University of Montana ScholarWorks at University of Montana Graduate Student Theses, Dissertations, & Professional Papers Graduate School 1999 Modeling the initiation and spread of crown fires Miguel Gomes da Cruz The University of Montana Follow this and additional works at: https://scholarworks.umt.edu/etd Let us know how access to this document benefits ou.y Recommended Citation Gomes da Cruz, Miguel, "Modeling the initiation and spread of crown fires" (1999). Graduate Student Theses, Dissertations, & Professional Papers. 2996. https://scholarworks.umt.edu/etd/2996 This Thesis is brought to you for free and open access by the Graduate School at ScholarWorks at University of Montana. It has been accepted for inclusion in Graduate Student Theses, Dissertations, & Professional Papers by an authorized administrator of ScholarWorks at University of Montana. For more information, please contact [email protected]. Maureen and Mike MANSFIELD LIBRARY Tlie University of JVEONTANA Permission is granted by the author to reproduce this material in its entirety, provided that this material is used for scholarly purposes and is property cited in published works and reports. ** Please check "Yes" or "No" and provide signature ** Yes, I grant permission K No, I do not grant permission Author's Signature Date /jcui/'S z ^ ^ Any copying for commercial purposes or financial gain may be undertaken only with the author's explicit consent. MODELING THE INITIATION AND SPREAD OF CROWN FIRES By Miguel Gomes da Cruz B.S. Forestry, Coimbra Polytechnic Institute, Portugal, 1994 Presented in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE UNIVERSITY OF MONTANA 1999 Approved by: airman, Board of£xaminers Dean, Graduate School i< Date UMI Number: EP36459 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. UMI OtsMTtation Publishing UMI EP36459 Published by ProQuest LLC (2012). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code uest* ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106- 1346 Gomes da Cruz, Miguel, M.Sc., September 1999 Forestry Modeling the Initiation and Spread of Cro;wn^Fir;^ Director; Ronald H. Wakimoto fj~- / The unknowns in fire phenomenology lead to a simplified empirical approach to build models designed to forecast crown fire initiation and spread. This information is needed to support decision making in a large array of fire management problems.The present study is based on a large fire behavior database from experimental fires in North American fiiel complexes. These fires cover a wide spectrum of fire environment and behavior conditions. Three types of models were develop in this study: A crown fire initiation model, a crovm fire spread model, and models to predict canopy bulk density. Crown fire initiation was modeled through a logistic approach using as independent variables wind speed, fuel strata gap, a surface fuel consumption class and dead fine fuel moisture content. Spread rates for active and passive crown fires were modeled through multiple non-linear regression analysis following physical reasoning. Independent variables used in the crown fire spread models were wind speed, canopy bulk density and dead fine fuel moisture content. Models to predict canopy bulk density in some common fuel complexes in the U.S. were developed by linking foliar biomass equations with stand data from the Forest Inventory and Analysis. Canopy bulk density was modeled as a function of species, stand density and stand basal area. The crown fire initiation rAodel correctly predicted 85 % of the cases in the dataset used for its construction. The active crown fire spread model yield a of 0.61. Comparison of predictions from both fire behavior models against an independent dataset from wildfire crown fire runs revealed good model performance. The crown fire initiation model correctly predicted all the wildfires as crovm fires. The active crown fire spread model yielded a mean absolute percent error of 34 % when compared against the independent dataset. The wide variation in fuel complex structure and fire behavior in datasets used to build the crown fire initiation and spread models gives confidence that the models might work well in fuel complexes different from the original ones, given an adequate description of the physical characteristics of the fuel complex. Keywords: Fire behavior modeling; Crown fire initiation; Crown fire spread; Canopy bulk density; Model evaluation. ii ACKNOWLEDGEMENTS This thesis was made possible by the assistance of many people. I am immeasurably indebted to Marty Alexander by his original