Dissertations / Theses on the topic 'Wildfire simulation'

[Truncated abstract] In this thesis, I address the modelling and computer simulation of spatial, eventdriven systems from a computer science perspective. Spatially explicit models of wildland fire (wildfire) behaviour are addressed as the specific application domain. Wildfire behaviour is expressed as a formal model and the associated simulations are compared to existing models and implementations. It is shown that the in- teracting spatial automata formalism provides a general framework for modelling spatial event-driven systems and is appropriate to wildfire systems. The challenge adressed is that of physically realistic modelling of wildfire behaviour in heterogeneous environments . . . Many current models do not incorporate the influence of a neighbourhood (the geometry of the fire front local to an unburnt volume of fuel, for example), but rather determine the propagation of fire using only point information. Whilst neighbourhood-based influence of behaviour is common to cellular automata theory, its use is very rare in existing models of wildfire models. In this thesis, I present the modelling technique and demonstrate its applicability to wildfire systems via a series of simulation experiments, where I reproduce known spatial wildfire dynamics. I conclude that the interacting spatial automata formalism is appropriate as a basis for constructing new computer simulations of wildfire spread behaviour. Simulation results are compared to existing implementations, highlighting the limitations of current models and demonstrating that the new models are capable of greater physical realism.

Wildfires have significant impact on both ecosystems and human society. To effectively manage wildfires, simulation models are used to study and predict wildfire spread. The accuracy of wildfire spread simulations depends on many factors, including GIS data, fuel data, weather data, and high-fidelity wildfire behavior models. Unfortunately, due to the dynamic and complex nature of wildfire, it is impractical to obtain all these data with no error. Therefore, predictions from the simulation model will be different from what it is in a real wildfire. Without assimilating data from the real wildfire and dynamically adjusting the simulation, the difference between the simulation and the real wildfire is very likely to continuously grow. With the development of sensor technologies and the advance of computer infrastructure, dynamic data driven application systems (DDDAS) have become an active research area in recent years. In a DDDAS, data obtained from wireless sensors is fed into the simulation model to make predictions of the real system. This dynamic input is treated as the measurement to evaluate the output and adjust the states of the model, thus to improve simulation results. To improve the accuracy of wildfire spread simulations, we apply the concept of DDDAS to wildfire spread simulation by dynamically assimilating sensor data from real wildfires into the simulation model. The assimilation system relates the system model and the observation data of the true state, and uses analysis approaches to obtain state estimations. We employ Sequential Monte Carlo (SMC) methods (also called particle filters) to carry out data assimilation in this work. Based on the structure of DDDAS, this dissertation presents the data assimilation system and data assimilation results in wildfire spread simulations. We carry out sensitivity analysis for different densities, frequencies, and qualities of sensor data, and quantify the effectiveness of SMC methods based on different measurement metrics. Furthermore, to improve simulation results, the image-morphing technique is introduced into the DDDAS for wildfire spread simulation.

Each year many forest fires have occurred and huge amount of forest areas in each country have been lost. Turkey like many world countries have forest fire problem. 27 % of Turkey&
#65533
s lands are covered by forest and 48 % of these forest areas are productive, however 52 % of them must be protected. There occurred 21000 forest fires due to several reasons between 1993 and 2002. It is estimated that 23477 ha area has been destroyed annually due to wildfires. The fire management strategies can be built on the scenarios derived from the simulation processes. In this study a GIS &
#65533
based fire simulating model is used to simulate a past fire occurred in Marmaris &
#65533
Ç
etibeli, Turkey, in August 2002. This model uses Rothermel&
#65533
s surface fire model, Rothermel&
#65533
s and Van Wagner&
#65533
s crown fire model and Albini&
#65533
s torching tree model. The input variables required by the model can be divided into four groups: fuel type, fuel moisture, topography and wind. The suitable fuel type classification of the vegetation of the study area has been performed according to the Northern Forest Fire Laboratory (NFFL) Fuel Model. The fuel moisture data were obtained from the experts working in the General Directorate of Forestry. The fire spread pattern was derived using two IKONOS images representing the pre- and post-fire situations by visual interpretation. Time of arrival, the rate of spread and the spread direction of the fire were obtained as the output and 70 % of the burned area was estimated correctly from the fire simulating model.

