ratios are 1971, 1973, and  12000-,  9000H  Area Burned A Potential 0>  a  >  X  Actual  6000-1  3000-  LA i  / \  v  x —I—  1965  1970  1975  — i —  1980  1985  Year  Figure 6.13 Relationship between adjusted potential area burned and actual area burned by year.  105 Table 6.7 Actual and adjusted potential burn areas by year, and ratio of potential to actual area. Year  Actual area (1 000 ha)  Actual percent burn  Adjusted potential area (1 000 ha)  Adjusted potential percent burn  Ratio of potential to actual area  1968  321.3  0.99  545.8  1.58  1.7  1969  26.0  0.08  584.8  1.70  22.5  1970  52.2  0.16  550.7  1.60  10.5  1971  61.4  0.19  2 540.9  7.37  41.4  1972  48.4  0.15  1 571.1  4.56  32.5  1973  10.1  0.03  434.4  1.26  43.0  1974  14.7  0.05  167.1  0.48  11.4  1975  5.3  0.02  142.5  0.41  26.9  1976  20.0  0.06  1 520.6  4.41  76.0  1977  9.2  0.03  433.8  1.26  47.1  1978  7.3  0.02  212.6  0.62  29.1  1979  190.3  0.59  1 582.4  4.59  8.3  1980  663.6  2.05  11 922.7  34.59  17.9  1981  1 348.8  4.16  2 498.3  7.25  1.9  1982  670.8  2.07  2 667.1  7.74  4.0  Total  3 449.4  5.29  7.9  Mean  27 374.8 0.71  106 1977. Some of the possible aspects to examine during those years which may have contributed to that apparent success include levels and types of presuppression preparedness, initial attack capability and back-up support, fire location and access factors, strategy or tactics, and money spent. The three years with the lowest ratios are 1968, 1981 and 1982, all three of which have been identified as problem years. During those years suddenly-imposed major fire starts coupled with severe burning conditions may have resulted in overwhelming of initial attack capabilities. Those and other factors could be evaluated in a focused review. The differences between actual and potential percentage area burn figures also give indications of relative fire season severity. A fourth approach to analysis is to plot the potential and actual areas burned on a scatter diagram for each of the fifteen years as illustrated in Figure 6.14. The regression shows those same three problem years of 1968, 1981 and 1982 as deserving of particular analyses. The year 1980 also warrants study in view of its apparently favorable results. Since this scatter diagram does not incorporate differences in FLI, costs should be given particular scruntiny.  6.3.4 Area saved from burning  An estimate of the area saved from burning is derived from the difference between actual area burned, and the adjusted potential area. Results are presented in Table 6.8 distinguished both by fire-cause and by fire-size class. The area saved from burning as a result of control of man-caused fires is over twice that for lightning-caused fires. This reflects the suppression success rate which is largely a result of faster detection and ease of surface access. The area saved as a result of action on fires in the smaller size classes is far greater than that of area saved from class E fires through a combination of successful attack on small fires, and possibly low burning conditions at the time of ignition which combined to keep the actual area relatively low. The average annual reduction in fire area and difference in average annual rates of burn is summarized in the following tabulation.  Figure 6.14 Scatter plot showing relationship between adjusted potential area burned and actual area burned among individual fire years.  108 Table 6.8 Total area saved from burning in 15 years, by fire cause and size-class categories. Fire Category  Potential area burned (adj.) (1 000 ha)  Actual area burned (1 000 ha)  Area saved from burning (1 000 ha)  10 323.3 17 051.7  2 689.6 759.8  7 633.7 16 291.9  5 103.5  3 402.4  1 701.1  and Actual Class A-D  22 271.5  47.0  22 224.5  C. Total  27 375.0  3 449.4  23 925.6  1 825.0  230.0  1 595.0  A. Fire Cause Lightning Man B. Fire size-class Actual Class E Potential Class E  D. Mean (15 years)  109 Lightning fires  Man - caused fires  Total  Area saved annually km  5 089  10 861  Potential area %burn  2.00%  3.29%  5.29%  Actual area %burn  0.55%  0.16%  0.71%  2  15 950  A summary of fire costs for the study area in constant 1981 dollars (Table 6.9) distinguishes between presuppression and fire fighting costs. The cost to save each hectare from burning by each of these categories, is shown as follows: Cost per hectare saved (1981 dollars) Cost Category  Cost per ha saved  Presuppression  $10.57  Fire suppression  $11.05  Total  $21.62  The total cost of $21.62 per ha saved during this recent period is less than the $28.88 per ha during the period 1909-1969. The lower figure may reflect contemporary economics of scale or greater efficiency through activities such as training and prevention, but the difference is not likely statistically significant. The relative fire fighting costs on Class E fires and those in the smaller classes compare graphically comprising $152.27 and $0.23 per hectare saved respectively. This also emphasizes the economic importance of effective initial attack if control of fires is an objective. A review of the fire suppression costs in Table 6.9 illustrates the great variation among years in response to fire season severity - both actual and, more recently, perceived. To develop a model illustrative of the relationship between fire control costs and benefits gained, fire season severity must  110 Table 6.9 Study area fire costs in constant 1981 dollars. Year  Pre-suppression costs ($1 000)  Fire Fighting costs ($1 000)  Total costs (FME) ($1 000)  1968  13 684  12 137  25 821  1969  15 058  5 636  20 694  1970  13 854  8 205  22 059  1971  14 871  11 952  26 823  1972  14 481  10 796  25 277  1973  16 417  2 943  19 360  1974  17 495 "  8 747  26 242  1975  16 447  6 942  23 389  1976  15 964  4 836  20 800  1977  16 122  3 283  19 405  1978  15 056  6 882  21 938  1979  16 016  14 375  30 391  1980  18 349  40 681  59 030  1981  25 296  59 302  84 598  1982  23 857  67 595  91 452  Total  252 967  264 312  517 279  Mean  16 864  17 621  34 485  Ill be considered as a third variable in a three-dimensional array.  6.3.5 Production function related to fire season severity (FLI) Incorporating fire season severity entails adding a series of production function curves, each representing an incremental unit of fire season severity, as postulated by Simard (1976). The continuum can be displayed more effectively as a three-dimensional graph, the axes of which comprise Fire Management Effort (FME), area burned, and an index of fire season severity. The Fireload Index (FLI), showing the strongest relationship to area burned of any index presently available, was employed. Potential Area Burned (PAB) was a possible alternative but not used since it had a lower correlation with Actual Area Burned and would be incorporated into determination of the Area Burned component of the graph. Such a representation for the northern forest area for the 15-year study period is illustrated in Figure 6.15. The highest potential area burned occurs as expected under conditions of greatest FLI with no FME. Area burned decreases with both increasing FME and conditions of lower FLI. The influence of FLI is particularly apparent. To generate such a graph required interpolating and extrapolating values for FME and area burned for various values of FLI in order to complete the grid. For any one year, area figures were only available for $0 FME (the potential area burned), and for the total FME expenditure (actual area burned). Intermediate and extended values for purposes of plotting were predicted mathematically by assuming the shape of the curve to be a negative exponential as illustrated by Sparhawk (1925), referred to by Chandler et al. (1983) and suggested earlier in this thesis in Figure 6.8. The final graph (Figure 6.15) was generated in a curve-smoothing program. An SAS statistical program was used to derive a multiple regression formula to predict grid values of area burned for any value of FLI. The following regression was produced: Area burned = EXP[-5.678 + 1.947 (log FLI) - 0.03568 (FME)] A stepwise regression (SAS Statistical Package) was used to calculate correlation of variables with area burned. The step which included FLI and FME showed both variables to be highly significant  Fire  Management  F i g u r e  » Eeffort ffort  (Smillion) v *  A .« F M E and FLI plotted from < n Relationship of area burned to F M E 6.15 n erived values.  ^^ .5  113  (p=0.0001) with an R square of 0.577. A note of caution in interpreting these results must be added. Since so few actual data points (30) were available, intermediate predicted points had to be added as described above, so the calculation is based on a combination of actual and predicted values. However, this was done to try to obtain a graphic representation of the relationships between area burned and the variables of FLI and F M E , using the data available. The relationship of area burned to dollars spent (FME) is much more tenuous than that of FLI, which was presented earlier, in light of the "efficiency" factor which comprises both economic and human elements. The economic aspect relates primarily to the most efficient apportionment of presuppression and suppression forces. Aspects under direct human influence include such factors as choice of attack methods, type and location of presuppression forces, and decisions on priorities of action. In fire suppression operations, with their fluidly dynamic situations, human judgement and performance become even more pivotal. The subject of efficiency cannot be addressed in this study, but its importance for future analyses is evident.  6.3.6 Least-cost-plus-loss analysis To address the question of least-cost-plus-loss or the optimal solution requires adding a value for damage. The subject of damage appraisal is beyond the scope of this thesis but, clearly, value figures are needed to complete the framework for analysis and to enable presentation of a final set of graphed cost relationships. The Alberta government fire program objective includes protection of a broad range of forest resource benefits," values for many of which are difficult to quantify. To simplify this aspect only timber values were considered, values which present the least difficulty in assignment of dollar equivalent figures. As Mills and Flowers (1983b) commented, despite the limitations, the dollar-valued net value change is still an important ingredient in fire management analyses. Two approaches to valuation are illustrated here, one based on the forestry sector contribution to the gross Alberta domestic product, the other on a timber replacement value developed by the Alberta Forest Service. 24  F . W . McDougall, Deputy Minister of Renewable Resources. Personal communication, April 1985.  114 The fire control objective of the Alberta government for the timber-producing capability of productive and potentially productive forested lands. Protection of presently utilized forest management units is effected to maintain the raw material supply required to sustain existing forest industries. Protection of the remaining under-utilized forest management units is intended to maintain commercial timber stands and sustain the existing allowable annual cut (AAC) in anticipitation that forest industries will soon view these areas as investment opportunities since they represent the last major unallocated forest resource in Canada. The government aim is to have these areas utilized to support an expanded forest industry in anticipation of the benefits which would accrue to the province through capital investment, economic diversification, and further contribution to the gross domestic product (GDP). Since these forested lands are in public ownership, reference to GDP is convenient in light of the owner's and manager's (government's) objectives. Although convenient, the GDP figure comprises both costs and benefits and therefore should tend to represent an overestimate of the contribution of timber alone. The province contains 19.20 million ha of productive forest (Canada 1982). In 1978-79 the Alberta forest industry contributed $502 million (in terms of 1981 dollars) to the GDP (Canada 1982). The timber to support this economic activity was derived from 4.48 million ha of managed forest. This equates to $112.00 per ha of productive forest land annual contribution. Productive forested lands comprise slightly less than 65% of provincial forested lands," so the mean potential value of all forested lands is $72 per ha. The second valuation figure is one developed by the Alberta Forest Service (Alberta 1981) as a basis for charging damages to forest values incurred as a result of mineral exploration activities. Damage to regeneration was based on reforestation costs, loss of merchantable stands reflected the extra cost of supplying equivalent timber to a mill, and immature stands were valued on reforestation cost compounded by stand age. For ease of administration, a mean value based on proportion of stands by area for the province was calculated which would be applied to all disturbed lands. The 1981 value was $175 per ha. For the purposes of this analysis, this value is less defensible than the previous one "D.E. Fregren - Personal Communication, April 1985.  115 based on contribution to GDP since it is developed more on an.administrative than economic base, and is not consistent with the provincial policy of charging reforestation as an accounting cost rather than an investment. However, it is useful as a comparison. 2t  Phillips et al. studied the economic rent of forest land in the Peace River country of Alberta based on production of conifer timber. Preliminary results indicated an average stumpage value of $220 per ha. If the value of deciduous timber was assigned half this value, and an adjustment made for non-productive lands, the mean stumpage value for all lands in the study area would be $108 per ha. This value is intermediate to the previous two and is, interestingly, higher than the GDP-derived figure. However, it suggests that timber-based values may reasonably lie within this general range. Both value figures employed disregard the questions of allowable cut effect and substitution, salvage, and other values. Since this study was conducted on a macro scale and not on an area-specific basis, the strategic planning aspects of forest management such as age-class distribution and regional characteristics could not be considered either. All fire is assumed to be either lethal to trees or to render them unsuitable for pulp since no measure of fire intensity was incorporated into the model. Van Wagner (1983) described how substitution or harvesting of younger stands instead of older stands which may have been destroyed by fire ameliorated the effect of fire on A AC. For example, the mean figures for the 15 years of this study showed a reduction of average annual burn from a potential 5.29 percent to an actual 0.71 percent - a reduction factor of 7.45 to one. An extrapolation of Van Wagner's curve relating A AC to rate of burn showed an A AC reduction ratio of only 4.02 to one, demonstrating the compensating effect of substitution. It would not be too difficult to add a sub-routine to a future computer program to calculate the differences in AAC reduction as they varied with percentage burn in order to approximate more closely the substitution effect. It would also be necessary to adopt appropriate yield table data. However, these refinements would not contribute a substantial difference to the results of these examples. Further, the question of substitution is still open. Reed and Errico (1984, 1985) showed for the northern interior of British Columbia that even "Phillips, W., G. Armstrong, J. Beck, and K. Banskota. 