Thesis (M.S.)--Oregon State University, 2010. / Printout. Includes bibliographical references (leaves 45-54). Also available on the World Wide Web.

Over the past century in the western United States, warming has produced larger and more severe wildfires than previously recorded. General circulation models and their ensembles project continued increases in temperature and the proportion of precipitation falling as rain. Warmer and wetter conditions may change forest successional trajectories by modifying rates of vegetation establishment, competition, growth, reproduction, and mortality. Many questions remain regarding how these changes will occur across landscapes and how disturbances, such as wildfire, may interact with changes to climate and vegetation. Forest management is used to proactively modify forest structure and composition to improve fire resilience. Yet, research is needed to assess how to best utilize mechanical fuel reduction and prescribed fire at the landscape scale. Human communities also exist within these landscapes, and decisions regarding how to manage forests must carefully consider how management will affect such communities. In this work, I analyzed three aspects of forest management at large spatiotemporal scales: (1) climate effects on forest composition and wildfire activity; (2) efficacy of fuel management strategies toward reducing wildfire spread and severity; and, (3) local resident perspectives on forest management. Using a forest landscape model, simulations of forest dynamics were used to investigate relationships among climate, wildfire, and topography with long-term changes in biomass for a fire-prone dry-conifer landscape in eastern Oregon. Under climate change, wildfire was more frequent, more expansive, and more severe, and ponderosa pine expanded its range into existing shrublands and high-elevation zones. There was a near-complete loss of native high-elevation tree species, such as Engelmann spruce and whitebark pine. Loss of these species were most strongly linked to burn frequency; this effect was greatest at high elevations and on steep slopes. Fuel reduction was effective at reducing wildfire spread and severity compared to unmanaged landscapes. Spatially optimizing mechanical removal of trees in areas at risk for high-severity wildfire was equally effective as distributing tree removal across the landscape. Tripling the annual area of prescribed burns was needed to affect landscape-level wildfire spread and severity, and distributing prescribed burns across the study area was more effective than concentrating fires in high-risk areas. I conclude that forest management can be used to reduce wildfire activity in dry-mixed conifer forests and that spatially optimizing mechanical treatments in high-risk areas can be a useful tool for reducing the cost and ecological impact associated with harvest operations. While reducing the severity and spread of wildfire may slow some long-term species shifts, high sub-alpine tree mortality occurred under all climate and fuel treatment scenarios. Thus, while forest management may prolong the existence of sub-alpine forests, shifts in temperature, precipitation, and wildfire may overtake management within this century. The use of PPGIS was useful for delineating the range of forest management preferences within the local community, for identifying areas of agreement among residents who have otherwise polarized views, and for generating modeling inputs that reflect views that may not be obtained through extant official channels for public participation. Because the local community has concerns about the use of prescribed fire, more education and outreach is needed. This may increase public acceptance of the amounts of prescribed fire needed to modify wildfire trajectories under future climate conditions.

No description available.

We present an autonomous driving agent on a simulated highway driving scenario with vehicles such as cars and trucks moving with stochastically variable velocity profiles. The focus of the simulated environment is to test tactical decision making in highway driving scenarios. When an agent (vehicle) maintains an optimal range of velocity it is beneficial both in terms of energy efficiency and greener environment. In order to maintain an optimal range of velocity, in this thesis work I proposed two novel reward structures: (a) gaussian reward structure and (b) exponential rise and fall reward structure. I trained respectively two deep reinforcement learning agents to study their differences and evaluate their performance based on a set of parameters that are most relevant in highway driving scenarios. The algorithm implemented in this thesis work is double-dueling deep-Q-network with prioritized experience replay buffer. Experiments were performed by adding noise to the inputs, simulating Partially Observable Markov Decision Process in order to obtain reliability comparison between different reward structures. Velocity occupancy grid was found to be better than binary occupancy grid as input for the algorithm. Furthermore, methodology for generating fuel efficient policies has been discussed and demonstrated with an example. / Vi presenterar ett autonomt körföretag på ett simulerat motorvägsscenario med fordon som bilar och lastbilar som rör sig med stokastiskt variabla hastighetsprofiler. Fokus för den simulerade miljön är att testa taktiskt beslutsfattande i motorvägsscenarier. När en agent (fordon) upprätthåller ett optimalt hastighetsområde är det fördelaktigt både när det gäller energieffektivitet och grönare miljö. För att upprätthålla ett optimalt hastighetsområde föreslog jag i detta avhandlingsarbete två nya belöningsstrukturer: (a) gaussisk belöningsstruktur och (b) exponentiell uppgång och nedgång belöningsstruktur. Jag utbildade respektive två djupförstärkande inlärningsagenter för att studera deras skillnader och utvärdera deras prestanda baserat på en uppsättning parametrar som är mest relevanta i motorvägsscenarier. Algoritmen som implementeras i detta avhandlingsarbete är dubbel-duell djupt Q- nätverk med prioriterad återuppspelningsbuffert. Experiment utfördes genom att lägga till brus i ingångarna, simulera delvis observerbar Markov-beslutsprocess för att erhålla tillförlitlighetsjämförelse mellan olika belöningsstrukturer. Hastighetsbeläggningsgaller visade sig vara bättre än binärt beläggningsgaller som inmatning för algoritmen. Dessutom har metodik för att generera bränsleeffektiv politik diskuterats och demonstrerats med ett exempel.