Wildfires pose serious threats to the society and environment. Simulation of wildfire spread and fire suppression remains a challenging task due to the complexity of wildfire behavior and fire suppression tactics. In previous work, a wildfire spread and suppression simulation model called DEVS-FIRE has been developed. Based on that model, this thesis develops a graphic user interface to support an interactive simulation environment for surface wildfire spread and suppression simulation. The developed environment allows users to dynamically set up fire spread simulations, and to interacticaly deploy firefighting agents to experiment different fire suppression tactics. This graphic user interface is implemented using the Java Swing framework, and is intergrated with the DEVS-FIRE model in a well-designed manner. The software architecture is described and the simulation environment and experiment results with different fuel, terrain and weather data are presented.

Simulation software is being increasingly used in various domains for system analysis and/or behavior prediction. Traditionally, researchers and field experts need to have access to the computers that host the simulation software to do simulation experiments. With recent advances in cloud computing and Software as a Service (SaaS), a new paradigm is emerging where simulation software is used as services that are composed with others and dynamically influence each other for service-oriented simulation experiment on the Internet. The new service-oriented paradigm brings new research challenges in composing multiple simulation services in a meaningful and correct way for simulation experiments. To systematically support simulation software as a service (SimSaaS) and service-oriented simulation experiment, we propose a layered framework that includes five layers: an infrastructure layer, a simulation execution engine layer, a simulation service layer, a simulation experiment layer and finally a graphical user interface layer. Within this layered framework, we provide a specification for both simulation experiment and the involved individual simulation services. Such a formal specification is useful in order to support systematic compositions of simulation services as well as automatic deployment of composed services for carrying out simulation experiments. Built on this specification, we identify the issue of mismatch of time granularity and event granularity in composing simulation services at the pragmatic level, and develop four types of granularity handling agents to be associated with the couplings between services. The ultimate goal is to achieve standard and automated approaches for simulation service composition in the emerging service-oriented computing environment. Finally, to achieve more efficient service-oriented simulation, we develop a profile-based partitioning method that exploits a system’s dynamic behavior and uses it as a profile to guide the spatial partitioning for more efficient parallel simulation. We develop the work in this dissertation within the application context of wildfire spread simulation, and demonstrate the effectiveness of our work based on this application.