1985. Economic consequences of forest- land losses to agricultural uses: an Alberta case study. Fac. Agric. and For., Univ. Alberta. Draft ms. in review. Feb. 1985.  116 modest rates of fire can result in very large reductions in long-run yield of white spruce. Mills and Flowers (1983b) concluded that probably one of the most controversial aspects of the net value change computation method was whether the substitution of unburned for burned resources should be reflected in the computation. They suggested that substitution should not be included in estimates intended for long-term planning applications because it clouded the underlying resource productivity impact of the fire with the sometimes ephemeral management constraints. Van Wagner (1983) concluded that his results implied that the real impact of fire in managed forests is properly judged by the effect on the harvest, not from data on area burned and volume killed. In his example dealing with AAC, that was certainly the case. However, this study illustrates that area itself is useful as an analog of fire effects from which interpretations of various aspects may be made. The questions of salvage and other values were not addressed in this study. Salvage of fire-killed timber can help to offset losses, but its applicability varies with fire intensity, stand age, species, location and product. For high quality pulp, for example, no char is acceptable. Estimates of salvageability could later be developed but was not attempted in this study. The forest protection organization in Alberta was developed primarily in response to protecting timber values, although many other related values were usually specified or implied as well, and the present policy clearly refers to other values. Consideration of timber values alone may not be adequate, but serves as a useful place to begin. Perhaps one of the more significant categories of other contemporary values at risk are physical structures within the forest including towns and communities of various sizes, petroleum and natural gas facilities comprising pumps, gathering stations, gas plants, storage tanks and sulphur extraction plants, and other structures such as sawmills, lodges, and resorts. Approximate values assigned to a list of structures in northern Alberta" totalled about one billion dollars. The potential average annual burn of 5.29 percent represents a fire cycle of only 19 years, the figure suggesting that any one facility might be threatened by fire once in 19 years in the absence of fire control action. It could be argued that one-19th of the combined value of facilities and "J. Skrenek, Forest Protection Branch. Alberta Forest Service. Personal Communication.  117 structures be added to a damage equation - $52.6 million per year or $30.84 per ha. There is not a simple relationship here, for there are alternative actions which may be taken to protect them. However, the example further illustrates the complexity of the valuation component. Using the predicted values of area burned illustrated in Figure 6.15, loss figures were calculated for both the $72 and $175 per ha values. The losses were added to the corresponding values of FME to generate cost-plus-loss figures, the plots of which are illustrated in Figure 6.16 and 6.17. The low points on the curved surfaces of the graphs were joined to show the lines of optimal solution, or points of least-cost-plus-loss, as related to FLI. The vertical axis represents cost-plus-loss so shows an increase with FME even at FLI values of zero. For values of $72 per ha, reflecting potential contribution to GDP, the optimal line follows one of essentially zero FME expenditure until FLI conditions of about 400 are reached. The maximum indicated FME of about $90 million is reached at FLI 1200, where the optimal line then exceeds the scale values. For the value of $175 per ha, reflecting the forest replacement costs, some level of FME is indicated after FLI values as low as 200, and the optimal FME exceeds the extended $90 million on the scale at FLI 1600. Despite the differences in values of the two loss figures, the shape of the curved surfaces is remarkably similar, as is the shape of the lines of optimal solution. This similarity suggested the possibility of a third graph to show the trend of a line of optimal solution for a range of forest resource values based on average conditions of FLI over the study period. In this way relationships may be seen regardless of the difficulty of assigning values. Figure 6.18 was derived from the 15-year mean values. The FME values indicated by the line of optimal solution, or low point on the cost-plus-loss surface, do not rise above nil until values of about $22 per ha are exceeded. However, the line curves sharply upward with increasing values. Of particular interest in this analysis is that the mean optimum point for the economic contribution value of $72 per ha is an FME of about $35 million. The actual mean FME over 15 years was $34.5 million, suggesting that the optimum mean FME has now been reached if the Alberta  118  o  s  o Ul  c o (0  c  2 0)  n Id i s o l  oilHtui $)  -  soO \  „  t  figure 6 . 1 ' SHOWIN8  E «  o  f  t  UM 01  w  aw  o O  F\te  ft** « ° SOOI  »A an a  6 1 &  Estivation  of op'  values!  121 government is to meet its stated fire management objective for timber alone. If property values-at-risk were included, the indicated FME would also be higher. The optimum FME point for the replacement value of $175 per ha is about $60 million. These cost and value figures must be further scrutinized since the costs do not consider efficiency, and the values disregard the strategic forest management planning aspects and regional characteristics. However, they provide a starting point for further study. These figures also make it possible to calculate a benefit-cost ratio. The cost to prevent area from burning, as outlined in section 6.3.4, was $21.62 per ha saved. Using the value of $72 per ha, the benefit-cost ratio would be 3.3:1 over the 15 years of study. Although Figure 6.15 showed the relationship between area burned and FME expenditure for various levels of fire season severity (FLI), it did not illustrate the probabilities of any given level of FLI occurring. It is this uncertainty over risk occurrence which makes planning for fire control difficult.  6.3.7 Application of a risk and uncertainty matrix One way to more readily visualize the funding options and consequences may be through the application of principles of decision making, particularly those related to risk and uncertainty. There is a spectrum of possible conditions facing the decision maker in these situations. Models in which various combinations of conditions are displayed, along with their probabilities of occurring, may greatly assist the decision making process. Many of these were summarized by Murphy (1982c), one of the most promising of which was an application described by Fight and Bell (1977) using an example applied to forest regeneration. As Fight and Bell explained, the appropriate model for decision making depends on the degree of knowledge concerning the future and the decision maker's attitudes towards uncertainty. Four major decision criteria are commonly considered, the selection of which depends on the nature of the problem and the decision maker. These were described succintly by Thompson (1971) as follows: 1.  Minimax. Perhaps the most basic criterion for decision making under uncertainty is termed minimax. This is also the criterion an avowed pessimist might choose. With each action the  122 decision maker associates that decision's worst possible consequence. The decision maker then chooses the action whose worst is best. In other words, assume the worst will happen and make the best of it. 2.  Minimin. The minimin criterion represents the opposite extreme from minimax. This criterion would be adopted by an extremely optimistic decision maker. He would assume that the best will happen and then choose the action with the most favorable best consequence.  3.  Minimax regret. As an alternative to minimax, the minimax regret criterion was suggested instead for decision making under uncertainty. The reasoning behind this criterion is that it is the differences in results, not absolute amounts, which are important. This criterion generally appeals to economists because of its similarity to opportunity costs. To apply the criterion, the minimum consequence for each State of Nature is subtracted from every consequence for that State. The new results are called regrets, hence the criterion name.  4.  Laplace, or principle of insufficient reason. This criterion for decision making under uncertainty assumes, since the decision maker is ignorant about the occurrence of the States of Nature, he should act as though they are equally likely to occur. Blattenberger et al. (1984) described a conceptual model for applying the factor of risk to fire  management decision making. Their analysis identified the data needed and they outlined the potential for various applications to both large- and small-scale problems. It formed an excellent framework within which future studies may be guided. No working models based on data presently available were found which dealt with fire. However, one example illustrated by Fight and Bell (1977) dealt with decision making under risk in which the probabilities of occurrence of "States of Nature" could be estimated. The following example illustrates the four decision criteria described as well as the additional one which considers risk. Fire season severity is used to describe the State of Nature to which Thompson referred, and classes of FLI are employed. The decision alternatives are described as levels of FME expenditure, and losses are presented in terms of cost-plus-loss. The economic contribution value of $72 per ha is used. Apart from these substitutions the decision tables are set up in the same format as those presented by Fight  123 and Bell (1977). Before presenting results, two weaknesses must be noted. The first is that the State of Nature for FLI is described in categories. Since the influence of FLI appears to be exponential, the mean figure for the class tends to mask the higher losses which may occur at the upper end of the class limits. The second weakness is one inherent in this study since total FME is employed, as outlined earlier. The decision matrix would mean more if it were based on levels of presuppression preparedness with losses comprising the extra fire suppression costs plus losses. Despite these weaknesses, the ability to consider probability is an interesting feature, and examples show the potential for using this approach. Table 6.10 illustrates the application of the minimax, minimin and Laplace decision options. The minimax criterion minimizes the maximum possible loss for the four FME level alternatives. In this case the FME of $90 million minimizes cost-plus-loss at $113 million and would be the option selected. The minimin criterion seeks to minimize the minimum possible loss. The FME of zero expenditure in the FLI class of 300 indicates the least possible loss at $16 million. The Laplace criterion falls between these two extremes, and requires that the alternative with the lowest average loss over all States of Nature be chosen. In this case the lowest average is $90 million for an FME value of $60 million. Calculation of the minimax regret criterion is illustrated in Table 6.11. This method required that the smallest loss for each state be subtracted from every other loss for that state. The largest resulting values or regrets for each alternative are compared, and the alternative with the smallest value is chosen - in this example $46 million in the $60 FME category. The final example (Table 6.12) is probably the most applicable to fire management planning situations where historic data may be used to give an estimate of the risk of any State of Nature occurring. In the example, the risk factors were determined by the frequency of occurrence of annual FLI within each of the four classes over the 15 years of the study. To minimize losses in the long run the level indicating the minimum expected loss would be selected, or $30 million FME.  T a b l e 6. 10 C o s t - p l u s - l o s s a s s o c i a t e d w i t h v a r i o u s l e v e l s o f FME a n d f i r e m i n i m i n , a n d L a p l a c e d e c i s i o n s a t v a l u e s o f $72 p e r h a . Level of FME ($m111 i o n )  season  severity  (FLI)  to  illustrate  minimax.  State of nature F i r e season s e v e r i t y (FLI) 300  850  1 350  Cost  +  loss  1 850  Worst possible cost + loss  Least p o s s i b1e cost + loss  Average 1 oss $ m 1 1 1 i on  $m111 i o n  o  16  124  306  566  566  ii  253  30  36  73  135  224  224  36  1 17  GO  62  75  96  126  126  62  90  90  91  95  102  113  1 13  91  100  M1n1 max  Minimin  Lap 1 a c e  So 1ut1 on ( u n d e r 11ned)  T a b l e 6.11 regret . Level FME  C o s t - p l u s -• l o s s  associated  l e v e l s o f FME a n d  of Fire  ($m111i o n ) cost 0  w i t hi v a r i o u s  +  300 loss  fire  season  severity  (FLI)  to  i  llustrate  minimax  Regret c a l c u l a t i o n season s e v e r i t y - FLI  850 ($m11 H o n )  1 850  1 350  Largest regret  566-113=453  453  224-1  11 1  16-16=0  124-73=51  306-96=210  30  36-16=20  73-73=0  135-96=39  60  62-16=46  75-73=2  96-96=0  126-113=13  46  90  91-16=75  95-73=22  102-96=6  113-113=0  75  13= 1 1 1  Table 6.12 C o s t - p l u s - l o s s expected with i l l u s t r a t e a p p l i c a t i o n of p r o b a b i l i t i e s .  