ideas, guidance and support given throughout this study. I would like extending my sincere thanks to my advisor Ron Wakimoto for his support, guidance and patience. Thanks also to my other committee members, namely Richard Lane for providing assistance on the statistical portion of the analysis, and Don Potts and Bobbie Bartlette for their support. I want to also thank Kelsey Milner and Hans Zuring for their advise and assistance in the analysis of the canopy bulk density data. Thanks also to Brian Stocks for making available to me some of the unpublished data used in this study. Thanks also to the Canadian Forest Service for inviting me to participate in the 1999 Phase III of the International Crown Fire Modeling Experiment, Fort Providence, NWT. I would like to recognize Domingos Xavier Viegas from the University of Coimbra, Portugal for his encouragement for me to seek an advanced educational degree, and the financial support given by ADAI - University of Coimbra. Financial support is also acknowledged from Funda9ao Luso-Americana para d Desenvolvimento, and Blackfoot Forest Protective Association. Thanks to all my friends who made my stay in Missoula so enjoyable. My sincere and heartfelt thanks to Isabel for her never-ending patience and constant support throughout the period of this study. A todos o meu muito obrigado. iii TABLE OF CONTENTS Abstract ii Acknowledgements iii Table of Contents iv List of Tables vii List of Figures ix Introduction 1 CHAPTER I - REVIEW OF THE STATE-OF-KNOWLEDGE AND MODEL EVALUATION 3 1.1. Objectives 3 1.2. Prediction of crown fire initiation 3 1.3. Prediction of crown fire spread 8 1.4. Model evaluation 18 1.4.1. Conceptual validity 19 1.4.2. Operational validation 21 1.4.2.1. Data validation 23 1.4.2.2. Sensitivity analysis 27 1.4.2.3. Crown fire models comparison 29 1.4.2.4. Predictive validation 32 1.4.2.5. Statistical validation 35 1.5. Conclusions 37 CHAPTER II - LOGISTIC CROWN FIRE INITIATION MODELING 39 2.1. Objectives 39 2.2. Review of pertinent variables influencing crown fire initiation 39 2.3. Methods 46 2.3.1. Database construction 46 2.4. Results 49 2.4.1. Variables analysis 49 2.4.2. Model building 54 2.4.2.1. Methods 54 iv 2.4.2.2. Model results 56 2.4.3. Model evaluation 60 2.5. Conclusions 67 CHAPTER III - CROWN FIRE SPREAD MODELING 69 3.1. Objectives 69 3.2. Review of pertinent variables influencing crown fire spread 69 3.3.Methods 77 3.3.1. Database construction 77 3.4. Results 79 3.4.1. Variables analysis 79 3.5. Model building 87 3.5.1. Modeling active crown fire spread rates 87 3.5.1.1. Modeling results 89 3.5.1.2. Model behavior 91 3.5.1.3. Model evaluation 94 3.5.2. Modeling passive crown fire spread rates 100 3.5.2.1. Modeling results 10^3 3.5.2.2. Model Behavior 104 3.6. Conclusions 106 CHAPTER IV - ASSESSING CANOPY FUEL BULK DENSITIES FOR SOME WESTERN US COMMON CONIFER FUEL COMPLEXES 107 4.1. Introduction 107 4.2. Objectives 108 4.3. Methods 108 4.3.1. Fuel complex selection 108 4.3.2. The Forest Inventory and Analysis data 109 4.3.3. Selection of foliage load equations 110 4.3.4. Procedure for calculating canopy bulk density at plot level Ill 4.4. Results 111 4.4.1. Canopy bulk density variability 112 4.4.2. Canopy bulk density modeling 114 4.5. Conclusions 120 V CHAPTER V - CONCLUDING REMARKS 122 Bibliographic references 129 Appendix 145 vi LIST OF TABLES Table 1. L Reference values used in sensitivity analysis 23 1.2. Forest stand structure for the various experimental fires 25 1.3. Relative sensitivity (RS) of rate of spread output of several crown fires spread models to fine dead fuel moisture content, 10 m windspeed and live fuel moisture content. Reference situation is Kenshoe Lake #5 28 1.4. Relative sensitivity (RS) of rate of spread output of several crown fire spread models to fine dead fuel moisture content, 10 m windspeed and live fuel moisture content. Reference situation is Lily Lake wildfire 29 1.5. Validation parameters for the Rothermel crown fire model 37 2.1. Fuel complexes and fire type distribution in the database used for building the crown fire initiation model 47 2.2. Correlation matrix for significant fire environment and behavior variables in the dataset 49 2.3. Coding scheme of the SFC category design variables 56 2.4. Estimated parameters and statistics for the probabilistic crown initiation model 56 2.5. Estimated parameters and statistics for the probabilistic crown fire initiation model 57 2.6. Classification table for logistic model fire type prediction 58 2.7. Values of input variables used for the evaluation of model behavior 63 3.1.