Les feux de forêt sont des incendies de végétation incontrôlés qui causent des dégâts importants à l’environnement, aux biens et aux personnes. Les actions de lutte contre de tels feux sont risqués et peuvent par conséquent bénéficier de techniques d'automatisation pour réduire l’exposition humaine. La télédétection aérienne est une technique qui permet d’obtenir des informations précises sur l’état d'un feu de forêt, afin que les équipes d’intervention puissent préparer des contre-mesures. Avec des véhicules aériens habités, elle expose les opérateurs à des risques élevés, qui peuvent être évités par l’utilisation de véhicules autonomes. Cette thèse présente un système de surveillance de feux de forêt basé sur des flottes de véhicules aériens sans pilote (UAV) afin de fournir aux pompiers des renseignements précis et à jour sur un feu de forêt. Nous présentons une approche pour planifier les trajectoires d’une flotte de drones à voilure fixe afin d’observer un feu de forêt évoluant au fil du temps. Des modèles réalistes du terrain, du processus de propagation du feu et des drones sont exploités, ainsi qu’un modèle du vent, pour prédire la propagation des feux de forêt et planifier le mouvement des drones. L’approche présentée adapte une méthode générique de recherche à voisinage variable (VNS) à ces modèles et les contraintes associées. L’exécution de la mission d’observation planifiée fournit des cartes des feux de forêt qui sont transmises à l’équipe d’intervention et exploitées par l’algorithme de planification pour déterminer de nouvelles trajectoires d’observation. Les algorithmes et les modèles sont intégrés dans une architecture logicielle permettant l’exécution dans des scénarios avec différents niveaux de réalisme, avec des drones réels et simulés survolant un feu de forêt réel ou synthétique. Les résultats de simulation mixte montrent la capacité de planifier les trajectoires d’observation d’une petite flotte de drones et de mettre à jour les plans lorsque de nouvelles informations sur l’incendie sont incorporées dans le modèle de propagation de feu
Wildfires, also known as forest or wildland fires, are uncontrolled vegetation fires occurring in rural areas that cause tremendous damage to the society, harming environment, property and people. The firefighting endeavor is a dull, dirty and dangerous job and as such, can greatly benefit from automation to reduce human exposure to hazards. Aerial remote sensing is a common technique to obtain precise information about a wildfire state so fire response teams can prepare countermeasures. This task, when performed with manned aerial vehicles, expose operators to high risks that can be eliminated by the use of autonomous vehicles. This thesis introduces a wildfire monitoring system based on fleets of unmanned aerial vehicles (UAVs) to provide firefighters with timely updated information about a wildland fire. We present an approach to plan trajectories for a fleet of fixed-wing UAVs to observe a wildfire evolving over time. Realistic models of the terrain, of the fire propagation process, and of the UAVs are exploited, together with a model of the wind, to predict wildfire spread and plan UAV motion. The approach tailors a generic Variable Neighborhood Search method to these models and the associated constraints. The execution of the planned monitoring mission provides wildfire maps that are transmitted to the fire response team and exploited by the planning algorithm to plan new observation trajectories. Algorithms and models are integrated within a software architecture allowing for execution under scenarios with different levels of realism, with real and simulated UAVs flying over a real or synthetic wildfire. Mixed-reality simulation results show the ability to plan observation trajectories for a small fleet of UAVs, and to update the plans when new information on the fire are incorporated in the fire model

Die Option Mechanism Dynamics (MDO) von Creo Parametric ist ein Tool aus dem Bereich Starrkörperdynamik. Im Kern des Vortrags steht jedoch die MDO-Funktion 3D-Kontakt, die einzigartig mit elastischen Körpern arbeitet. Im Vortrag wird die Vorgehensweise bei der dynamischen Simulation eines Rutschreifens auf GFK-Wasserrutschen erläutert, abschließend werden Animationsbeispiele präsentiert.

There is growing recognition of the importance of preserving Canada’s forests. Canada’s 348 million hectares of forest land cover 35% of its land area, representing 9% of the world’s forests and 24% of the world’s boreal forests. As a renewable resource, forests offer significant environmental, economic and recreational benefits and innumerable services contributing to the quality of life. Canada has recently entered an era of increased frequency and severity of natural disasters. Ecosystems and communities especially in western Canada have recently undergone a trend of increasing pressures from natural disturbances. These disturbances include wildfires associated with increased fuel load levels from past fire suppression regimes and a widely spread infestation of the mountain pine beetle in addition to changes in weather patterns. Wildfire activity has reached extreme levels in many of the recent years. This thesis profiles an area of western Canada within the Montane Cordillera covering the Nechako Lakes Electoral District in central British Columbia and assesses its vulnerability to the specific hazard of wildfires caused by natural and man-made sources. The objectives of this research are to review, simulate and assess the impact of various fuel management strategies in a sub-section of the Nechako Lakes Electoral District called the Southside. Values at risk include private property and old growth forest in respectively timber supply areas, provincial parks, woodlots and community forests. Simulation results show that firebreaks are effective in significantly reducing the area burned in different parts of the landscape. The performance of different strategies shows large variation. Although this has not been investigated further, such variation has likely been caused by topographic aspects and the positioning of firebreaks in the landscape in relation to climatic parameters. These results can therefore not be extrapolated beyond the simulated area, but do give an indication of the performance variation that may be expected when similar firebreaks are applied elsewhere. The results also show that model performance of all firebreak strategies is heavily and fairly consistently influenced by weather stream parameters. Sensitivity analyses of weather stream parameters show that although the reduction in total area burned varies, the ranking between strategies in their overall performance is consistent regardless of the weather pattern. Combined dry, warm and windy weather conditions lead to a 3.44-fold increase in total area burned as compared to the scenario with average weather conditions. In favourable weather conditions represented by wet, cold and nearly windless conditions, the model shows an 85% reduction in total burned area as compared to the average scenario. These results illustrate the significant impact of uncontrollable variables on the overall result.