various  Level of FME (Smillion)  l e v e l s o f FME  and  fire  season  severity  (FLI) at values  o f $72  S t a t e of Nature F1re season s e v e r i t y ( F L I ) 300 cost  9  60% +  loss  850 @> 2 0 %  1 350 @  13%  1 850  9  7% Expected 1 OSS  ($m111 i o n )  0  16  124  306  566  113.8'  30  36  73  135  224  69 . 5  60  62  75  96  126  7 3 ..5  90 Probabi1i ty of occurrence  91  95  102  1 13  94 . 8  60%  20%  13%  7%  'Expected  loss  = (0.60x16) + (0.20x124) + (0.13x306) + (0.07x565)  =  113.8  p e r ha  to  126  The actual mean F M E during this study period was $34.5 million. Only the last example which included possibilities of occurrence indicated this option which is also the level indicated in the curve of optimal solution (Figure 6.19). This result suggests that application of the matrix approach to planning, especially the method which incorporates probabilities of occurrence, is worth further study. Smaller classes may be necessary to increase sensitivity to the curvilinear relationship of FLI and FME, or a continuous function approach may be effective. This general risk aversion approach based on probabilities is basic to the more sophisticated model described by Blattenberger et al. (1984). Two other decision criteria, Laplace and minimax regret, suggested FME levels of $60 million, the second-ranked choice of the previous criterion. These methods could be useful in situations where probabilities of risk could not be determined.  6.3.8 Geographic distribution of mean fire season severity Simard (1973) portrayed forest fire weather zones based on 10-year (1957-66) mean values of the Fire Weather Index of the CFDRS, illustrated previously in Figure 3.1. In this study 15-year mean values of the Fireload Index were calculated, and a map showing the geographic distribution of mean fire season severity was produced by plotting iso-lines of FLI values (Figure 6.19). The iso-map of FLI shows peak values in the Lac La Biche forest in the southeastern portion of the study area. The terms 'high' and 'low' are relative, but the mean FLI of over 1100 does indicate the very high potential which manifested itself in that region. The 'low' mean values in north-central and northeastern parts of the province mask the high potential which has been realized there on occasion, but were certainly relatively lower than most of the study area. The Rocky Mountains typically experience higher rainfall amounts induced by the orographic effect. The pattern in the southwest reflects this, showing a downslope gradient of increasing FLI with decrease in rainfall. It would be interesting to know whether or not that pattern was also exhibited on the southern forest areas. The pattern is different than that of Simard (1973), but not unexpectedly given both the technical and temporal differences. Simard's high areas around Grande Prairie and Fort Vermillion on  Figure 6.19 I so-lines of 15-year mean values of Fireload Index.  128  the lower Peace River probably reflect the FWI response to the higher temperature and lower relative humidities in those parkland agricultural areas. The FLI combines potential for fire spread with numbers of fires thereby incorporating the influence of wind, fuel type and ignitions. The variation among fire seasons also suggests the difficulty of comparing techniques between different study periods without controlling the influence of weather. An iso-line map for the same 15-year study period based on both techniques would be more amenable to comparison. The iso-line patterns of FLI could also be prepared each year to facilitate comparison between and among fire years, and by mean 20-day periods to illustrate changes with advancing season. Plots of mean FLI distinguished by fire cause could assist fire control planning. FLI data for man-caused fires could help to identify more prime area-specific targets for intensified prevention programming, while lightning-caused FLI plots might help in deciding initial attack resource locations.  6.3.9 Intimations of resilience and surprise, and policy implications The average potential rate of burn for the northern study area is 5.29 percent. The reciprocal of this rate yields a fire cycle of about 19 years. Whether or not a rate of burn of this magnitude could actually be sustained bears discussion. If all fires were stand-destroying in intensity, the mean stand age in northern Alberta would become about 20 years if this rate was sustained. It is debatable whether most coniferous stands could support a burn frequency of this rate. There could be a sparsity of fuels to provide the horizontal continuity necessary to carry fires surficially, depending on site and location. The low buildup of fuels would also result in lower fire-intensities, although they could still be lethal to young thin-barked trees. The fire growth model in this study does not attempt to distinguish stand-destroying from lesser-intensity fires, but much of the estimated potential area burned would be in the form of surface fires of a searing or lesser intensity. Plant communities under deciduous hardwoods could certainly support fires at 20-year or less intervals, but only during the spring and fall seasons. Except for some of the semi-open montane-type conifer stands it is conjectural whether coniferous forests could support surface fires at 20-year intervals. Some could, but the slow rate of necessary fuel buildup  129 would preclude frequent fire in most cases. However, frequent fire would probably result in changes in the plant community to one in which deciduous trees prevailed and which favored herbaceous plants, all of which would be able to sustain annual fires during spring or fall. Holling (1973) described his concept of ecosystem resilience, and elaborated upon it in 1981 (Holling 1981). The essence of his concept is that success and failure in ecosystem management are partners in the sense that success in one aspect can lead to "failure" in another. In the case of forest fire, fire control activities, as he put it, have --- "been dramatically successful in reducing the frequency and extent of fires". As a consequence of that success, fuel has accumulated in protected areas, raising the threat of more damaging and extensive fires than those that occurred naturally. The result of this ecosystem response was termed "surprise" among the human managers which has led to changes in forest fire control policy. The U.S. National Park Service experience and policy changes as outlined by Kilgore (1976) were especially cited. Holling elaborated on his concept with examples illustrating the hypothetical relationship of fire behaviour to fire intensity and fuel condition to show how, with fire exclusion, both potential for fire intensity through fuel buildup, and horizontal fuel continuity increased. The speed with which the changes are manifested affects the speed of human response to them. Fuels changes, in the case of fire, have been slow, not appearing significant in the United States until the 1970's after active fire control since the 1800's. Policy change response was also shown to have been slow. He contrasted this to the much faster policy changes in response to the rapid buildup in spruce bud worm populations in Eastern Canada, where response time could be measured in years rather than decades. In summary, Holling tentatively suggested these features in determining an adaptive response: 1.  Failures tend not to be perceived until they become crises, but the earlier they are detected the easier is the response.  2.  Presumption of success blinds one to detection of error and to the need for further policy analysis and technological development.  3.  A climate receptive to change must emerge. That requires: a. perceptive understanding of the source of surprise; b. availability of alternative policies; c. availability of the techniques to implement the policy; d. time to absorb the new situation and the new understanding; and  130 e. communication and dialogue among publics. 4. Failure is more easily detected, appreciated, and responded to adaptively if the parents of the original policy have retired. 5. The earlier the problem is detected, the less the cost of change. Hence there is a great need to monitor the correct variables and to develop incentives to respond in the face of inertia and commitments to the past."  These views have many implications in this study, five of which bear some discussion. Holling's comment that the frequency and extent of fires have been reduced through human intervention is true for his references to situations in the United States. In Alberta, the surficial extent of fires has certainly been reduced, but the frequency has not. In the developing Alberta economy, such activities as land-clearing for agriculture, mineral exploration and development, and recreational use have increased the numbers of fires. From a control standpoint, more user-specific fire prevention activities are suggested. Maximum fire sizes generated by the model indicate the latent potential for large fires, given the continuity of fuels. The largest potential fire in 15 years was 2.94 million ha. Although much larger than any of the actual fires in the last 15 years, it is not as large as the 1919 burn in eastern Alberta and central Saskatchewan which was estimated at five million ha (Murphy 1983a) as outlined earlier. Several other fires ranged between that and the approximate size of the 1950 burn on the British Columbia-Alberta border estimated at 1.61 million ha (Murphy 1978), when fire burned uncontrolled over a period of 5 months. Fires of this size can occur when fuels are available, but would become less common in an environment subjected to a sustained annual burn of over 5 percent. Historic rates of burn show great variation, as previously illustrated in Figure 6.9, showing 20-year mean rates ranging from 0.5 percent to 3 percent per year. These natural variations appear related primarily to changes in weather conditions. There may be fuel aspects involved as well, with less continuity after periods of extensive burns. The "surprise" response to which Holling (1981) referred was clearly illustrated during the 1979 fire season in the Northwest Territories (Murphy et al. 1980). Strong adverse reaction of native hunters and trappers to extensive and uncontrolled fires led to formation of a Ministerial Panel, public  131 hearings, a major report, and resulted in changes in fire control policy. Whether or not the extensive area burned was a result of ecosystem response to reduced levels of fire control or a function of variability in fire season severity could not be determined. However, in northern public opinion, the large burns were clearly a result of reduced control action. The development of vertical fuel continuity which enhances fire spread into tree crowns was also described by Holling (1981) as one of the fuel changes which can occur with fire exclusion. His point is correct, but this is also a fuel condition which is inherently characteristic of many of the northern conifer fuel types in which vertical movement of fire may occur on bark scales, 28  branch-to-branch, and by arboreal lichens. Niederleitner observed that many fires during the severe seasons of 1980 and 1981 were exhibiting crowning behaviour virtually from the time of ignition, creating conditions beyond immediate control. This behaviour is believed to be largely a result of the extreme weather conditions, but an increase in opportunities for crowning as a result of fire exclusion would certainly enhance development of additional crown fires. Changes in Canadian fire control policies have been slow to develop, substantiating Holling's (1981) comment. This is understandable in Alberta where effective fire control only became possible in the 1950's, in contrast to the early 1900's in the U.S. National Parks. The recurrence of large, high-intensity fires introduces an added cautionary note about reducing fire control effort until agencies become more confident in their abilities to control problem situations as they emerge. It is therefore not possible to conclude that the Alberta situation yet reflects a great deal of stress and response, particularly in light of variation in fire season severity and the relatively recent advent of effective fire control. At the same time, though, the study indicates a substantial accumulation of areas which would have otherwise been burned, therefore increasing the potential for the fire problems to which Holling referred. If the mean rate of burn of 5.29 percent could not be sustained indefinitely, Holling's concepts suggest that it is a rate of burn which might result during the succeeding decade if fire control activities were abruptly and totally withdrawn, and if the same level of human activity, and hence fire-starts, 28  J . Niederleitner, Head of Fire Control Planning, Alberta Forest Service, Personal communication, March 1985.  132  was sustained. The political action of totally withdrawing fire control services is highly unlikely, but the concept has profound implications for areas such as the Canadian national parks and wilderness areas for which suggestions have been made that the 'natural role' of fire be restored. Given the stress on the system in those areas, with their earlier record of suppression activity, policy changes must be made gradually and with the introduction of alternative treatments such as prescribed burning and mechanical . modifications to fuels such as falling or harvesting. The alternative would have to be a willingness to accept large high-intensity fires. Sudden withdrawal of control would undoubtedly result in a 'surprise' reaction when weather conditions were favorable for the start and spread of fires. At the same time, continuation of the present policy of fire exclusion will ultimately lead to the failure described by Holling when the fire environment created by fuel accumulation and continuity, conditioned by weather, and triggered by multiple ignitions leads to fire behaviour rendering them uncontrollable. In the Alberta provincial forests the assumption is that harvesting will eventually replace fire as the effective agent for removal of old stands and setting the stage for renewal, and that other treatments or prescribed fire may be substituted for wildfire in intermediate stand treatment such as brush and debris control in semi-open stands. Where it is planned to reintroduce fire, the risks of large and high-intensity fires should be anticipated and accepted. An alternative would be to approach the change in policy in two stages - the first of which would consist of the re-creation of the normal situation, or a capital phase, followed by a maintenance phase of planned or permitted burns. In both stages a combination of prescribed fire, relaxation of fire control standards, and mechanical treatments could be used.  7. S U M M A R Y A N D C O N C L U S I O N S This study revealed a strong influence of fire management effort in reducing area burned." There was a remarkable similarity among the historical, age-class and fire growth modelling studies. Review of the five hypotheses provides an orderly framework for review of the salient points. 