Vortrag über die Vorteile konstruktionsbegleitender Wärmesimulation direkt durch den Konstrukteur am nativen CAD-System. Die Elinter AG als Spezialist für Wärmesimulationen und Strömungssimulationen empfiehlt FloEFD zusammen mit Pro/ENGINEER Wildfire. Schnelle thermische Bewertung von Konstruktionsvorschlägen und Designvarianten direkt durch den Konstrukteur. Die frühzeitige Wärmesimulation bzw. Strömungssimulation direkt im Desginprozess ist mit minimalem Aufwand möglich. Simulationen mit den nativen CAD Daten direkt und assoziativ in Pro/ENGINEER Wildfire geben den entscheidenden Zeitvorteil und machen thermische Betrachtungen innerhalb weniger Minuten möglich. So können neue Konzepte oder Geräte von Anfang an auf optimale Leistung bei niedrigsten Kosten ausgelegt werden. Teure Überraschungen bei Prototypen oder im Werkzeugbau werden vermieden. Mit frühzeitigen Simulationen bauen Sie ein tiefes Verständnis über Ihr eigenes Produkt auf und sichern den Vorsprung gegenüber von Mitbewerbern. FloEFD ist speziell für die Anforderungen in der frühen Produktentwicklung konzipiert und erlaubt Konstrukteuren schnelle und sichere Aussagen über ihre Konstruktion auch wenn sie sich bisher nicht oder nur sporadisch mit Strömung und Wärme befassen. FloEFD rechnet Wärme, Strahlung, Konvektion, Wärmeleitung, Erwärmung durch elektrischen Strom, Strömung von Gasen und Flüssigkeiten und sogar mechanische Verformungen aufgrund ungleichmässiger Temperaturverteilungen und Druckverteilungen. FloEFD ist vollständig in Pro/ENGINEER Wildfire integriert und verfügt zudem über eine Anbindung an Pro/Mechanica. Mehr Iterationen in kürzerer Zeit bei niedrigeren Kosten. FloEFD/Pro machts möglich. Tryout bei www.elinter.ch oder email an sven.klett@elinter.ch

La vulnérabilité est la composante du risque d’incendie de forêt la moins connue. Elle est généralement évaluée à dires d’experts. Plus objectivement elle peut être évaluée a posteriori en mesurant les dommages causés par un sinistre, si le détail des caractères de ce sinistre est connu. Nous proposons un modèle d’évaluation de la vulnérabilité formulé par une analyse multicritères des dires d’experts. Ce modèle est validé en utilisant un modèle physique d’exposition. Ses entrées sont fournies par un modèle de propagation. La calibration se fonde sur une analyse des dommages engendrés par un sinistre réel. Les résultats valident l’approche de modélisation de la vulnérabilité par des variables spatiales
Vulnerability is not a well-known component of forest fire risk. It is usually assessed through experts’ opinions. It can be assessed more objectively after a disaster par measuring damages, if the attributes of the disaster are known. We propose a model for vulnerability assessment formulated with a multi-criteria analysis of experts’ opinions. This one is validated by using a physical model for exposure assessment. Its inputs are provided by a fire propagation model. The system is calibrated based on the analysis of damages induced by a real wildfire. Results demonstrate the consistency of a vulnerability model based on spatial variables

Thesis (M.S.)--University of Missouri--Rolla, 2007.
Vita. The entire thesis text is included in file. Title from title screen of thesis/dissertation PDF file (viewed November 16, 2007) Includes bibliographical references (p. 46-47).

It has been well established that under certain circumstances wildfire is capable of producing water repellent or hydrophobic soils. Hydrophobic soils can dramatically alter runoff and erosion processes and as such have been the subject of considerable research activity. Wildfires in chaparral vegetation are recognized as being particularly susceptible to hydrophobic soil development. A comparison of chaparral fire soil heat profiles from DeBano (1989) and Weirich (unpublished) indicates that under higher fire intensity situations in chaparral a different soil heating mechanism other than just conduction heating may be at work. In contrast to the slow moving low temperature increases expected in conduction heating a much faster heat pulse resulting in more rapid temperature rises and higher temperatures at depth can also occur in chaparral wildland fires. This suggests that a better understanding of the heat transfer processes that occur at extreme fire intensities is both important and is needed. The specific aim of this study was to observe heat flow under a variety of particle sizes using a laboratory based wildfire simulator operating at intensities and durations similar to those experienced in chaparral wildfires. The wildfire simulator system consisted of a propane burner array, an array of thermocouples to measure temperatures at varying locations and depths, and a data logging system to record the results of the heating experiments. Using the simulator homogenous sand, silt, clay, and heterogeneous clay loam were subjected to 600ºC, 900ºC, and 1200ºC peak intensities with two different heating durations or treatments (H1 and H2). The heating levels and durations used were based on data from field based chaparral fire experimental temperature data previously collected by Weirich (unpublished). The system design allowed the user to control the intensity and duration of the heat treatments and the thermocouple sensor arrays measured temperatures at the flame to a soil depth of 15cm. The apparatus and experimental treatments allowed for the investigation of peak heat intensity, heat duration, slope, and most importantly particle size on heat transfer processes. The higher soil temperatures at depth, shorter times to peak temperatures at depth, and observed temperature spiking seen during some of the simulator experimental runs (specifically with respect to larger particle sizes such as sand) call into question the view that slow moving conduction may not be the only soil heat transfer process at work in high fire intensity situations such as those seen in chaparral wildfires and in particular chaparral wildfire underlain by larger particle sizes fractions such as sand.

The Chihuahua desert is one of the most biologically diverse ecosystems in the world, but suffers serious degradation because of changes in fire regimes resulting in large catastrophic fires. My study was conducted in the Sierra La Mojonera (SLM) natural protected area in Mexico. The purpose of this study was to implement the use of FARSITE fire modeling as a fire management tool to develop an integrated fire management plan at SLM. Firebreaks proved to detain 100% of wildfire outbreaks. The rosetophilous scrub experienced the fastest rate of fire spread and lowland creosote bush scrub experienced the slowest rate of fire spread. March experienced the fastest rate of fire spread, while September experienced the slowest rate of fire spread. The results of my study provide a tool for wildfire management through the use geospatial technologies and, in particular, FARSITE fire modeling in SLM and Mexico.

Forest Park, a 5,000 acre heavily-forested park within the city limits of Portland, Oregon was selected as the study area for performing a fire simulation analysis. A well-documented fire swept over a large area of the park in 1951, and provides both direct inputs, including the ignition point, and context for the present day fire simulations. The goal of the research was two fold. First, determine the difference between small area simulations using standard and custom surface fuel models. Second, determine if fire simulation can be an effective tool in assessing fire danger and behavior in a wildland-urban interface environment like Forest Park.