1.  Descriptive historical accounts of fire policy and fire seasons can be verified by analyses of actual annual expenditures on fire. The  bar charts of costs (Figure 6.1 and Figure 6.2) reflect the descriptive historical  accounts of both general levels of forest service support and influence of fire seasons. However, the descriptive records provide only comparative relationships and cannot provide quantifiable measures of relative differences. Of particular value, the historical descriptions provide a perspective by which to judge contemporary actions. 2.  There has been a decrease in area burned which is related to increased fire management effort. Analyses of age classes indicates a progressive 80-year decline in periodic average annual rate of burn for Alberta from 2.85 to 0.25 percent. Figure 6.8 portrayed the decline as related to increases in FME  as a negative exponential as suggested by Sparhawk (1925). If the relationship is  in the form of the negative exponential, the results suggest that recent increases in fire load prevented as great a reduction in area burned as predicted. However, it is possible that the relationship is not the negative exponential, but there were not enough data points in this study to enable suggestion of a more appropriate shape. The  analyses also confirmed the substantial variation in periodic annual rates of burn both  among time periods and geographic areas, with rates ranging from 0.25 to 3 percent per year. Analyses of age classes by potential site productivity suggested the possibility of increased rates of burn on the more productive sites. Additional studies would be needed to confirm and explain this. 3.  Variation in fire season severity can be better described by considering both the potential for fire spread and the actual number of fires.  133  134 Indices of fire season severity were tested. The Season Severity Index developed by Williams (1959) and modified by Van Wagner (1970) showed a coefficient of determination of 0.197 to area burned. A Daily Spread Potential index (DSP) based on potential for spread alone had a coefficient of determination with area burned of only 0.018, and was therefore of no value. However, the fire load index (FLI) which combined potential for spread with numbers of fires showed a coefficient of determination of 0.693 with actual area burned. Although this index can be refined and improved, it does show the advantages of including numbers of fires in the description of fire season severity. This index helps to address the suggestions put forward by MacTavish (1965), and can help to quantify the variations in fire season severity referred to by Smith (1971). 4.  Potential area burned in the absence of fire control may be estimated by means of a fire growth model. The WHATIF.FIRE growth model was developed to try to estimate the potential extent of area burned in the absence of fire control. Total estimated areas were obviously too high, but since the calculations were all based on mathematical formulae, the results were amenable to mathematical adjustment. Adjusted potential areas burned were derived with reference to historic rates of burn related to occurrence of lightning-caused fires. The adjusted figures indicated that the potential area burned of 5.29 percent per year was reduced to the actual area burned of 0.71 percent per year, resulting in an average annual area 2  saved from burning of about 16 000 km . The estimates of area saved from burning helped to address the data omissions described by Davis (1971), Martell (1978) and Mills and Flowers (1983b). The cost data for expenditures on fire management indicate a cost of $21.62 per ha prevented from burning. The table of annual variation and scatter diagram of potential and actual areas burned per year make it possible to identify years such as 1971, 1973, 1976 and 1977 in which particularly notable results were achieved in reducing areas burned, and serious years such as 1968, 1981 and 1982 which might bear closer scrutiny for the lessons which might be learned from them. 5.  There is a relationship between fire season severity, fire management effort, and area saved from  135 burning. There was a high coefficient of determination (0.693) between the Fireload Index (FLI) and actual area burned, as outlined. The relationship to fire management effort (FME) was not as strong, due in large measure to the many human-related variables affecting the efficiency of fire management expenditures. However, the coefficient of determination is 0.577. The efficiency factor needs further detailed study. The relationships postulated in the isometric graph (Figure 6.15) are further amenable to economic analyses to determine optimal levels of FME for various degrees of FLI and forest resource values. These studies suggest that area is useful as a surrogate measure in evaluation of fire effects on a general basis. An unfortunate aspect of this study was that cost data were not clearly distinguishable between presuppression and suppression expenditures. Neither was it possible to distinguish the different types of expenditure within the presuppression category. As a result, the question of efficiency could not be addressed. However, the total amount spent (FME) alone is revealing in the analyses. The subject of appraisal of forest values at risk from forest fires was beyond the scope of this study. A value for the economic contribution of timber to the Alberta gross domestic product was selected for the economic analysis, one related to the Alberta government objective for fire management. No other values were considered in this particular analysis. The line of optimal level of FME expenditure ranged from 0 until FLI conditions of about 400 were reached, and increased in a curvilinear manner to over $90 million at FLI of 1200. For the average conditions reflected in the total 15 years of study, the optimum FME for the value of $72 per ha are $35 million for timber alone. This compares to the actual FME of $34.5 million suggesting that a break-even point has been reached considering government policy and timber values alone. Further increases in FME would be indicated if increased values per unit area could be shown, for instance by including the value of facilities and structures at risk, and if the average FLI increased, as suggested may be the case in the three years since the end of this study. The ratio of benefit to cost over the 15 years is 3.3 to 1. The isometric graphs show more clearly the curvilinear relationships between fire  136 season severity, fire management effort, and area saved from burning as reflected in costs. Application of these results to analyses of risk and uncertainty suggests possibilities for use of this technique in planning for levels of FME. The criterion which incorporates an estimate of the probability of fire risk shows the greatest promise, generating results close to that of the model relating FME, FLI and area. Finally, it can be argued whether or not the northern forest could sustain the accelerated rates of burn over 5 percent per year as indicated by the model. A closer approximation could undoubtedly be obtained later through forest type-specific analyses of vegetative response to burns of various frequencies. No one sweeping answer would likely be applicable to the area as a whole. However, the results indicate that if the present level of FME was relaxed or curtailed, those rates would likely prevail at least for the next decade as the resilience factor described by Holling (1973, 1981) asserted itself. Such a major drastic policy change would not likely be instituted on forest management areas where harvesting is expected to substitute for fire in setting back successional stages. However, there would be profound implications for management of national parks and wilderness areas where a high degree of fire exclusion has been achieved and where no other alternative disturbances may be acceptable.  8. FURTHER STUDIES SUGGESTED Many suggestions for further study emerged during this project. Among these are included the following. 1.  Refine the process of fire growth modelling for predicting area burned. a.  Use the next series of spread formulas in preparation by Alexander et al., publication pending, cited earlier.  b.  Combine these with plotted output models which could be spatially located in a data file in a Geographic Information System which would help to:  c.  1)  consider major wind direction  2)  eliminate overlap areas  3)  introduce a reburn-free interval for fires in conifer fuel types  4)  consider the few major physical obstacles to fire spread  Adapt as soon as possible the area-specific model developed by Kourtz et al. (1977) and which is being made operational at the Petawawa National Forestry Institute.  d.  Add FWI to the WHATIF.FIRE model to distinguish area burned above the FWI threshold values which would yield sufficient intensity to cause tree mortality and the more profound environmental effects of high-intensity fires.  2.  Apply the WHATIF.FIRE model to the more recent years to add to the data reference base and extend comparison studies of the effectiveness of F M E .  3.  Develop efficiency studies to determine the optimum mix of presuppression and suppression funds to minimize cost plus net value change. The optimum solution varies with fire season severity, but if applied on an "average" basis there will be inefficiencies in both low and extreme years, as outlined by Mills and Bratten (1982). It may be an improvement to consider expenditures in three categories instead of the usual two: a.  presuppression costs incurred in advance of the fire season,  b.  pre-fire load buildup - comprising mobilization of extra resources during the fire season in anticipation of an increasing fire load,  137  138 c.  suppression costs - extra costs of fighting actual fires.  This approach would enable a lower presuppression funding investment and lower total costs. To make this possible would require a) a budgetary process which would allow pre-fire load buildup on demand, b) availability of the extra resources as required, and c) ability to predict increased fire load. The last point is pivotal but some interesting work in this connection has been done by Cunningham and Martell (1973, 1976), Martell et al. (1984), Martell and Otukol (1985) and others. If predictive capability of fire starts and FWI could be combined to predict Fireload Index, it could provide the improved data needed. 4.  Use this model on a basis for assessing the effects on timber supply applying such harvest allocation models as Timber-RAM (Navon 1971), and considering the aspects of timber supply and substitution put forward by Van Wagner (1983) and Reed and Errico (1984, 1985) and Dempster and Goudie (1984). These results could help to refine calculation of the effects of fire on optimal rotation from an economic standpoint as outlined by Martell (1980) and Reed (1984).  5.  Study more closely the effect of site on rate of burn as suggested by Smith (1981) and Dempster and Goudie (1984). If there is a significantly higher rate on good sites, it would certainly affect yield calculations in harvest allocation models.  6.  Focus on fire damage appraisal techniques applicable to Alberta, considering the multiplicity of resource values, physical structures, and government policy objectives. With a better means now available through the FLI for estimating fire potential, these studies may be carried further than previously possible.  7.  Begin more specific studies of efficiency in fire control to lead to optimization of FME. Basic to this must be a major revision of the fire expenditure accounting system so that a clear distinction may be made between presuppression and suppression expenditures, and among the various categories of presuppression expenditures. Application of computer-assisted data handling programs should make this possible without major disruption to existing systems.  8.  Although the relationship of FLI to actual area burned shows a strong relationship, the effect of the probably highly significant surges of particularly high FLI should be analyzed in greater detail  139 to improve assessment of response needs and the cost-effectiveness of providing adequate responses. 9.  Growth and spread of wild fires is governed by complex relationships among such characteristics as fuel extent, quantity, arrangement, and various size, moisture and chemical qualities along with atmospheric and topographic factors. As Albini (1984) explained, fire behaviour and spread has been partially explained by theoretical models, and predicted to a certain extent by empirically-based models. However, he described several behaviour phenomena which were poorly understood including the theoretical basis of predicting the commonly-observed elliptical shape, why fire whirls develop, and why unburned strips of live trees are frequently left in running crown fires. Imprecisions in fire growth modelling were also exemplified in this study. Recent popularization of chaos phenomena or chaotic dynamics (Lees 1984, Taubes 1984) suggests an intriguing possible application to fire-related questions. Chaos is a term used to describe the apparently disordered, turbulent state of behaviour in contrast to an orderly state predictable by simple physical laws. The transition from laminar flows in fluids to turbulent flows is a commonly-cited example. Taubes described how mathematicians had begun to recognize that chaos is not the impenetrable mystery that it appeared to be, that even in chaos there is order. Lees referred to some physicists who suggested this may represent a major shift in the whole philosophy of science and the way that man looks at his world. By better comprehending the routes and patterns of chaos it might be possible to recognize the signs of incipient chaos and how to prevent its advance. In particular, studies of fluid dynamics may well help to reveal insights into the fire behaviour questions posed by Albini. Lees (1984) referred to the work of T.D. Rogers, professor of mathematics at the University of Alberta. Discussion with Rogers" reinforced the view that chaos theory could eventually assist in addressing questions of fire behaviour, as it has in studies in population biology (Rogers 1981). However, of more immediate value may be the application of other studies such as the measurement of shape (Rogers and Trofanenko 1979) as related to describing fire size,  "Personal communication, April 1985.  