Ce travail de recherche est dédié à l’étude numérique du comportement des feux se propageant librement à travers un couvert végétal homogène (depuis l’ignition jusqu’à la propagation), à l'aide du modèle "FireStar3D". Différentes configurations sont abordées dans cette thèse : les feux de litière réalisés à l’échelle du laboratoire en milieu confiné (dans le tunnel à feu du laboratoire de Missoula), et à plus grande échelle, les feux de prairie où deux types de ligne d’allumage ont été considérés (ligne de longueur finie et quasi-infinie). Les simulations ont été réalisées à l'aide de deux modèles de turbulence : le modèle k- des équations de transport moyennées (approche RANS) et la simulation numérique des grandes structures (LES). La comparaison avec les données expérimentales concerne principalement la vitesse de propagation du feu, l'intensité du feu, la fréquence des fluctuations des flammes et la longueur d'onde caractérisant la structuration en crête et en creux du front de flammes dans la direction transversale. Les résultats numériques ont mis en évidence la compétition entre les forces de flottaison et les forces d'inertie du vent dans la détermination du comportement du feu, ainsi que la similarité dynamique du front de flammes (intensité et structuration 3D) à petite et grande échelles. Cette thèse a été également menée dans le cadre du développement et de la validation du modèle "FireStar3D"
This research work aims to numerically study the behavior of a fire front propagating through a homogeneous solid-fuel layer, using "FireStar3D" model. Laboratory fire experiments have been reproduced numerically and grassland fires have been simulated with finite and quasi-infinite firelines. The simulations were carried out using both Reynolds-averaged Navier–Stokes equations approach (RANS) and Large Eddy Simulation approach (LES). The comparison with the experimental data concerned mainly the Rate of Spread (ROS) of fire, the fireline intensity, the frequency of flames fluctuation, and the wavelength characterizing the crest-and-trough structure of the fire front along the transverse direction. The numerical results highlighted the competition between buoyancy forces and the wind inertial forces in governing the fire behavior, as well as the similarity of fire-front dynamics (intensity and 3D structuration) at small and large scales.This research work was also carried out in the context of developing and validating "FireStar3D" model. The level of details in the physical modeling, the properties of the used numerical method, and the good agreement obtained with the experimental and numerical data reported in the literature, all-together place "FireStar3D" in a good position at an international level among other numerical tools used to study the behavior of wildfires

As sensor data becomes more and more available, there is an increasing interest in assimilating real time sensor data into spatial temporal simulations to achieve more accurate simulation or prediction results. Particle Filters (PFs), also known as Sequential Monte Carlo methods, hold great promise in this area as they use Bayesian inference and stochastic sampling techniques to recursively estimate the states of dynamic systems from some given observations. However, PFs face major challenges to work effectively for complex spatial temporal simulations due to the high dimensional state space of the simulation models, which typically cover large areas and have a large number of spatially dependent state variables. As the state space dimension increases, the number of particles must increase exponentially in order to converge to the true system state. The purpose of this dissertation work is to develop localized particle filtering to support PFs-based data assimilation for large-scale spatial temporal simulations. We develop a spatially dependent particle-filtering framework that breaks the system state and observation data into sub-regions and then carries out localized particle filtering based on these spatial regions. The developed framework exploits the spatial locality property of system state and observation data, and employs the divide-and-conquer principle to reduce state dimension and data complexity. Within this framework, we propose a two-level automated spatial partitioning method to provide optimized and balanced spatial partitions with less boundary sensors. We also consider different types of data to effectively support data assimilation for spatial temporal simulations. These data include both hard data, which are measurements from physical devices, and soft data, which are information from messages, reports, and social network. The developed framework and methods are applied to large-scale wildfire spread simulations and achieved improved results. Furthermore, we compare the proposed framework to existing particle filtering based data assimilation frameworks and evaluate the performance for each of them.