140  configuration, and ecological diversity. A related study in progress by Rogers may result in the type of area-specific fire growth model needed in order to better evaluate fire management effectiveness.  9. REFERENCES CITED Alberta 1947. Annual report of the Department of Lands and Mines for the fiscal year ending March 31,1946. Rep. of Director of Forestry, p. 31. Alberta 1971. Alberta forest inventory statistics by forest management units and forest. Timber Manage. Br., Alberta For. Serv. File R4.111 Economics May 1971. 85 p. Alberta 1981. Calculation of the 1981-82 provincial timber damage assessment rate (Green Area). Timber Manage. Br., Alberta For. Serv., June 1981. 11 p. Alberta 1984. Area by forest. Forest area summary. Timber Manage. Br., Alberta For. Serv. 30 August 1984, File T/M I 4.48. 1 p + 11 att. Alberta 1985. Instructions for completing the fire report form FP-48. For. Prot. Br., Alberta For. Serv. March 1985. 18 p. Albini, F.A. 1976. Estimating wildfire behaviour and effects. USDA For. Serv. Gen. Tech. Rep. INT-30. 92p. Albini, F.A. 1984. Wildland fires. Am. Scientist. 72: 590-597. Alexander, M.E. 1982. Calculating spring drought code starting values in the prairie provinces and Northwest Territories. Can. For. Serv., North. For. Res. Cent. For. Manage. Note No. 12. 4 p. Alexander, M.E. 1983a. Analysis of the Canadian forest fire weather index for the 1968 Lesser Slave Lake fire. For. Rep., North. For. Res. Cent., Can. For. Serv. Vol. 28, pp. 8-10. Alexander, M.E. 1983b. Tables for determining spring drought code starting values in west-central and northern Canada. Can. For. Serv. North. For. Res. Cent. For. Manage. Note No. 19. 8 p. Althaus, I.A. and T J . Mills. 1982. Resource values in analyzing fire management programs for economic efficiency. USDA For. Serv. Gen. Tech. Rep. PSW-57. 9 p. Anderson, H.E. 1983. Predicting wind-driven wildland fire size and shape. USDA For. Serv. Res. Pap. INT-305. 26 p. Arnold, R.K. 1949. Economic and social determinants of an adequate level of forest fire control. Ph.D. diss. Univ. Mich. Ann. Arbor. 205 p. In Gorte and Gorte 1979. Beall, H.W.T948. Forest fire danger tables (provisional), 2nd ed. rev. Dom. For. Serv., For. Fire Res. 141  142 Note 12. 73 p. Baumgartner, D.C. and A.J. Simaid. 1982. Wildland fire management economics: a state of the art review and bibliography. USDA For. Serv. Gen. Tech. Rep. NC-72. 46 p. Bellinger, M.D., H.F. Kaiser and H.A. Harrison. 1983. Economic efficiency of fire management on nonfederal forest and range lands. J. For. 81: 373-375, 378. Blattenberger, G., W.F. Hyde, and T.J. Mills. 1984. Risk in fire management decisionmaking: techniques and criteria. USDA For. Serv. Gen. Tech. Rep. PSW-80. 9 p. Bratten, F.W. 1982. Probability model for analyzing fire management alternatives: theory and structure. USDA For. Serv. Gen. Tech. Rep. PSW-66,11 p. Brown, A.A. and K.P. Davis. 1959. Forest fire control and use. McGraw-Hill, New York. 686 p. Byrne, A.R. 1968. Man and landscape change in the Banff National Park area before 1911. Studies in land use history and landscape change. Nat. Park Ser. No. 1. Ed. R.C. Scace. Dep. Geog., Univ. Calgary. 173 p. Canada. 1897. Annual report of the Department of the Interior for the year 1896. Sessional papers (No. 13). A. 1897. Ottawa, p. 27. Canada. 1934. Statement showing expenditure incurred in and for the Province of Alberta for the fiscal years 1905-06 to September 30, 1930. Department of the Interior, Forest Service. July 3, 1934. Accession No. 81.312/16 Provincial Archives of Alberta, Edmonton. 2 p. Canada. 1956. Forest fire danger tables, 1956 - Alberta east slope. For. Br., Dep. Nor. Aff. Nat. Res. 14 p. Canada. 1959. Forest fire danger tables - Alberta. For. Br., Dep. Nor. Aff. Nat. Res. 14 p. Canada. 1982. Alberta's forests. Nor. For. Res. Cent., For. Rep. No. 25. 12 p. Canada. 1983. Historical statistics of Canada. 2nd Ed. Statistics Can. pp. K33-34. Canada. 1984. Consumer Price Index. Cat. 62-001. Statistics Can., Edmonton. 1 p. printout. Chandler, C , P. Cheney, P. Thomas, L. Trabaud, and D. Williams. 1983. Fire in forestry. Volume II. John Wiley & Sons, Toronto. 298 p. Chornoby, E. and J. Holwein. 1984. Summary of northern Alberta forest expenditure for the years  143 1975-76 to 1981-82. Program Support Br., Alberta For. Serv., Dep. Energy and Natural Res. 24 August 1984.10 p. Clawson, M. 1975. Forests for whom and for what? Johns Hopkins Press, Baltimore. 175 p. Cunningham, A. A. and D.L. Martell. 1973. A stochastic model for the occurrence of man-caused forest fires. Can. J. For. Res. 3: 282-287. Cunningham, A.A. and D.L. Martell. 1976. The use of subjective probability assessments to predict forest fire occurrence. Can. J. For. Res. 6: 348-356. Davis, L.S. 1971. The economics of fire management. In: Planning in fire management. Proc. Southwest Interagency Fire Council, Phoenix, Ariz. p. 60-69. In Gorte and Gorte 1979. Deeming, J.E., J.V. Lancaster, M.A. Fosberg, R.W. Furman and M.J. Schroeder. 1972. National fire-danger rating system. USDA For. Serv. Res. Pap. RM-84. 165 p. Dempster, W.R. and J.W. Goudie. 1984. "Yield estimation for timber management in western boreal forests. W.R. Dempster Assoc., Carvel, AB. Unpub. rep. June 1984. 13 p. Edgecombe, G.H. and P.Z. Caverhill. 1911. Rocky Mountain Forest Reserve. Report of boundary survey parties. Bull. No. 18. For. Br. Can. Dep. Interior, 16 p. Fight, R.D. and E.F. Bell. 1977. Coping with uncertainty - a conceptual approach for timber management planning. USDA For. Serv.Gen. Tech. Rep. PNW-59. 18 p. Flint, H.R. 1928. Adequate fire control. J. For. 26(5): 624-638. Flowers, P.J., T.P. Hunter and T.J. Mills. 1983. Design of a model to simulate large-fire suppression effectiveness. Proc. 7th Conf. Fire and Meteorol., Am. Meteorol. Soc. pp. 168-173. Fuglem, P.L. 1983. Value systems for forest protection - a development proposal. For. Prot. Br., B.C. Min. For. Unpub. rep. 7 p. Gonzalez-Caban, A., C.W. McKetta and T.J. Mills. 1984. Costs of fire suppression forces based on cost-aggregation approach. USDA For. Serv. Res. Pap. PSW-171. 16 p. Gorte, J.K. and R.W. Gorte. 1979. Application of economic techniques to fire management - a status review and evaluation. USDA For. Serv., Gen. Tech. Rep. I NT-53. 26 p. Gray, H.W. and B. Janz. 1983. Initial-attack initiatives in Alberta - a response to the 80's. Proc.  144 Intermt. Fire Council, Fire Manage. Workshop, Banff, Alberta. 25 October 1983. ms. 14p. Green, R., and A.H. Laycock. 1967. Mountains and plains. Chapter 4, Alberta - A natural history. Ed. W.G. Hardy. Hurtig, Edmonton, pp. 69-89. Green, D.G. 1983. Shapes of simulated fires in discrete fuels. Ecol. Modelling. 20: 21-32. Green, D.G., A.M. Gill and I.R. Noble. 1983. Fire shapes and the adequacy of fire-spread models. Ecol. Modelling. 20: 33-45. Harrington, J.B., M.D. Flannigan and C.E. Van Wagner. 1983. A study of the relation of components of the fire weather index to monthly provincial area burned by wildfire in Canada 1953-80. Can. For. Serv., Inf. Rep. PI-X-25. 65 p. Harvey, D.A. and B. Janz. 1982. A comparison of fire weather severity in northern Alberta during the 1980 and 1981 fire seasons. Alberta For. Serv., Edmonton. Unpub. ms. 14 p. Holling, CS. 1973. Resilience and stability of ecological systems. Ann. Rev. Ecol. and Systematics 4: 1-23. Holling, CS. 1981. Forest insects, forest fires, and resilience. Proc. Fire Regimes and Ecosystem Properties. Dec. 1978. USDA For. Serv. Gen. Tech. Rep. WO-26. pp. 445-464. Hornby, L.G. 1936. Fire control planning in the northern Rocky Mountain region. USDA For. Serv., Missoula, Mont. 179 p. In Gorte and Gorte 1979. Johnson, E.A. and C.E. Van Wagner. 1985. The theory and use of two fire history models. Can. J. For. Res. 15: 214-220. Kiil, A.D., R.S. Miyagawa, and D. Quintilio. 1977. Calibration and performance of the Canadian fire weather index in Alberta. Can. For. Serv., Inf. Rep.NOR-X-173. 45 p. Kilgore, B.M. 1976. Fire management in the National Parks: an overview. Proc. Tall Timbers Fire Ecol. Conf. 14: 45-57. Kourtz, P.H., S. Nozaki and W. O'Regan. 1977. Forest fire in the computer - a model to predict the perimeter location of a forest fire. Fish. Env. Canada. Inf. Rep. FF-X-65. 26 p. Kun, S.F. 1958. An analysis of forest fire control standards. M.Sc. thesis. Dep. Forestry, Univ. Brit. Col. 104 p.  145 Lawson, B.D. 1972. An interpretive guide to the Canadian forest fire behaviour system. Can. For. Serv., Pac. For. Res. Cent. Rep. BCP-3-72.19 p. Lees, D. 1984. A science called chaos. Maclean's. 97(40): 76-79. Lewis, H.T. 1977. Maskuta: the ecology of Indian fires in northern Alberta. West. Can. J. Anth. 7(1): 15-56. Longley, R.W. 1967. Climate and weather patterns. Chapter 3, Alberta - A natural history. Ed W.G. Hardy. Hurtig, Edmonton, pp. 53-67. Longley, R.W. 1968. Climatic maps for Alberta. Alberta Climatological Committee. Dep. Geography. Univ. Alberta. July 1968. 8 p. Lortie, M. 1979. Arbres, forets et perturbations naturelles au Quebec. Univ. Laval, Quebec. 172 p. Lotan, J.A. 1977. Fire management perspective: USDA For. Serv. Western Wildland, Summer 1977. Mont. For. Conserv. Exp. Sta., Univ. Mont., Missoula. Mactavish,-J.S. 1965. Economics and forest fire control. Can. Dept. of For., Public. No. 1114. 24 p. Martell, D.L. 1978. A subjective measure of forest fire management effectiveness. For. Chron. 59(3): 159-162. Martell, D.L. 1980. The optimal rotation of a flammable forest stand. Can. J. For. Res. 10: 30-34. Martell, D.L. 1982. A review of operational research studies in forest fire management. Can. J. For. Res. 12: 119-140. Martell, D.L., R.J. Drysdale, G.E. Doan, and D. Boychuk. 1984. An evaluation of forest fire initial attack resources. Interfaces 14(5): Sept.-Oct. pp. 20-32. Martell, D.L. and S. Otukol. 1985. Development of mathematical models for predicting daily people-caused forest fire occurrence in Ontario. Fac. Forestry, Univ. Toronto, March 26, 1985. 83 P-  Marty, R.J. and R.J. Barney. 1981. Fire costs, losses, and benefits: an economic valuation procedure. USDA For. Serv. Gen. Tec. Rep. INT-108. 11 p. McClement, F. 1969. The flaming forests. McLelland and Stewart, Toronto. 220 p. McLean, J. and B. Coulcher. 1968. Seven days in May - meteorological factors associated with the  146 Alberta forest fires of May 18-25,1968. Alberta For. Serv. 92 p. Mills, T.J. 1979. Economic evaluation of alternative fire management programs. Symp.: Fire control in the 80's, Oct. 29-Nov. 1,1979. Missoula, Mont. pp. 75-89. Mills, T.J. and F.W. Bratten. 1982. FEES: design of a fire economics evaluation system. USDA For. Serv. Gen. Tech. Rep. PSW-65. 26 p. Mills, T.J. and P.J. Flowers. 1983a. Completeness of model specification: an illustration with fire-induced timber net value change. Proc. 7th Conf. Fire and Meteorol., Am. Meteorol. Soc. pp. 153-158. Mills, T.J. and P.J. Flowers. 1983b. Estimating fire-induced net value change in resource outputs. Intermt. Fire Council Symp., Banff, October 25-27, 1983. In press. Draft 25 p. Murphy, P.J. 1978. Fire research at the University of Alberta. In: Fire Ecology and Resource Management. Comp. D.E. Dube. Nor. For. Res. Cent. Inf. Rep. NOR-X-210. pp. 90-91. Murphy, P.J. 1981. Computer applications to forest fire management. Term pap. for J.H.G. Smith, Fac. For., Univ. Brit. Col. 125 p. Murphy, P.J. 1982a. Historic development of policy for the control of prairie and forest fires in Alberta. Term pap. for Dr.J.V. Thirgood, Fac. For., Univ. Brit. Col. 305 p. Murphy, P.J. 1982b. Describing the occurrence of fire in the forest - a review of methods. Term paper for Prof. A. Johnson. Fac. For., Univ. Brit. Col. July 1982. 70 p. + app. Murphy, P.J. 1982c. Economics of forest fire management - a review of literature and some thoughts for further studies. Term pap. for Dr. D. Haley, Fac. For., Univ. Brit. Col. July 1982. 72 p. Murphy, P.J. 1983a. The great fire of 1919 - Alberta - Saskatchewan. Fac. Agr. & For., Univ. Alberta. 15 June 1983. p. Legend 4 p + map. Murphy, P.J. 1983b. Determination of parameters for a fire-growth model for Alberta - summary of a workshop 27 September, 1983. Fac. of Agric. and For., Univ. Alberta. 4 November, 1983. Unpub. ms. 10 p. Murphy, P.J. 1984. Summary of forest administration and protection costs: Alberta 1905-1982. Fac. Agric. and For., Univ. Alberta. September 1984. 27 p.  147 Murphy, P J . 1985. History of forest and prairie fire control policy in Alberta. Alberta Dep. Energy and Natural Res., ENR Rep. No. 1/11. 408 p. Murphy, P.J., S.R. Hughes and J.S. Mactavish. 1980. Forest fire management in the Northwest Territories - a review of 1979 forest fire operations and forest fire management policy. Forest Fire Review Panel. Can. Dep. Indian and North. Aff. 164 p. Navon, D.I. 1971. Timber RAM, a long-range planning method for commercial timber lands under multiple-use management. USDA For. Serv. Res. Pap. PSW-70. 22 p. Nimchuk, N. 1983. Wildfire behaviour associated with upper ridge breakdown. For. Protection Br., Alberta For. Serv. Dep. Energy Natural Res. ENR Rep. T/50. 46 p. Noste, N.V. and J.B. Davis. 1975. A critical look at fire damage appraisal. J. For. 73(11): 715-719. Parks, G.M. 1964. Development and application of a model for suppression of forest fire. Manage. Sci. 10(4): 760-766. Quintilio, D. and A.W. Anderson. 1976. Simulation study of initial attack fire operations in the Whitecourt Forest, Alberta. Can. For. Ser., Inf. Rep. NOR-X-166. 35 p. Quintilio, D., G.R. Fahnestock, and D.E. Dube. 1977. Fire behaviour in upland jack pine: the Darwin Lake project. Can. For. Serv., Inf. Rep. NOR-X-174. 49 p. Reed, W.J. 1984. The effects of the risk of fire on the optimal rotation of a forest. J. Env. Econ. Mgt. 11: 180-190. Reed, W.J. and D. Errico. 1984. The impact of forest fire in the Fort Nelson TSA. Dep. Math., Univ. Victoria. Unpub. rep. no. 780-13-162. 23 p. Reed, W.J. and D. Errico. 1985. Assessing the long-run yield of a forest stand subject to the risk of fire. Can. J. For. Res. 15: 680-687. Rogers, T.D. 1981. Chaos in systems in population biology. Prog. Theor. Biol. 6: 9-146. Rogers, T.D. and S.C. Trofanenko. 1979. On the measurement of shape. Bull. Math. Biol. 41: 283-304. Rothermel, R.C. 1972. A mathematical model for predicting fire spread in wildland fuels. USDA For. Serv. Res. Pap. INT-115. 40 p.  148 Rothermel, R.C. 1983. How to predict the spread and intensity of forest and range fires. USDA For. Serv. Gen. Tech. Rep. I NT-143.161 p. Rowe, J.S. 1972. Forest regions of Canada. Can. For. Serv. Pub. No. 1300.172 p. Schweitzer, D.L., E.V. Andersen and T.J. Mills. 1982. Economic efficiency of fire management programs at six National Forests. USDA For. Serv. Res. Pap. PSW-157.-.29 p. Simard, A.J. 1973. Forest fire weather zones of Canada. Can. For. Serv., 1 p. map. Simard, A.J. 1976. Wildland fire management - the economics of policy alternatives. Can. For. Serv. Tech. Rep. 15. 52 p. Simard, A.J., A. Young and R. Redmond. 1977. AIRPRO. An air tanker productivity computer simulation model. The Fortran program (Documentation). Can. For. Serv., Inf. Rep. FF-X-64. 341 p. Simard, A.J., G.A. Young and R.D. Redmond. 1978. AIRPRO. An air tanker productivity computer simulation model. (Application). Can. For. Serv., Inf. Rep. FF-X-69. 17 p. Simard, A.J. and A. Young. 1978. AIRPRO. An air tanker productivity computer simulation model. The equations (Documentation). Can. For. Serv., Inf. Rep. FF-X-66.191 p. Smith, C.B. 1983. Political implications of fire management. Proc. Intermt. Fire Council, Fire Management Workshop, Banff, Alberta. 25 October 1983. ms. 13p. Smith, J.H.G. 1971. How much forest protection is needed? For. Chron. 47(1): 1-3. Smith, J.H.G. 1981. Fire cycles and management alternatives. Proc. Fire Regimes and Ecosystem Properties. Dec. 1978. USDA For. Serv. Tech. Rep. WO-26. pp. 511-531. Smith, J.H.G. and D.E. Gilbert. 1976. Rates of spread and fire damage to timber cover types in British Columbia. Proc. Tall Timber Fire Ecology Conf. No. 15. pp. 135-154. Sparhawk, W.N. 1925. The use of liability ratings in planning forest fire protection. J. Agric. Res. 30(8): 693-762. Stocks, B.J. and R.J. Barney. 1981. Forest fire statistics for northern circumpolar countries. Gt. Lakes For. Res. Centre, Can. For. Serv. Rep. O-X-322. 23 p.  149 Storey, T.G. 1972. A computer simulation model for fire control planning. Fire Tech. 8(2): 91-103. Taubes, G. 1984. The mathematics of chaos. Discover 5(9): 30-39. Thompson, E.F. 1971. Some approaches for considering uncertainty in forestry investment decisions. For. Econ. Sect., IUFRO 15th World For. Congr., Gainesville, FL., 19 March 1971. 10 p. Turner, J.A. 1973. A fire load index for British Columbia. Can. For. Serv. Inf. Rep. BC-X-80. 15 p. Turner, J.A. and B.D. Lawson. 1978. Weather in the Canadian forest fire danger rating systems: a user guide to national standards and practices. Can. For. Serv. Inf. Rep. BC-X-177. xx p. USDA Forest Service. 1981. Fire suppression policy. FSM 5130.3, Washington, D.C. Cited in Gonzalez-Cabin et al. 1984. Van Wagner, C.E. 1969. A simple fire-growth model. For. Chron. 45(2): 103-104. Van Wagner, C.E. 1970. Conversion of Williams' severity rating for use with the fire weather index. Can. For. Serv., Inf. Rep. PS-X-21. Van Wagner, C.E. 1973. Rough prediction of fire spread rates by fuel type. Can. For. Serv. Inf. Rep. PS-X-42. 11 p. Van Wagner, C.E. 1974. Structure of the Canadian forest fire weather index. Can. For. Serv. Pub. No. 1333. 44 p. Van Wagner, C.E. 1978. Age-class distribution and the forest fire cycle. Can. J. For. Res. 8:220-227. Van Wagner, C.E. 1981. Forest fire research in Canada - background and potential. Prep, for PPRIC survey of forest research. Can. For. Serv., Petawawa Nat. For. Inst. 12 p. Van Wagner, C.E. 1983. Simulating the effect of forest fire on long-term annual timber supply. Can. J. For. Res. 13: 451-457. Van Wagner, C.E. and T.L. Pickett. 1975. Equations and Fortran IV program for the 1976 metric version of the Forest Fire Weather Index (FFWI). Can. For. Serv. Inf. Rep. PS-X-58. 20 p. Williams, D.E. 1959. Fire season severity rating. For. Branch. Can. Dep. Nor. Aff. Nat. Res. Tech. Note 73. 13 p. Williams, D.E. 1969. Economics of forest fire control. Pulp & Pap. Mag. Can. 70(Sep.): 97-98.  10. APPENDIX Table of Contents Appendix  Page  10.1  Fire growth model computer program WHATIF.FIRE.  151  10.2  Alberta forest fire statistics 1918 -1983.  159  10.3  Summary of Alberta forest administration and protection costs 1905-06 to 1982-83.  161  150  Appendix 10.1  Fire growth model computer program  WH ATI F.FIRE  152  PROGRAM:WM4TIf  f l » C  F I R E G R O W T H  M O D E L  OEVELOPEO  Br  P R O G R A M M E D  S E P T E M B E R  THIS  MODEL M  N O R T H E R N  ALBERTA  M U R P H Y  1 9ft* < 1 »ftS - 0 3  ( R E S E A R C H  - 0 5  A S S I S T A N T )  I  CALCULATES:  DAILY  POTENTIAL  71  A  SPREAD  3)  A  FIRE  PIKES  LINEAR  POTENTIAL  LOAD  4)  TOTAL  5)  POTENTIAL  POTENTIAL  AREA  PROGRAM  SPREAD  BURNED  P O T E N T IAL  AS  AN  DISTANCE SEASON  (DSP)  BY  AND NUMBER  AND YOUNG  CONSISTS  RATIO  187ft)  OF  A  DISTANCE  FOR EACH  SPREAD  E L L I P S E ! VAN  LENGTH-TD'WIDTH  (SIMARD  ON D S P  LINEAR  TOTAL  CALCULATED  SPREAD  POR EACH  BASED  STATION  OF  EACH DAY  ON T O T A L  WITH  FIRE  INDEX  INDEX  BURNING  BASED  THE  F O R  P . J .  B V B . P E D D E R S E N  WAGNER  MAIN  FIRE  1 1 ( 1 I  ADJUSTED  US INC  FOR EACH FIRE  DISTANCE BY  MAXIMUM  PROGRAM  WINOSPEED  REPORTEO  A N D TWO  WINOSPEED.  SUBROUTINES  UNITS: INPUT  1  OUTPUT  -  D S R - F I L E S  (GENERATED  2  -  3  -  F W I - F I L e S  4  -  CLASS  T  -  ANNUAL  6  -  RATINGS  E  DAJLY  FIRES  SPREAD  VARIABLE  POTENTIAL  ON  REVISED  D S R - F I L E  -  SUMMARY  L I N E - F I L E  12  -  TEMPORARY  IS  BY  CAUSE  AND OF  BURNING  INDIVIDUAL  •  WITH  SFFW21  SEVERITY,  DISTRIBUTION  11  OBJECTMODULE  C  SUMMARY  10  «  BFFW21  IN  PERIOD  INFORMATION  DSR  COMPILED  «  FOR SEASON  LENGTH -  IN  A P S F I R ER E P O R TS f  SUMMARY  FREQUENCY f  ( FROM  (GENERATED  F I L E  OSP  FOUND  IN  IN  FIRES  OF A L L FIRES  TO R E C O R D  (TAPEFILE)  STARTING  SUBROUTINE  DAY  RATING  FINOS  FORTRANVS  DOCUMENTATION  C C  COMMON:  C C  STLOAD  -  CONTAINS  TSTRIB  •  PROM SUBROUTINE RATING SUMMARY OF ANNUAL FREQUENCY  SPREAD  RATING  DISTRIBUTION  SSR  C C  TOPS TOTAL SPREAD POTENTIAL WE I C H T E O RAT ARRAY POR LENGTH OF MONTH  C C  TCOUNT  •  COUNTER DETERMINING MAX. PREOUENCV FOR DISTRIBUTION OP B U R N I N G LENGTH PERIOD  ONLISI  -  ARRAY FOR DIURNAL ADJUSTEMENT INITIAL SPREAD I N D E X ! I S I I  REAL:  SCASONAL SEVERITY SUBROUTINE RATING  POTENTIAL(DSP)  C C  C C C  •  DAIL V  AREA  OF  WEICHT CONTAINS R LINEAR SPREAD  C C  RACC - S U M OF L I N E A R SPREAD / F I R E IN M E T E R RR1 SUM OF LINEAR SPRCAO/FIRE IN KM  C  RR  -  SUM OF  L l  C C  S  C C  AR 1 AR 2 -  ACTUAL ACTUAL  AREA AREA  C  AR 3  ACTUAL  AREA  C C  A T I • A T I -  -  RATIO  LINEAR  C  SPREAD/FIRE  FOR AREA  -  A T ]  C  WS  C C  U  POTENTIAL POTENTIAL  -  -  POTENTIAL  AREA(HA)/YEAR IN  INDEX T L 0AD  C C  C  F I R E LOAD INDEX SUM OF INDEX FOR ONE  C C  CAUSE! 2 ) < ACT<2) SUMS TDPS1 LOCAL VARIABLE  C  AOPS  LOCAL  VARIABLE  ARRAY WITH OATA THE CALCULATION  -  C  SSRt  C  AS R  C  PLOAD  AREA  -  FFMC DC -  KM/H  IN M I L E S / H 1B83 FORMULA)  TL 0AD 1 -  -  POTENTIAL TD W E I G H T  TOTAL  VARIABLE  WEIGHTED  AREA  OMC -  C C  ISI W •  C  CO  OUFF  WEIGHTED  TOTAL  MOISTURE  CODE  COST  LEGAL • LEGAL DESCRIPTION 0 T , DT1 .0 T 2 - OAY  C  MT.MH1.MH2 •  *  COUNTER  -  FUELTVPE  FN  •  FIRE -  STARTING  TIMM  C C  CA CAUSE HH3 CONTROL  STARTING  C  DAT -  ARRAY IV  SSR  BY  X  AREA  STLOAD  CODE  (INPUT)  (INPUT)  (INPUT)  PERIOD/FIRE  NO.  C  C  •  OF  (INPUT)  FOR BURNING  FU  C  TIMH.H1  TOPS  MONTH  C C  t  (INPUT)  C C  TO  OF  BY  TIME  TIME  VARIABLE  ADDRESS  D A Y ON A  HOURS  TENTH  OF  SORTING  SCALE  HOURS  FOR ASPEN FROM  AREA  ARE COEFFICIENTS  OF S S R  INITIAL SPREAO INDEX WIND SPEED I INPUT) -  T L 0AO  AND A C T U A L AREA TOPS BY X AREA  TO WEIGHT  TOTAL  FINE FUEL MOISTURE DROUGHT CODE (INPUT)  C  STATION  TO WEIGHT  BY F U E L T V P E . DATA OF LINEAR SPREAD  WEICTHEO  LOCAL AREA  RATIO  (HA ) / Y E A R  M A X .WINOSPEED  "  ON  AR E A ( H A ) / F 1 R E AREA(HA)/STATION  M A X .WINOSPEED <FOR ANDERSON -  BASEO  IN H A / F t R E ( N A ) / S T A T I O N  C  C  MILES  CALCULATION  C C  INTEGER:  IN  STATION  B-AX1S OF E L L I P S E FOR AREA CALCULATION DETERMINED BY MAXIMUM WINOSPEED  C  C C C  FOR EACH  IN  BY  C C  -  X OP A R E A PER DAY  CALCULATED  FUNCTION  FIRELOAD t - U I  C  TAT -  CUM. FREQUENCY  FOR ANNUAL  C  TOT -  C U M . FREQUENCY  FOR 0 ISTRIBUTI  DISTRIBUTION 0 N / S T A T I ON  FOR  RESPECTIVELY  153 TCOST - TOTAL OIRECT COST NF - NO OF F I R E S / S T A T I O N ANF - NO. OF F I R E S / Y E A R COUNT! 2 ) - SUMS NO. OF F I R E S 6 V CAUSE C 0 S T I 2 ) • SUMS COST BV CAUSE F I L O A D • K E E P S TRACK OF DAYS WITH F I R E S A N 0 SUMS HQ , OF F I R E S PER DAY ( S C A L E OF 1 TO 16 6 1 LENGTH - LOCAL V A R I A B L E TO ADJUST DATES TO THE S C A L E OF 1 TO ICS AS TR IB - ANNUAL D I S T R I B U T I O N OF BURNING SCOST • SUMS F I R E COST PER S T A T I O N ACOUNT - LOOP CONTROL FOR AS TR|S SI - S T A R T I N G DAY I S I SDC - S T A R T I N G DAY DROUGHT COOE SF • S T A R T I N G OAY OMC OS - START 1NG DAY FW - FWI CHARACTER  LENGTH  PERIOD  SN - S T A T I O N ABB. ( I N P U T ) ST • CONTROL V A R I A B L E TO START COMPUTATIONS T I T L E - S T A T I O N NAME ( I N P U T ) NAME - CONTROL V A R I A B L E TO S T A R T COMPUTATIONS C A U S . C L A B 1 CODE CONTAIN DATA NECESSARY FOR OUTPUT  I M P L I C I T INTEGERI A-Z ) COMMON S T L O A O I 1SB ) , TSTR IB( 130) ,SSR , TOPS , TCOUNT COMMON / R A T / D A Y S ! 1 2 ) REAL D N L t S l ( 2 4 ) / . S 5 . . S S . . S S . . S S , S O . . 5 0 . . S O . . S S , t . 6 0 . .65. .70, .76. .62. .SO. , t S , .17, l . O . .90. .60, .75, .68. (.65..60..SS/ R E A L WEICHTI I 6 > / . O 6 1 . . 13*. O i l , . 0 1 0 . .O A O, .OSS, .026. 1 O X , .062, . 04 9 , . OS 1 , .066. . 08 4 , .OS 7, .046, . 0 2 1 / REAL R,WS.RR.U.L1,S.RR1 AR1.AR2,AR3.AT1.AT2,AT3.RACC REAL STLOAO.INDEX,SSR,TOPS,TLOAD,TL0A01 REAL C < 3 . 2 > , C A U S E ( 2 ) . A C T ( 2 > , A D S P . S S R 1 , T D P S 1 , A S R . P L 0 A 0 INTEGER FFMC,OC.DMC,ISI.W,C0.0T.MT.AR.T0.DAYS.MH3 INTEGER FU,FN,TIMH,T1MM,DT1,D T2,MH1,MH2,YEAR,CA.H1 INTEGER DAT,TAT,TOT,TSTRIS,TCOST,NF,ANF,TCOUNT I N T E G E R C 0 U N T ( 2 ) . C 0 S T ( 2 ) , F I L O A O (1681 INTEGER L E N G T H ( t ) . A S T R I B ( 1 2 0 ) , S C O S T , A C O U N T . L E G A L CHARACTER*4 MONTHS(12) DATA MONTHS/* J A N . ' , *FEB .*, 'MAR.*,*APR .*, 'MAY '. 6 'JUNE* . 'JULY * . 'AUG. ' . ' S E P . ' , 'OCT . ' . 'NOV . * , 'DEC . ' / CHARACTER"2 S N . S T ( I S ) DATA S T / * BM * . ' ED * , *KN' , * AO' , * BP * . *VA*. 'FT' . 'PU' , • 'CP*. 'HL', 'CU'.'NO*. 'MN'.'RE', 'SD*. 'TY'/ DATA LENCTH/O.O.O, * 14, 1 S . 4 7 . 7 7 . 106. 1 3 9 / CHARACTER•1O C A U S ( 2 ) DATA CAUS/* L I G K T N I N G '.'OTHERS MEN*/ CHARACTER'S C L AB13) DATA C L A B / ' S L A S H " , 'JACK P I N E ' . * A S P E N */ CHARACTER "8 C O O E ( l O ) OATA C O D E / ' S P R U C E * , ' P I N E ','ASPEN ','MUSKEG','BOG •'BRUSH '.'GRASS '.'BURNS ','LOGGED'.'OTHERS'/ DATA C / 1 . 3 4 , . OAS , . 2 4 9 . 1 . O O , 1 . 7 6 . 1 . 0 1 / CHARACTER-8 N A M E ( 1 S I , T I T L E DATA NAME/*BITUMONT'.'CDRA ','KEANE '.'ADAIR •'BUFFALO *.'YATES *.'PINTO •'PUSKWASK',*COWPAR L*.'HEART L K * . * C A O O T T E *. *'NOTIKEWN' , 'MERIDI AN' . 'REDEARTH' . * SWANDIVE ' . • * TONY */ WR I T E < 4 . 3 0 0 7 ) 1  INITIALIZATION  OF LOCAL V A R I A B L E S  WR I T E ( 1 \ . 4 O 1 S > WRITE( 11,4016) C AT2-0 AT3«0 AR2-0 AR3»0 A N F - O  NF-O H«0 SCOSTtO ACOUNT*O T C O S T - O A S R - O  AOSP-0 00 3 CC* 1 , 120 ASTRIB(CC}*0 3 CONTINUE DO 13 K « 1 , 3 COUNT(K)>0 CAUSE(K)*0 ACT(K)*0 COST (K I •O 13 C O N T I N U E C C C  CHANGE  16 C C C  C C C  OP LOOP  COUNTER  FROM  OAT TO L L  DO 19 LL > 1 , 1 69 F I LOAD! LL )>0 S O I L L l>0 CONTINUE  COUNTER TO CONTROL ARE A L I K E  READS  INPUT  DATA  THAT  FROM  S T A T I O N A B B R E V I A T I O N AND S T A T I O N  FIRE  NAME  REPORTS  AOOING V A R I A B L E F POR F O R E S T A B B R E V I A T I O N 1O R E A D ( 2 . 4OOO,ENO-214) F , S N , F H , L E G A L , C A , TIMH, TIMM, 0 T1 ,MM1 .FU.AR.CO TD•O IF(AR.EO.O.ANO.CO.eO.O) GOTO 10 CHECKS WHETHER IF F I R E BURNED  SUPPRESSION ANY AREA  E F F O R T WAS  I F < S N . NE . ST( N> > GOTO 2 1 4  TAKEN  AND  C C  C C C  C H E C K S WHE T H E R  DATE  ON F | R E R E P O R T  I ¥<MH I , I T . 4 » MH I » 4 I F < MH1 . E O . 4 . A N O . 0 T 1 . L T . 1«l REWIND 3 GOTO 10 I F ( OT I , EO . O I DT I * 1  FALLS  WITHIN  SEASON  LENGTH  D T H IS  R E A D S T I T L E F R O M W E A T H E R 0 A T A A N D G O E S TO t * S T A T E M E N T | F S N F R O M F I R E R E P O R T AND T I T L E FROM WEATHER D A T A INDICATE THE SAME S T A T I O N 1 1 READ<3.4001 .END' lOOITITLE.YCAR IF< T I T L C . EO . N A M E < N \ I GOTO It READ( J , 400C I GOTO 1 1 1 « SCOS T - S C C S T•CO REAO(3.4O0S)  C  C C C  C  C C C  C C  C C C C C C C C  READS FOUND 12  W E A T H E R 0 AT A IN F W I - F I L E  STARTING  READ(],4OO2.EN0itOO) M H 2 . OT 2 IFIMH2.EO.MH1.AH0.DT2.EO.0T1) GOTO 12  START 100  UNTIL  OF F I R E  SPREAD  DATE  GOTO  SIMULATION  OF F I R E  IS  lOO  END $  END O F 2 0 0 •  LOOP  M N 1 • M H1 0 I «0T1  CONVERSION  OF C A L E N D A R  DATE  TQ S C A L E  (1  -  1S9)  D A ' ( L E N G T H < MH1 ) I * 0 T 1 H 1 •TIMH IF(TIMH CONVERSION  . LT .  1)  H1 * 10  OF T E N T H  OF HOUR  INTO  M t MUTCS  M1•(TIMM I* C C O N V E R S I O N O F C A L E N D A R D A T E TO C U T O F F I N S P R I N G AND RE I N T R O D U C T I O N IN FALL  0AT E  FOR A S P E N  MODEL  I P ( M N I .EO. S > MH1*100 IP(MH1.GT.•> MH1>120 I F ( M N 1 . E O . t t MH1«2O0 0ATE MH1*0T1 a  C C C C C  D E P E N D I N G ON F U E L T V P E F R O M F I R E R E P O R T A N D 0 A T E , F U E L T V P E S ( F U ) A R E A S S I G N E D TO F U E L M O D E L S (I). 2« J A C K P I N E 3 ' A S P E N U S L A S H A S P E N MODEL I S USED U N T I L I S J U N E AND FROM I S S E P T E M B E R . S L A S H M O D E L I S U S E D FOR F I R S T DAY O N L Y . I«2 IF(FU.EO.3.OR.FU.CO.«.Oft.FU.EO.7 U . 3 IF <I . C Q . 3 . A N D . D A T E . C T . I I S . A N D . 0 A T E . L T . 2 1 S ) I«t 14  ISO  10  MN2 • t D2«30 H2-24 M2-00 RACC'O. WS • O BACKSPACE  C  GOTO  START  OF  3  ZOO-LOOP  C C C C C C  MN1 MN2 -  S T A R T I N G MONTH S E P T E M B E R OR L E N G T H O F B U R N I N G PERIOD WEATHER C O N D I T I O N S E X T I N G U I S H THE F I R E NATURAL L V 0 0 1 • S T A R T I N G OATE DD2 - L A S T DAY  IT  DO 2 0 0 M M " M N 1 , M N 2 001*1 IP(MM.CO.MN1 )001 •01 D02*0AVS(MM) I F ( M M . C O . M N 2 >0D2>D2  C C C  STORING  C C C  READS  STARTING  DATE  IP(MM.C0.MN1JOS-OA DO 2 0 0 D D - 0 0 1 . D D 2 T0*TO* 1 WEATHER  DATA  REA0(3 tSI.CND>21O»W,PPMC.OMC.DC.1S,FW P I LOAD < O A ) * P I L 0 A O ( O A 1•1 r  MTBMM DTIOD  OA*DA* 1  C C C  STORING  C C C C C C C C C  T H R E S H O L O L E V E L S POR E X T I N G U I S H M E N T WS F I N D S M A X . W I N D S P E E O D U R I N G B U R N I N G PERIOO P F M C OF L T . S S I S E X T I N G U I S H M E N T P O R A S P E N M O O E L DMC OF L C S S T H A N 2 0 OR DC OF L E S S T H A N l O O I S E X T I N G U I S H M E N T FOR C O N I F E R MODEL A T P P M C OP S S - T T A S P E N P I R E S S M O U L O C R (IS1«1I P R O M I S J U N E TO I S S E P T E M B E R A S P E N F I R E S 0 0 N O T S P R E A O .  