Zhao Yibin.
Thesis submitted in: November 2001.
Thesis (M.Phil.)--Chinese University of Hong Kong, 2002.
Includes bibliographical references (leaves 103-111).
Abstracts in English and Chinese.
Abstract --- p.I
ACKNOWLEDGEMENT --- p.V
Table of Contents --- p.VIII
List of Tables --- p.IX
List of Figures --- p.X
Chapter Chapter 1 --- Introduction --- p.1
Chapter 1.1 --- Research background and problem statement --- p.1
Chapter 1.2 --- Research objectives --- p.6
Chapter 1.3 --- Methodology --- p.7
Chapter 1.4 --- Significance of this study --- p.9
Chapter 1.5 --- Organization of the thesis --- p.10
Chapter Chapter 2 --- "Literature review: wildfire behavior simulation, Web GIS and public participation GIS" --- p.11
Chapter 2.1 --- Introduction --- p.11
Chapter 2.2 --- Investigating wildfire behavior --- p.12
Chapter 2.3 --- Modeling wildfire with GIS --- p.20
Chapter 2.4 --- Emergence of the Web GIS --- p.27
Chapter 2.5 --- New agenda from public participation --- p.28
Chapter 2.6 --- Summary --- p.31
Chapter Chapter 3 --- System design: requirements analysis and feasibility analysis --- p.34
Chapter 3.1 --- Introduction --- p.34
Chapter 3.2 --- Analysis of functions requirement --- p.35
Chapter 3.3 --- A host of solutions --- p.41
Chapter 3.4 --- Summary --- p.52
Chapter Chapter 4 --- Simulating the wildfire --- p.53
Chapter 4.1 --- Physical Setting of experiment area and data preparation --- p.53
Chapter 4.2 --- Adaptation and formularization of the Rothermel's fire behavior model --- p.60
Chapter 4.3 --- Fire spreading algorithm --- p.66
Chapter 4.4 --- Defining wildfire with Object Oriented Design (OOD) method --- p.71
Chapter 4.5 --- Summary --- p.74
Chapter Chapter 5 --- Participation process with interactive tools empowered by IT technologies --- p.76
Chapter 5.1 --- Comprehending the problem in an interactive way --- p.76
Chapter 5.2 --- Performing wildfire simulation --- p.81
Chapter 5.3 --- Submitting of end users comments --- p.84
Chapter 5.4 --- Discussion bulletin board --- p.94
Chapter 5.5 --- Summary --- p.96
Chapter Chapter 6 --- Discussions and conclusions --- p.98
Chapter 6.1 --- Research limitations and discussions --- p.98
Chapter 6.2 --- Conclusions --- p.99
BIBLIOGRAPHY --- p.103
Appendix 1 .Defining MapService with ArcXML --- p.112
Appendix 2.Defining MapNotes with ArcXML --- p.112

O flagelo dos incêndios é um acontecimento que tem vindo a aumentar nos últimos anos em todo o mundo, nomeadamente em Portugal. De 2015 a 2017, a área ardida multiplicou-se oito vezes, sendo que as áreas mais afetadas estão principalmente no Norte e Centro de Portugal. Para que os bombeiros possam atuar com mais rapidez, eficiência e eficácia, é necessário aprimorar as técnicas atuais de combate a incêndios. Dito isto, a utilização de ferramentas e técnicas para prever a forma como o fogo se propaga sem colocar vidas humanas em perigo, é uma solução vantajosa em muitos aspectos, nomeadamente em termos de segurança para a população e economicamente viável. O processo utilizado durante este trabalho é descrito desde a recolha e análise dos dados, passando pelo processamento dos dados e respetivo desenvolvimento do programa, terminando na análise dos resultados. Este documento apresenta uma proposta de simulador de risco de incêndio e área ardida baseado em técnicas de Machine Learning, mais precisamente numa abordagem de Deep Learning utilizando dados de eventos de incêndios anteriores em Portugal conjugados com dados geográficos e climáticos. A solução apresentada revela uma boa capacidade de predição da área ardida especialmente em relação à percentagem de área ardida, revelando uma boa correspondência entre a percentagem prevista e a real.