STARTING-OA Y  VALUES  FOR  IF(DO.EO.ODI.AND.MM.EO.MNI)  2Of  DC.ISI.DMC GOTO  20*  IP(I.10•3.AND.MM.EO.B.AND .DO.GE.IB) I P ( I S . B O . O ) GOTO 2 1 0 I S * P L O A T ( IS ) W* F L O A T 4 W > I F ( W . G T . W S ) WS-W I P d . C Q . S . A N D . PFMC . L E . TT ) I S * 1 t P ( O O . C O . O I ) GOTO 2 1  GOTO  21O  155 IFII . E 0 . 3 .AND. PrMC.LT.Ill GOTO 210 IFI1.E0.2 . A M 0 . DMC . I T . 30 1 COTO 210 IPII.EQ.2 .AND. D C . I T . I O O ) COTO 210 C C C C C  THIS I S T H E R A T E OF S P R E A O IROS) FORMULA FROM A O J U S T E O TO M/OAV I1440), BT . 1 , AND ICNORINC I440(MIN/OAYI • .4 ( I S 1 A D J U S T M E N T ) >11S2  VAN THE  WACNER BUI.  (11731  R«C(<I>.II.(ISI)».C(<I).2I.IIS2 RACC>AACC«fi GOTO 200 C C C  ADJUSTMENTS 21  C C C C C  FOR  DAY  1  HH1>H1 HH2• 2 4  S T A R T OF 2 2 0 • LOOP ROS U S E S THE SAME FORMULA. AND APPLIES THE DIURNAL CURVE EMPLOYED BY O U I H T I L 1 0 AND ANDERSON llBTfi)  VARIATION  DO 2 2 0 HH> HH1 .HH2 MHHH.HH.1 IPIHHHH.GT.24IHHHH«HHHH-24 R * C U [ l , l l > I O N t [ S I I H H H N I M S W * > C K W . 2 l C C C C  THIS TIME  IS THE PERIOD  A C C U M U L A T I O N OF T H E S P R E A D C A L C U L A T E D AS "TIME"  TIME'SO 1F(MM.EO.MN1.AND.DD.E0.D1.AND IF(MM.EO.MN2.AN0.D0.eO.02.AND R A C C R A C C R ' T I M E HHHI « H H MMM1« O MMM2.0 IFIMM.EO.MN1 AND.DD.eo.01.AND HHH2 « HH. 1 IF1MM.EO MN2.AND.DO.EO.02.AND IP I H M H 2 . G T . 2 4 > H H H 2 ' H H H 2 - 2 4 IF(MM.EO.MN2.AND.00.EO.02.AND  OVER  THE  INTERVENING  -HHt.E0.HH1TIME.TIME-MI . H H 2 - E 0. H H ) T I M E . T I M E . M 2 - B O  .HHI.EO.HH)MMMI.M1 .HH2.E0.HH)HHH2*HH .HH2.EO.HHIMMM2.M2  C C  IF  FUELTYPE  IS  SLASH  IT  IS  SET  TO  JACK  PINE  (CONIFER)  AFTER  DAY  1  C  220  i p i r u . e o . •> CONTINUe  200  COHTINUE  i>2  c C C  END BNO  OP OF  220 200  LOOP LOOP  c C C C  TO  STORE  STARTINC-OAY  VALUES,  JUMPEO  TO  20t  STATEMENT  COTO 210 SI-IS S D O D C SF.DMC C O TQ 20B  204  C 210  RR1.IRACCI/1OOO WRITE<B.2300)TITLE.YEAR,SN,PN NF.NF.1  C C C C C  SIMARD AND YDUHC'S I V M. ALEXANDER. SIMARD AND VOUNC'S  LENCTH RATIO  TO •  WIDTH  WIDTH  RATIO  TO  ADJUSTED  LENCTH  TO  THEREFORE  METRIC 1/L1  1 1 . < « K R ( - . 0 2 S 7 . . S 2 1 3 7 > W S . • 1 .2 I S > I R R 1 > «L1 C C C C C  CORRECTED ELLIPSE FORMULA FORMULA POR AREA (A . IA . • I> 3 . 1 4 / 4 } THE CONSTANT 7S.S CONVERTS SOUARE  KM  INTO  HA  ((3.14/4)  K l O O ) .  AT 1 • ( S ) • ( R R 1 ) • ( 7 4 . S ) ARI-(FLOAT(AR 1)4.4 IF(FU.E0.OI FU.IO WRITE!».2301ICOOE(FUI.CLAB(I) TO 1'TO-1 W R I T E ( S . 2 4 O 0 ( M O N T H S ( M N 1 I , 0 t . M O N T HS ( M T ) . D T . T O . A T 1 . AR I . L 1 . WS . R R 1 C C C  CHARACTER  ARRAYS  ARC  REASSIGNED  TO  VARIABLES  WITH  SHORTER  NAMES  Y.CODE(FU)  • •CLASHI) C C C  OUTPUT  FOR  TAPE  SUMMARY  FILE  W R I T E ( 1 ! . 4 ) S R . F . F N . Y , B . 0 I . S 0 C . S F . S I . 0 A . 0 C . D M C . IS.TO.AT 1 .AR1.RRI RR 1 "O S D I D S I ' S D I O S >.1 C C C  CHECK  222  POR  CLASS-E  PIRES  I F ( A T I .CT.2O0.OR.AR I.CT.200 I WRITE(B.lOOO) DMC,PPMC,DC.IS AT2.AT2.AT1 AR 2 " AR 2 .AR 1 K.2 IPICA.EQ.1> K•1  COTO  444  COUNTIKI.COUNTIKI.I CAUSelK)«CAUSe(K).AT1 ACT(K >"ACT(K>•AR1 COST<KI.COST IX).CO COTO IO  c C  »44  STATEMENT  WRITES  CLASS  E  PIRES  ON  UNIT  4  C BBS  WRI T B I 4 . J O 0 1 I S N , F N . Y E A R , A T I . A R 1 . L E G A L  214  IPIN.CT.1ICOTO 21S WR I T E < 7 , 4 0 O S ) Y E A R  COTO 2 2 2  C C C  21S 215  STATEMENT CALL  STARTS  I F I N . f 0 - 14 ) N"N. 1  COMPUTATIONS  RATING  COTO T T 1  POR  STATION  TOTALS  22 7 75  NN•N -1 TLOAD'O OD 22 t C • 1. TCOUN1 ASTRIB<SC)*ASTRI«<SC.*TSTR1S<SC) CONTINUE IFITCOUNT - C T . ACOUNTI ACOUNToTCOUNT AT3«AT3*AT2 Alt 3 * Aft 3 * A R 2 WR 1 T E < 4 . 3 0 0 2 I N A M E I N N I , Y E A R I NDE X *©. DO 33 O " t . 1 S I I N O E X i ( S T L 0 A O ( O I I * ( F U 0 A O ( O N 1F t I N O E X . CT .O > COTO 77 I F t S D I Q I . C T . F I L O A Of O I > 0 COTO 33  77  O)  COTO  7<  W R l T C ' S . 3 O 0 3 - 0 , S T L 0 A D ' 0 )  7ft 114  33  , F I LO A O I O * . INOCX  IF tS O IO > . C T . O ) COTO 75 COTO 114 W R I T E * 1 2 , Z 0 O 4 , N A M E 4 N N > . 0 . I N D E X TLOAD*<TLOAD>+<INDEX) S TLOAD ( O H O F U O A D I O t t O SD < 0 ) * 0 CONTINUE T L 0 A D 1 * ( T L O A D ) « W E I C H T <NN > r*lOAD*(RLOADI*TLOAD1 W R 1 T E ( T , 4 0 1 0 I N A M E I N N t . S S R , T O P S . T L O A D , A T 2 . A R 2 . N F . S C O S T SSR1*SSR"(WE1CHT1NN) > T0PS1*TDPS*<WEICHTtNN>) ASR*ASR*>5SR f ADSP*AOSP+TOPS1 A T 2 * 0 A R 2 * 0 A N F • A N F •HF N F * 0 TCOST*TCOST*SCOST t C O S T i O I F ( N . E O . 17 t BACKSPACE 2 GOTO 10  COTO  7TB STATEMENT BEGINS ON A N A N N U A L BASIS 77 S  1 IO  1  TTS  CALCULATIONS  FOR  A L L  STATIONS  W R I T E ( 4 . 4 0 1 2 ) V E A R DO 11© P * 1 , 2 W R I T E « 4 , 4 0 1 3 > C A U S < P } , C O U N T ( Pi . C A U S E < P > , A C T < P > , C O S T I P ) CONTINUE W R I T E ( 4 . 4 0 1 4 I A H F , A T 3 . A R 3 , T C O S T W R I T E ! * , 1 0 1 O ) YEAR DO 9 T* 1 ,ACOUNT T T• T - 1 TAT* TT• I A S T R I B ( T >) W R I T E ( S , 1 0 1 S ) T T , A S T R I B ( T ) , T A T CONTINUE WR I T E ( 7 , A O 1 1 ) A S R , A O S P , P L 0 A D . A T 3 , A R 3 , A N F . T C O S T  FORMAT STATEMENTS FOR MAIN C*«****«*«* • • * * * * • • • • • • * • * * • • * • • * * » •  c  BBS lOIB 1010  F O R M A T < 1 S X . i a , 1 2 X , I 3 , 1 X , 3 < 1 X . l 3 ) ) FORMAT<T7. I 7 . S X . 2 < 1 7 ) > F O R M A T ( * 1 *. T S . ' A N N U A L D I S T R I B U T I O N OF BURNING LENCTH P E R I O O * . • I X , M N ' , I X . 1 4 . I X . ' F O R A L L S T A T I 0 N S ' . / , T 4 , « S < ' - ' ) . / , T 1 O , • ' P O T E N T I A L ' . I K . *FREOUCNCV• , I X .' T O T A L ' . / . T 1 2 . ' D A Y S ' . 1 S X , ' D A Y S ' 2004 PORMAT<AIO, I 4 , F t 2 . 0 ) 2 3 0 0 FORMAT <«</-,'ReCULTS F 0 R * , 1 X . A A , 1 X . I 4 . 2 X . A 2 , * - * . I 2 . • / , S © < ' - ' M 2301 FORMAT< ' F U E L T V F E FROM REPORT * * , I X , A S , / , • *PUE LMODE L USED * ' . 1 X . A 1 0 ) 2 4 0 0 F O R M A T ( / / , ' S T A R T I N G DATE* , 1 X , A 4 , 1 3 , * THE F I R E RAN U N T I L * . I X , • A 4 , 1 3 , Z X , ' T O T A L D A T S * " 1 3 , / • ' P O T E N T I A L AREA BURNED • ' . P 1 2 . 2 . 1 X , ' N A ' , 2 X . • ' A C T U A L AREA BURNED * * . P 1 2 . 2 . 1X. * H A ' . / , • * INPUT VALUES POR AREA C A L C U L A T I O N : R A T I O * . I X , F S . 1 , * :1 * , / , • S X . ' B A S E D ON M A X I M U M W I N O S P E E O D F * , P S . O , I X , ' K M / H ' , / , • S X , ' P R E O I C T E D LINEAR SPREAD •' . P 7. 2 , 1X. 'KM * ) 3 0 0 0 FORMAT<'PIRE EXTINGUISHED N A T U R A L L Y AT OMC O P ' , 1 3 , I X , • 'PPMC D P ' , I 3 , 1X , * A N O I S I O P ' .1 3 ) 3002 FORMAT I * 1' ,T 2 0 , 3 S ( ' * ' ) , / . T 2 0 . ' F I R E L O A D INDEX F 0 R ' , A 1 O . 1 X , I 4 . / • , T 2 0 . 3 S < ' * ' ) , 3 < / ) . T 2 0 . 3 S ( * - ' ) . / . • T 2 0 . ' D A Y ' , I X , ' D A I L Y S P R E A D ' . S K . ' N O . O F * . • S X , ' F I R E LOAD' , / . T 2 S . ' P O T E N T I A L ' , 4 X .' F I R E S ' . • X , ' I N O E X ' . / . • T 3 7 . * (BUR NINC) ' , / , T 2 7 , * (M/OAY) * , 3 X .*(THAT D A Y )' , 3 X .' ' M / D A Y I '. • / , T 2 S . ' < 1 ) ' , I 0 X . *<2>*.4X. M 3 > * < 1 * K < 2 > * . / . T Z 0 , 3 S ( * - ' ) , / ) 3003 F O R M A T ( T 2 0 . I 3 , F S . O , S X , I 3 . F 1 2 . 0 ) SOOT F O R M A T ( 3 ( / ) . T S , ' S U M M A R Y OF C L A S S E F I R E S 0 2 0 0 H A , * . / . • TS,4S< *-' l . / . T S . ' F I R E ' , I X ,' V E A R ' . S X , ' P O T E N T I A L ' , 4 X . ' A C T U A L * . • 3X, * LEGAL * . / , • T S . ' N O . * 13 X . ' A R E A * . T X . ' A R E A ' , 1 1 . ' D E S C R I P T I O N ' , / . T S . 4 S ( * - * ) > 3O0S FORMAT < T 4 , A 2 , ' - * . 1 2 , 2 X , I 4 , F 1 2 . 0 , F 1 0 . 0 , S X , I 7 I 4OOO F O R M A T ( A 2 . 1 X . A 2 . 4 X . I 2 . S X . I T . I X . I I . I X , 1 2 . I I . 2 ( 1 2 ) , 3 X . I I . I X . 1 7 . 4 0 0 1 F O R M A T ( 2 ( / > , 1 2 X . A o , 1 K , I 4 r 4 0 0 2 F O R M A T ( 4 X , I I , I X ,1 2 1 4 0 0 S F O R M A T ( 4 ( / ) I 4 00S F O R M A T ( 1 7 3</ ) > 4 00B  4 0 1 0 4 0 11  4012  • • • • • • • • • • • • • • • •  FORMAT( ' 1* , T 2 0 , 'SUMMARY F O R ' ,I X . I 4 . / . T 1 8 . 2 1 I ' • * >. 3 </ I , 7 2 ( ' - ' » . / . ' S T A T I O N '. I X . ' S E A S O N A L ' , I X . ' S P R E A D ' . 3 X .' F I R E * , 4 X .' P O T E N T I A L ' . 3 X . ' A C T U A L ' ; 3 X .' N O . O F ' , 2 X ,' D I R E C T * . / . T 1 0 . ' R A T I N G ' , 3 X . ' R A T I N G ' , 3 X . ' L O A D ' . S X , * A R E A * . S X ,' A R E A ' . 4 X , ' F I R E S ' , 3 X . ' C O S T ' . / T2S , *INDEX ' , 3X. ' ( H A ) ' , S X . ' ( H A ) ' , 2 4 X . * < S ) * . / , T 1 0 . * ( 1 ) * . S X , M 2 ) ' S X . * ( 3 > ' , / . .7 2 < ' - ' I , F O R M A T ( A t . * | * , 1 X , F 4 . 2 . ' | ' , 1 X . F S . O . ' | * , F i . O . * | * . F 1 0 . O . ' | ' . F 1 0 . O , * | * . I 3 . * | * . 1 7 ) F O R M A T ( • 1 ( * - * ) . / . * TOTAL ' . * | * , 1 X , F 4 . 2 . ' | * . I X . P « . 0 . ' | * . 2 X . F S . 0 . ' I ' . F I O . O . ' I ' . F I O . O . *| * . 1 3 . * | ' . I S . / , ' N . A L T A ' . / , S I < • - ' > . / / . ' • • • COLUMN 1 - 3 A R E A R E A * . I X . 'WEICH TED A V E R A G E S ' , / . ' i 1) V A N WAGNER AND P I C K E T T 1 1 T S ' . / . ' ( 2 ) SEASONAL RATING OF A V E R A G E DAILY LINEAR SPREAD POTENTIAL . 1 X . '( M / D A V ) ' . / , ' ( 3 ) AVERAGE RATING PER STATION BAS EO O N : ' . / , T 2 O . ' 0 A I L V LINEAR SPREAD POTENTIAL K N O . OF F I R E S ' ) FORMAT! ' 1 ' , T 2 0 . 3 S ( *•*) , / . T 2 2 . 'SUMMARY OF P I R E S BV C A U S E FOR*. 1 X . I * , / , « • ( * - ' I . / , ' C A U S E * . C X , ' INO. F I R E S (' . I X . * | P O T E N T I A L |*. 1 X , * | A C T U A L |*. I X . ' ( T O T . <0ST | * . / . T 2 4 , 1  157 • ' |AR E A  (MA) |',IX.'|AAEA (HA) |'.1R.'|0IREC1(t>|'. •/.•4('•' I I 4 0 13 F O R M A T < A 1 O . ' | * . 4 1 , 1 4 . ' | * , 4 1 , F M . O , • | ' , F 1 0 . 0 . 2 X . * | ' I # . I 4 0 t 4 FORMAT! TOTAL * . 3 X . | * , l S . * | ' . F 1 i . 0 ' | * . F 1 2 . O . ' | I t l 1  ,  C C C  ADDED 401S 4014  FORMAT  ,  1  1  STATEMENTS  FORMAT! S S I , 'SUMMAR V O F ' , 1 1 , ' F I R E S ) F O R M A T < 2 I / I . 1«X, ' S T I F | R E | F U E L TYRE | * , 11K, ' S T A R T I N G ' , 1 7 1 , • *| EXTINGUISHMENT |T 0 T j ',I X , ' A R E A ' , • X , * | LINEAR | SUPPR. |' 1  • 1 « X , * |  *  |  AF S  M 0 0 E L | O A V  DC  DMC  ISI  OSR  DSP  F L I ' .  •«R , * | D A Y DC OMC I S I | O A v | P O T E N T I A L * . •41,'ACTUAL | S P R E A D | COST |*,/,S7X,'M/0AY•,34X, • MHA1',«X.*(HA(",3X,*|',2X,*<KM) | * . 3 X . * ( S I - . / , 1 4 X . 1 1 7 < - i 4 FORMAT i t 4 X . A 2 . * | , A 2 , I 2 . ' | * , A S , * | * , A 5 , * | ' , I 3 , 3 ( l 4 l , 2 2 K . ' | . • 4 ( I 4 > , 1 X , * | ' . 1 3 , ' | ' . F I 0 . 2 , 2 X , F S . 2 , * | ' , F 7 . 2 , t X . ' | * , I« t 777 STOP END ,  • •  '••••SUBROUTINES'  SUBROUTINE  T H I S  T  RATING  S U B R O U T I N E  S E A S O N A L W H I L E  R A T I N G  C A L C U L A T E S O F  C A L C U L A T I N G  R A T I N G  A N D  D A I L Y T H E  R E T U R N E D  D S P  T O  A  T E N - D A Y ,  S P R E A D T H E  M A I N  F O R  IMPLICIT INTEGER <A-Z) COMMON L O A D ( 1 S I > , O S T R I B ( 1 2 0 1 COMMON / R A T / L M O N ( 12 I REAL LOAD,ASR,SDPS INTEGER LMON.SCOUNT.OSTRIB LOCAL  4 •  1002  7  A  U S E  L O A D T N  T H E  A N D  ( D S P ) . I S  F I L E D  F I R E  W I T H  L O A D  T H E  D A I L Y  I N D E X .  ,ASR,S0PS,SC0UNT.STD<1S9I  VARIABLES ADPS MOPS SOPS DSP 0SP1 DSP 2 DSPM  2  M O N T H L Y ,  P O T E N T I A L  C O M M O N  - l O DAY R A T I N G • MONTHLY R A T I N G - S E A S O N A L R A T I N G R E T U R N E D TO M A I N OAILV SPREAD POTENTIAL • S U M OF O S P * FINDS MAXIMUM SPREAD POTENTIAL - F I N D S S P R E A D P O T E N T I A L FOR L O N G E S T BURNING LENGTH PERIOO  INTEGER DO.FFMC,DMC.OC.ISI.PWI.PLUS.FREO REAL D S R , C D S R , D S P , R I S I . S S R . R D P S ( 1 7 ) , M O P S ( S ) R E A L A D P S , A D P S t , M O P S 1 . M 0 P S 2 , S D P S 1 , O S P 1 . 0 S P 2 . DSPM I N T E G E R M M . M Q , M N , R O , S S . C H E C K , T A T , TOT , I I I CHARACTER-t TIME(17) DATA T I M E / * I S / 4 - 2 4 / 4 ' . ' 2 S / 4 4 / S * . ' S / S - 1 4 / S ' . ' 1 S / B - 2 4 / S * . •'2S/S3/8'.* 4/S-13/S*. •*14/8-23/S','24/S3 / 7 ' . ' 8/7-13/7", '14/7-23/7*. '24/72/8*. • * 3 / 8 - 1 2 / 8 ' . ' 1 3 / 8 - 2 2 / 8 ' . '23/S- !/•',' 2 / 1 • 1 I / I ' , ' 1 J / l • 2 1/I' . • * 22/S- S O / 9 */ CHARACTER*TO ILINE CHARACTER * 4 MON(S > D A T A MON / ' A P R . * , ' M A Y JUNE'.'JULY*.'AUG.'.'SEP.*/ C H A R A C T E R * 2 0 SN ACOUNT-O T AT • O D S P 1«0 0SP2*O DO 2 K « 1 , 1 2 0 O S T R I 8 ( K > mO CONTINUE 00 4 J * 1 . 1 7 R D P S ( J >"O. CONTINUE DO S Y * 1 , S M O P S < Y > «0 CONTINUE PLUS-0 SCOUNT"0 SOPS 1 * 0 CHECK-0 0«O M0*O R0*O F-O ADPS-O MDPS1«0 READ(1.lOOO,END*SSB)ILINE FORMA T ( S ( / ) I 00 7 M - 1 , • READ(1,lOOO,END«S0S)1LINE WR I T E ( 1 0 , l O O O I I L I N E CONTINUE READ( 1.lOO1 ) WRITE < IO. 2001 ) R EAO( 1 . 3 0 0 I )SN WR I T E ( 1 0 , 3 0 0 6 ) S N 00 S P* 1 , 4 READ(1,1OOO) ILINE WR I T E < 1 0 , l O O O ) I L I N E CONTINUE 0 0 t O N * 1 , 1S t READ 4 1 , 1003 ,END*«t»>MM,0.FFMC.DMC.OC. I S I PLUS*PLUS* 1  ISI  .FWI.OSR.COSR  A D J U S T M E N T TO  RISI"(FLOAT(  I S I ) ) • .»  THE NOON I S I R E A D I N G I S A O J U S T E D BY A N A V E R A G E V A L U E D E R I V E D FROM O U I N T I L I O A N D A N D E R S O N ET A L ( 1 S 7 7 ) . UNTIL I S J U N E . O S P I S C A L C U L A T E D B V T H E A S F EN M O D E L . FROM I S J U N E • SO S E P T , O S P I S B A S C O ON T H E J A C K P I N E M O O E L . D S P I S C A L C U L A T E D ON A 2 4 - H O U R B A S I S F R O M l O O O TO 2 2 0 0 H O U R S . A L L OTHER WEATHER CONDITIONS ARE AS IN M A I N . IF(N.CT.SI)  GOTO  44  CO LA  > M 4 X o o  M ti ti n  o X o u X h.  uOn  o o *O  — « u o o f« o  O O • l> c a. w * x • - t* *  -  —ft.01 MM 3 O -J » 4. r*  > O O ft. * • ' M - - o o  I M O i a xO X • o o x  « f t • O »- •* —ft.<l D • M »- U — Q • — X *- *W « *- 3 o a. a • M — u — o - M  4 «  o o • I* X  ra ~ k. ¥• <~HM nn-i-JI — ftJ HI hi W H — —M . * - V B - | » * 0 « 1 > c w x « x «J x ai x x « *- - o • • • K * * • x • « o * • K x IX« XxXX — O X • • • • M l * - K O < • o M « * - O — o « M — — -J —-J J (ft ft* X X3 —~ XW • • . M M M — P.* — . M a Q • O K O " * . — '-— O -i «-i .J •» .4 •* _ J M M O M — «t ~- ft ttl ft . . - N - •ftlX 4 X • •ft.. — . . . O a — -t — •» ~. — i - — — —u —— 4 "s. • x * o n n h - n M> « - M M •«—* *« • ~ O O N — O • 0 O — M M Xft)— — — •M« • • •M M -~ « M •ftl• 4 4 4 M M O O O O O N O O WOO0 O •0 • - O • - * t» 0 O •> • • - - U ~ft.. . . . 3 •O O O O O > S O 'OOO O- 00 "OO O M O OOw •3 • 0 i • • > •(!•)•>> • • N. • • X • • J ft. • •ftlN *• • « *• . M H M « 'hi • *ftift,. — ftf • - * M M « a ft X • < ^ N N f | H » - > \ l ^ ^ \ f \ 4< H< I N K H « 3 — X * 4 3 . < 4 0 - 4 3 ^ - ' « 3 * X « 0 3 b , « k . 