Climate exerts considerable control on wildfire regimes, and climate and wildfire are both major drivers of forest growth and succession in interior Northwest forests. Estimating potential response of these landscapes to anticipated changes in climate helps researchers and land managers understand and mitigate impacts of climate change on important ecological and economic resources. Spatially explicit, mechanistic computer simulation models are powerful tools that permit researchers to incorporate climate and disturbance events along with vegetation physiology and phenology to explore complex potential effects of climate change over wide spatial and temporal scales. In this thesis, I used the simulation model FireBGCv2 to characterize potential response of fire, vegetation, and landscape dynamics to a range of possible future climate and fire management scenarios. The simulation landscape (~43,000 hectares) is part of Deschutes National Forest, which is located at the interface of maritime and continental climates and is known for its beauty and ecological diversity. Simulation scenarios included all combinations of +0��C, +3��C, and +6��C of warming; +10%, ��0%, and -10% historical precipitation; and 10% and 90% fire suppression, and were run for 500 years. To characterize fire dynamics, I investigated how mean fire frequency, intensity, and fuel loadings changed over time in all scenarios, and how fire and tree mortality interacted over time. To explore vegetation and landscape dynamics, I described the distribution and spatial arrangement of vegetation types and forest successional stages on the landscape, and used a nonmetric multidimensional scaling (NMS) ordination to holistically evaluate overall similarity of composition, structure, and landscape pattern among all simulation scenarios over time. Changes in precipitation had little effect on fire characteristics or vegetation and landscape characteristics, indicating that simulated precipitation changes were not sufficient to significantly affect vegetation moisture stress or fire behavior on this landscape. Current heavy fuel loads controlled early fire dynamics, with high mean fire intensities occurring early in all simulations. Increases in fire frequency accompanied all temperature increases, leading to decreasing fuel loads and fire intensities over time in warming scenarios. With no increase in temperature or in fire frequency, high fire intensities and heavier fuel loads were sustained. Over time, more fire associated with warming or less fire suppression increased the percentage of the landscape occupied by non-forest and fire-sensitive early seral forest successional stages, which tended to increase the percentage of fire area burning at high severity (in terms of tree mortality). This fire-vegetation relationship may reflect a return to a more historical range of conditions on this landscape. Higher temperatures and fire frequency led to significant spatial migration of forest types across the landscape, with communities at the highest and lowest elevations particularly affected. Warming led to an upslope shift of warm mixed conifer and ponderosa pine (Pinus ponderosa) forests, severely contracting (under 3�� of warming) or eliminating (under 6�� of warming) area dominated by mountain hemlock (Tsuga mertensiana) and cool, wet conifer forest in the high western portion of the landscape. In lower elevations, warming and fire together contributed to significant expansion of open (<10% tree canopy cover) forest and grass- and shrubland. The compositional changes and spatial shifts simulated in the warming scenarios suggest that climate change is likely to significantly affect forests on this landscape. Warming and associated fire also tended to increase heterogeneity of forest structural stages and landscape pattern, resulting in a more diverse distribution of structural stages, especially in lower elevations, and a more divided landscape of smaller forest stands. The NMS ordination emphasized the dissimilarity between the severe +6�� scenarios and the other two temperature scenarios. The +0�� and +3�� scenarios differed from each other in composition (mainly because cool forest was lost in the +3�� scenarios), but within a given level of fire suppression they remained remarkably similar in terms of overall composition, structure, and landscape pattern, while the +6�� scenarios separated noticeably from them. Such decisive differences suggest that under the simulated ranges of precipitation and fire suppression, the interval between 3 and 6 degrees of warming on this landscape may capture an ecological threshold, or tipping point. Additional simulation research that incorporates (for example) management actions, insects and pathogens, and a wider array of precipitation scenarios could help illuminate more clearly the possible range of future landscape conditions. Still, these results provide a glimpse of potential divergent outcomes on this important landscape under possible future climates, and suggest that these forests will undergo considerable changes from both historical and current conditions in response to higher temperatures expected in this area. Some changes may be inevitable with warming, such as the upslope shift of warm forest types, but careful planning for fire and fuels management might allow land managers to modulate fire behavior and steer vegetation dynamics toward the most desirable outcome possible.
Graduation date: 2013