3 « t * - ft 4- 3 • •ftlftX ft « . . w « 4 « X - « • X- • - X - • •X — - X X - • . X - - »• x - ~ ~ •3 * »-»-»-»-»-»-»-»-03»-»-»->-t-K»-X0»- • »- »- f r X • - • H n ' l | x w | y « - M H H - - ' * W H | u « M - » a | | - H W ^ . « . | i ft « ft 4 4 4 4 4 4 4 4 M 0 > 4 4 4 4 4 4 4 * n > f t ) 4 4 4 X »- «t X - O K »- O O hO »- - O O K t- * I - M »- • K M »- - « M »- O O •) • • I X ! I I I X ! 4 « XI I Z II I • • - - I X I 3 X 3K 4NI4«N 4-X 4 4 - X •-• — 4. X —ft.X « ft ^ H X 4 4ft.• > - » « « « « K X X « - X X X X K X X X f t i 4 « » - K « K » - 0 O J M O U t O « U ( ( O I I K O O I I I I | [ O g O | O O O I t O O K O x O | 1 0 I I U D ft 4 ft O O O O O O O O M t b O O O O O O O * * - - > O O O W X u i O D i | u i i ) D ( | u o i i | u | o f i u o M U f a - ) . i u « i i • X • •.ftft.ft.ft.ft.ft,rt--ft.ft.ft,ft.ft.ft.»,>-t-K^ft.ft.ft.«ft» ft X ft • • • • • • O n « •> «J • * o ft o — H t f t h f t " nftO-nno — e* m m •4 •* O O O O O O O " --OOOOO OOOfti • « 0 O 0 0 0 O O 0 OOOOOOO OOOfti - —— - r N N N N H nnri u u u u u o  ti  * MftO 4 O t- f* • X 3 O H x M M O • M O M M O t-o ft. * ft. a. t- •. • a, 4. U O O • o o OOO O O a O XXX M U 4 4 u> x  m  • I M X --O0IOII*!- O I a. i a *. O M U O [ M -  "  v  n  H  v  v  V  t  v  v  v  v  v  v  v  v  v  v  V  H  v  v  v  -»  — n v. — O V M 4— O ~ M v. >X 4K4 *- 4 • 3 Q X O X N X XO U U | H X Of O XI ^ O X X H SIU"-BJ «• > m •»  o o  Appendix 10.2  Alberta forest fire statistics 1918-1983  Appendix 10.2 Alberta Forest. Fire Statistics 1918-1983  Year  No. Fires  Area Burned ha  1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950  279 332 376 840 1 758 605 641 271 279 231 396 396 270 622 379 288 240 97 248 376 521 499 313 445 215 265 325 284 275 122 177 323 248  25 271 336 789 15 080 21 874 209 186 24 151 249 063 89 505 89 963 14 973 89 323 111 216 31 277 244 793 55 087 37 946 23 825 2 557 82 742 338 030 711 603 173 673 191 927 548 540 131 857 208 481 292 460 87 713 110 193 32 399 119 571 576 993 281 449  '1918-1980 from Stocks and Barney (1981) 1981-1983 from R.S. Miyagawa, Alberta Forest Service pers. comm.  Year  No. Fires  Area Burned ha  1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983  84 192 140 85 232 258 180 464 469 474 811 278 554 361 252 371 796 617 548 816 854 737 478 598 693 774 556 657 000 348 522 257 756  9 659 196 008 135 211 52 157 81 464 '281 056 6 267 93 483 35 597 8 078 78 328 1 824 7 126 7 287 21 989 28 309 9 395 400 400 29 200 52 432 63 060 49 292 10 691 18 394 4 898 22 475 15 738 7 839 194 601 672 427 1 365 532 688 349 2 818  1 1 1 1  Appendix 10.3  Summary of Alberta forest administration and protection costs  1905-06 to 1982-83  162 SUMMARY OF FOREST ADMINISTRATION AND PROTECTION COSTS ALBERTA 1905 -06 TO 1982-83 - COSTS X $1000 — Columns Descriptions: A  — fiscal year; 1 April to 31 March  B  — total cost of all Forest Service operations including fire fighting costs  C  — fire presuppression (fixed) costs not including fire fighting - proportions estimated from budget codes  D  — fire fighting costs - extra suppression  E  — total fire-related costs - (columns 1-C + 1-D)  F  — total Forest Service budgeted costs not including fire suppression (column 1-B - 1-D)  G  — total Alberta government expenditure on both Income and Capital accounts  H  — Consumers Price Index 1981 = 100. Years 1905-13 estimated from Wholesale Price Index  I to N  — cost figures adjusted by CP I to common 1981 base  SUMMARY  OF  FOREST  ADMINISTRATION  ALBERTA -  YEAR  TOTAL FOREST SERVICE  FIRE PRESUPPR  FIRE FIGHT  TOTAL FIRE  OPERN CODES  1905-06 COSTS  X  TOTAL ALBERTA BUDGET  B  1905-06 1906-07 1907-08 1908-09 1909-10  TO  $1000  AND  PROTECTION  COSTS  1982-83 -  CNSUM PRICE INDEX  ADJST FOREST SERVICE  H  I  ADJST PRE-  ADJST FIRE FIGHT  FIRE  ADJST TOTAL FIRE  ADJST OPERN COOES  ADJST ALBERTA BUDGET  M  N  6 9 6 14  .9 .4 .6 .5 33 .4  -  -  -  -  6 ,, 2 8 ., 5 5 .. 9 1 3 .. 1 30. 1  6..9 9 .4 6 . 6 14 . 5 33 .4  -  -  1 0 ., 4 1 0 .. 0 1 0 ., 9 1 0 ., 9 11 . 1  66 .3 94 .0 60 .6 133 .0 300 .9  -  -  -  -  -  -  5 9 ,. 6 85 .0 5 4 ,, 1  -  -  -  -  1 2 0 ,, 2 2 7 1 ,. 2  66 . 3 94 .0 60 .6 133 . 0 300 .9  14 . 2  -  -  -  -  1 2 .. 8  1 4 ,. 2  -  -  1 0 .. 7  131 . 0  -  -  -  -  1 1 8 ,. 0  1 3 1 ,. 0  INTERMEDIATE MEANS  SUMMARY  OF  FOREST  ADMINISTRATION  ALBERTA -  YEAR  TOTAL FOREST SERVICE  FIRE PRESUPPR  FIRE FIGHT  TOTAL FIRE  OPERN CODES  1905-06 COSTS  X  19 1 9 - 2 0 1920-21 1921-22 1922-23 1923-24 1924-25 1925-26 1926-27 1927-28 1928-29 1929-30  .0 . 8 .6 . 2 . 1 . 8  INTERMEDIATE MEANS  PROTECTION  COSTS  1982-83 -  ADJST FOREST SERVICE  ADJST PRE-  ADJST FIRE  FIRE  FIGHT  H  53. 7 9 4 ., 5 1 3 9 .,4 1 9 4 .. 3 2 0 7 .. 1 212.. 1 210., 9 186.. 2 170 , 5 2 5 2 .. 5 2 2 7 ., 2 2 5 5 ,,9 299.. 1 272 . 7 309 287 334 311 345 394  $1000  CNSUM PRICE INDEX  TOTAL ALBERTA BUDGET  B  1910-11 191 1 - 1 2 1912-13 1913-14 1914-15 1915-16 1916-17 1917-18 1918-19  TO  AND  237 .9  -  -  -  -  - -  - -  -  -  4 8 ,, 3 85 .0 1 1 9 ,. 6 1 6 2 .. 3 1 7 1 ,. 2 175 .8 183,.8 165 .8 150 , 1 2 2 4 ,, 9 195 .4 224,.5  347 .8  309 287 334 311 345 394  - -  202 .8  237 .9  2 6 8 ,.8 149 .2 272 . 7 247 . 5 287 .8 271 .5 303 .6  53..7 94.. 5 1 3 9 ..4 1 9 4 .. 3 2 0 7 .. 1 212.. 1 210..9 186 . 2 1 7 0 .. 5 252..5 2 2 7 ,. 2 2 5 5 ,, 9 299,. 1 272,. 7 .0 . 8 .6 . 2 . 1 .8  K  -  -  1 1 .. 2 1 1 .. 6  -  -  1 2 ,. 2 1 1 ,. 9 1 2 ,. 2 12 . 4 13 .4 15 . 9 18 . 0 1 9 ,. 8 22 .9 20 .2 18 . 5 18 . 5 18 . 2 18 . 4 18 . 6 18 . 3 18 . 3 18 . 6  17O0..5 1885..8 2122,.6  16 . 5  1428 .2  -  -  -  -  ADJST TOTAL FIRE  4 7 9 ..5 8 1 4 . ,7 1 142. 6 1632 . 8 1697..5 1710.. 5 1573 , 9 1171,. 1 9 4 7 .. 2 1275. 3 992.. 1 1266. 8 1616 . 8 1 4 7 4 .. 1 1697 . 8 1 5 6 4 .. 1 1798 . 9  -  -  - -  -  -  - -  4 3 1 .. 3 7 3 2 .. 8 980. 3 1363 . 9 1 4 0 3 .. 3 1417,.7 1371 , 6  CODES  ADJST ALBERTA BUDGET  M  N  ADJST OPERN  4 7 9 .. 5 8 1 4 ., 7 1 142 6 1 6 3 2 .. 8 1697 .5  1 1 1 1 ..4 1 4 5 3 .. 0 806 .5 - 1 4 9 8 .4 1345 . 1 1 5 4 7 .. 3 1483 .6 1 6 5 9 , .O 1869 .9  . 5 .9 . 1 . 2 1275 , 3 992 . 1 1266 . 8 1616,.8 1474 . 1 1697 .8 1564 . 1 1798 .9 1700 .5 1885 . 8 2 122 . 6  1217 .0  1428 .2  1 0 4 2 ,. 8 8 3 3 ,. 9 1 135. 9 8 5 3 ,. 3  1710 1573 117 1 947  SUMMARY  OF  FOREST  ADMINISTRATION  ALBERTA -  YEAR  TOTAL FOREST SERVICE  FIRE PRE . SUPPR  A  B  C  1930-31 1931-32 1932-33 1933-34 1934-35 1935-36 1936-37 1937-38 1938-39  251 . 6 394 . 7 177 . 4 169.,2 1 8 7 ,, 0  2 5 7 ., 4 141 . 1 1 19 . 6  1 3 1 ., 2 254 . 2 195,.6 366., 1 287..0 248 . 2 3 5 9 ,.7 281 . 7 284 . 3  1939-40 1940-41 1941-42 1942-43 1943-44 1944-45 1945-46 1946-47 1947-48 1948-49 1949-50  378 .0 357 . 1 3 8 3 ,.4 417 . 1 579 .6 1093 . 8  _  _  COSTS  X  TO  $1000  PROTECTION  COSTS  1982-83 -  FIRE FIGHT  TOTAL FIRE  OPERN CODES  TOTAL ALBERTA BUDGET  CNSUM PRICE INDEX  ADJST FOREST SERVICE  ADJST PREFIRE  ADJST FIRE FIGHT  ADJST TOTAL FIRE  ADJST OPERN CODES  ADJST ALBERTA BUDGET  D  E  F  G  H  I  J  K  L  M  N  215.. 2 3 5 7 ,, 1 1 5 7 .,4 1 5 2 ., 0 1 6 8 ., 0 114,.3 235 .9  2 5 1 ,, 6 295,,0 161 . 1 136. 8 1 5 1 .. 7  175,.9 345,. 1 266., 3 223..5 3 3 3 .4 249,.9 250 . 1 339 .2 3 1 8 ,. 8  _  _  347 . 2 403 .4 615 . 3  9 9 ., 7 16 . 3 3 2 .. 4 35.,3 1 .. 1 1 19 . 8 52,.5 2 0 9 ,.4 136 , 9 63 , 4 1 6 1 ,. 9 3 7 ,. 2 18. 9 80 .8 5 9 ,. 2 4 9 ,. 2 18 . 5 87 . 1 266 . 3  2 2 2 .O  81 . 4  1 3 2 ., 7 113., 2 1 1 6 ., 1 1 2 3 ,, 4 135.. 7 129.,4 160,, 1 1 7 1 ,. 5 2 1 2 ,. 7 2 3 1 ,. 2 258 .4 259 6 290 8  1905-06  AND  _  _  _  1 3 6 7 ,. 4 2 3 7 7 ., 7 1 174 . 8 1 1 7 5 ,. 0 1280, 8 892,.5 1 6 9 4 ,. 7  340,.0 365 . 7 490 . 5 881 .6  143,. 1 156., 7 150,, 1 184 , 8 1 9 7 .. 8 244 .5 2 6 5 .4 297 .2 2 9 7 ., 9 3 3 4 ,. 2 398 .6 492 .5 827 .5  3 0 2 8 4 . .6 26041, 8 2 6 1 9 0 .7 25792..3 31785. 2 2 4 2 5 8 , .7 2 4 4 5 4 , .2 2 7 1 9 4 , .8 2 7 5 3 6 , .7 30502 .8 1 1 5 7 7 1 .. 7 4 4 8 8 2 . .9 6 5 9 3 4 , .7 76740 .9 162584 .6  1 8 ,. 4 16 . 6 1 5 ,, 1 1 4 ,. 4 1 4 ,. 6 14 . 7 15 . 0 15 . 4 1 5 ,. 6 15 . 5 16 . 1 17 . 0 17 . 9 18 . 2 18 . 3 18 . 4 19 . 0 20 .8 23 .7 24 .5  1270 . 1 2346, 8 1851 , 6 1 5 4 1 ,. 6 2115,.9 1 5 7 3 ,.7 1562 . 1 2065 .6 1940 .8 2017 .9 2005 . 3 2445 .6 4464 . 5  1412 .0 1410, 9 1530,.5 1669,.2 1702,. 1 251 1 .4  2 9 9 ,.O  262,.5  4 4 6 5 7 , .2  17 . 5  1858 . 2  1209 . 1  130.. 1 1 3 4 ,. 4  34645. 8 24777. 8 2 3 4 4 8 . .4 2 5 6 5 8 . .6  _  1550..6 934 . 4 830. 6 908. 9 770., 1 7 7 4 .. 0 8 0 1 ,. 3 8 6 9 ..9 834, 8 9 9 4 .. 4 1008,.8 1 1 8 8 ,. 3 1270,.3  _  _  6 0 0 .6 1 0 7 .. 9 225..0 2 4 1 ,. 8 7 .5 7 9 8 .7 340 .9 1342 .3 8 8 3 .2 393,.8 9 5 2 .4 207 .8 103 .8 441 321 258 88 367 1086  .5 . 7 •9 .9 .5 .9  1 1 6 9 ., 6 2 1 5 1 ,. 2 1042 . 4 1 0 5 5 ,, 6 1 1 5 0 ,. 7 7 7 7 ,. 6 1 5 7 2 ,. 7 1 142 .2 2 2 12 , 2 1718., 1 1 3 8 8 ,, 2 1 9 6 1 ., 2 1396 . 1 1374 .2  1 3 6 7 ,. 4 1 7 7 7 ,. 1 1 0 6 6 ,, 9 950.,0 1 0 3 9 ,. 0 885,.0 896 .0 929 . 2 1004 , 5 968,, 4  1853 .5 1732 . 6 1789..5 1758 . 2 2 0 6 9 .6 3 5 9 8 .4  1 147 1 163 1365 1458 1624 16 19 1758 1916 2078 3377  .. 8 ,. 5 '  1645 .7  14 19 . 7  ,. 9 ,. 2 .0 ,0 .. 9 . 3 . 1 .6  _  _  208709. 6 164091. 4 162836 . 1 1 7 5 7 4 3 ,,9 2 0 6 0 1 7 .8 173612,.0 170069.. 4 165335 3 205065. 8 150675. 1 143848. 3 151926,. 3 1 5 1 3 0 0 , ,6 1 6 6 6 8 1 ,. 9 629194. 3 2 3 6 2 2 5 , ,8 3 1 6 9 9 3 .6 323801 , 3 6 6 3 6 1 0 .8  INTERMEDIATE MEANS  339 . 8  461 .6  2 4 0 3 0 1 .6  SUMMARY OF FOREST ADMINISTRATION AND PROTECTION COSTS ALBERTA  1905-06 TO 1982-83  - COSTS X $1000 -  YEAR  TOTAL FOREST SERVICE  FIRE PRESUPPR  FIRE FIGHT  TOTAL FIRE  OPERN CODES  TOTAL ALBERTA BUDGET  CNSUM PRICE INDEX  ADJST FOREST SERVICE  ADJST PREFIRE  ADJST FIRE FIGHT  ADJST TOTAL FIRE  ADJST OPERN CODES  ADJST ALBERTA BUDGET  D  E  F  G  H  I  J  K  L  M  N  3767 , .9 3005 , .7 3296 . 1 4427 , .2 4883 . 2 5679 , .3 8457 .4 9545,.3 15855,.2 17271 .6 , 15879,.0 23485,. 1 16640,.9 17918,. 1 18062,.3 18782,. 1 21475 .6 23412 .9 30683 . 7 24654,.9  4697 .2 4887 .5 , .6 5019 . 5680,.9 . 1 6076 , 6767 , .7 7553 . 3 1 1 125. .5 13313 . 1 16128 .4 18262 .4 19957,.3 20946,.9 20664 , 1 22123,. 5 23186..5 23745 .2 25048 . 2 24062 . 1 26622 . 7  899834 . 1 446680 .6 565938 .0 757765 .4 887758 .4 993133 .9 965335 .4 1235294 ,0 1009268 . 2 1490722 .0 1309440 .0 1 175377.0 1315144 .0 1323 12 1.0 1431241 .0 1808544 .o 21791 13 .0 237939 1.0 2409852 .0 2483724 .0  14359 .2  15293 . 4  135333 1.0  A  B  C  1950- 51 1951- 52 1952- 53 1953- 54 1954-55 1955- 56 1956- 57 1957- 58 1958- 59 1959- 60 1960- 61 1961- 62 1962- 63 1963-64 1964- 65 1965- 66 1966- 67 1967- 68 1968- 69 1969- 70  1443 .8 1391 . 7 1496.. 7 1794 . 1 1800..5 2086..5 2856 ,2 3403.. 3 5573 . 1 6234 . ,2 6060. 9 8758 .5 . 6910.. 5 7387 . .0 7804 .0 . 8486 : ;3 9906 . 3 1 1525. 3 14615 .9 13240-.6  689 . .4 810..5 873..3 1066..5 1322 .9 . 1460 .9 1770 .9 2756 6 3352 .4 4119 .8 4659 .5 4969.. 3 5117 .6 5 190..8 5537 . 7 5783 .0 601 1.4 6163 .0 6187 .5 7116 .6  949.. 5 1 183. .7 260.. 1 28 . 1 838. 6 1363..6 66 . 1 939 .4 . 1430 .6 1252 .9 . 1607 .7 186..4 1731 .7 68. 8 1391 . 7 1618. 6 1928 .8 157..7 2444 .2 . 2182 .9 673..3 87 .9 , 2844 , 5 3315 .4 1499..3 4851 , . 7 4073 .8 1234 .4 . 5354 . 2 4999 .8 326 .5 4986..0 5734 .4 7421 . 3 6306 .5 2452..0 207 .5 . 5325.. 1 6703 .0 5841 . 3 6736..5 650..5 5996..7 7345 .0 459..0 7883 . 4 602 .9 6385..9 1548 .0 7559 .4 8358 .3 9142 .6 2382 . 7 8545..7 5472 . 3 1 1659.8 9143 .6 2671 .4 9788 .0 10569 . 2  226758,. 1 124623,.9 161292,. 3 214447,.6 253011 . 1 283043,. 1 278981 .9 , 3681 17,.9 308836 . 1 462123 .9 411 164,. 2 371419 .4 420846 . 1 431337,.8 475172 . 1 614905,. 1 767047 .8 868477 .6 915743 .8 986038 .6  25. 2 27 ,9 , 28 ,5 28..3 28 .5 28..5 28 .9 29. 8 30,.6 31 ,0 . 31 ,4 31 ,6 32,.0 32..6 33,.2 34..0 35,.2 36,.5 38,.0 39.. 7  3748 .0  1051 . 7  447169 .0  31 .6  INTERMEDIATE 6138 .8 MEANS  5729..4 4988. 2 5251 .6 6339 .6 6317.. 5 7321 . 1 9883 .0 11420. 5 18212 .7 20110..3 19302., 2 27716..8 21595,.3 22659.,5 23506..0 24959..7 28142..9 31576 .2 38462..9 33351,.6  2735., 7 1032 . 1 2905.,0 100,.7 3064. 2 231 .9 , 3768 ,6 . 658 .7 4641 . .8 241 .4 , 5126..0 553 , .3 6127..7 2329 .8 295 , 9250..3 .0 10955..6 4899 , .7 13289., 7 3981 .9 14839., 2 1039 .8 . 15725..6 7759 . .5 15992., 5 648 . .4 15922.. 7 1995 .4 . 16679..8 1382 .5 . 17008..8 1773..2 4397 .7 17077,.8 16884,.9 6527 .9 16282,.9 14400 .8 17925,.9 6729 .O  •  4799 . 7  5087 .0  18342 . 3 11310,. 2  3048 .9  SUMMARY  OF  FOREST  ADMINISTRATION  ALBERTA -  YEAR  A  1970-71 1971-72 1972-73 1973-74 1974-75 1975-76 1976-77 1977-78 1978-89  TOTAL FOREST SERVICE B  -  OPERN CODES  TOTAL ALBERTA BUDGET  CNSUM PRICE INDEX  ADUST FOREST SERVICE  ADUST PREFIRE  ADUST FIRE FIGHT  ADUST TOTAL FIRE  ADUST OPERN CODES  ADUST ALBERTA BUDGET  c  D  E  F  G  H  I  U  K  L  M  N  123840. 0 153247. 0  30114. 0 31469. 0  INTERMEDIATE 46984. 9  9 5 5 3 .. 9  $1000  1982-83  TOTAL FIRE  12475. 37182. 60128. 75241.  TOTAL MEANS  X  COSTS  FIRE FIGHT  6762 .0 7471 . 0 7620. 0 9303 .0 10997. 0 11454. 0 11954. 0 13032. 0 13246. 0 15387. 0 19419. 0  MEANS  COSTS  TO  PROTECTION  FIRE PRESUPPR  15280. 0 17729. 0 16765. 0 17154. 0 25771 .0 26594. 0 27542. 0 28237. 0 34445. 0 45979. 0 78221 .0  1979-80 1980-81 1981-82 1982-83  1905 06  AND  4546. 0 5654. 0 4896. 0 2367 .0 6101 .0 4373. 0 3386. 0 2645. 0 5244. 0  11308. 13125. 12516. 11670. 17098. 15827. 15340. 15677.  0 0 0 0 0 0 0 0 0 0 0  10734. 0 12075. 0 11869. 0 14787. 0 19670. 0 22221 . 0 24156. 0 25592. 0  1128628. 0 1266730. 0 1370889..0 1504000. 0 2076903. 0 2 7 2 0 7 5 1 .,0 2920034. ,0 3396885. .0 3768749.,0  4 1 .. 0 42..2 44..2 47 . 6  37268.. 3 4 2 0 1 1 ,.9 37929,.9 36037. 8  52,.8 58 .5 62 .9 . 67 .9 73 .9 8 0 .7 88 .9  4 8 8 0 8 , .7 45459 .8 43787 .0 41586 .2 46610 .3 56975,.2 8 7 9 8 7 ,.6  19579 .5 19004 .8 19192 .9 17924 . 2 19066,.9 2 1 8 4 3 ,.6  100 .0 1 10 . 8  123840 .0 138309 .6  0 0  18490. 27862. 56601. 90242. 0 106710. 0  29201. 0 33504. 0 41039. 0 63712. 0 78006. 0  4 5 3 1 3 5 2 . ,0 5548605. ,0 7 0 4 3 2 8 3 . .0 8998773. 0  14479. 1  17249. 1  31728. 2  29735. 8  3559657. .0  67 .0  5 1 4 2 ,. 4  4 7 4 6 . ,5  6648.. 2  6 3 8 9 . .6  1078219..0  28 .6  0 0  16492..7 17703,,8 17239,.8 19544,, 1 20827 6  1 1 0 8 7 ..8 13398,, 1 11076..9 4972 . 7  2 7 5 8 0 . .5 31101. 9 2 8 3 1 6 .7 24516. a  26180 . 5 28613,. 7 26852,.9 31065.. 1  11554. 9 7 4 7 5 ..2 5383., 1 3 8 9 5 ,.4  3 2 2 2  3 7 2 5 3 , .8 3 7 9 8 4 . .6  2 3 8 2 . ,6 7054.. 7 4 3 8 7 . ,9 3 0 8 8 , .4  2752751 .0 3001730 .0 3101559 .0 3159663 .0 3933528 .0 4650856 .0 4642344 .0  15458,,5 41824,, 5  2 5 0 2 0 , .3 3 4 5 2 5 , .4 63668., 2  3 8 4 0 3 ,,8 37690,. 7 39514 . 2 41516 . 7 46163 . 1  3 0 1 14 . 0 28401 .6  60128,.0 67907,.0  90242,,0 9 6 3 0 8 , ,6  63712,.0 70402,, 5  5002776 .0 5099796 .0 5615058 .0 6241400 .0 7043283 .0 8121637 .0  60508 .6  20533 .5  2 0 0 9 6 .8  40630 .3  4041 1 .8  4797410  15639 .0  9925 .2  6365 .5  1 1 195 . 1  1 1395 . 3  1807667 .0  7096,. 1  0  ON —1  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items