PHD Computer Science / Διδακτορικές Διατριβές στην Επιστήμη των Υπολογιστών

Amidst globally changing wildfire activity, increasing attention is focused on the role of the governance system as one component interacting with wildfire regimes and shaping their associated impact. Using an interdisciplinary approach with methods from the natural and the social sciences, the aim of this dissertation was to explore characteristics of governance systems for wildfire regimes theoretically and empirically based on two case studies. In Chapter 1, the significance of the problem was illustrated. A literature review was conducted to define key terms and to outline the rationale of why they are relevant for this work. The concepts of wildfire regimes and governance systems are set out and linked together, to explain how both are intertwined and interact as dynamic entities with physically measurable and socially constructed dimensions. Examples of recent work were introduced to summarise the state of art in the field, point out gaps in the literature, and to highlight this dissertation’s unique contribution to knowledge through the formulation of three research objectives. In Chapter 2, the methodological framework was introduced. To this end, the overall use of discourses methods was set out, together with a reasoning on temporal and spatial choices guiding the further research. An overview of material, methodologies and study design was provided for each research objective, to complement detailed methodology as provided in each respective Chapter. Ethical considerations were elaborated on, and followed by reflections on the limitations of this work such as existing biases due to the use of language, research scope, and deployed theory. The first research objective was investigated in Chapter 3, aiming to advance the field theoretically by systematically retrieving contributions from studies on governance systems for wildfire regimes to date. By engaging sampled work with recent advances in the broader field of natural resources scholarship, this approach allowed to identify components within governance systems that require closer attention. Four themes in the resulting analytical framework pointed to governance processes for actor participation, for collaboration and co-production, to the role of path-dependencies and local place-based characteristics, and of actor adaptation and anticipation. The four themes informed and guided the remaining, empirical research conducted in this dissertation. The second research objective was presented in Chapter 4, aiming to understand interactions across wildfire regimes and the governance system when focusing on regional and national level organizational structures. To this end, an empirical case study was developed and applied to examine the case of Italy. Qualitative data were collected across key stakeholder sectors to identify forces at play and conflicts of interest in the status quo. Survey respondents indicated satisfaction with the governance system mostly concentrated at regional levels that allows to account for variability in the wildfire regime at country-scale. At the same time, that responsibilities for wildfire mitigation and response were found to be fragmented across a range of decentralized institutions and network. While promising to develop locally adapted measures, the fragmented governance system poses coordination challenges across actors, organizational levels and sectors. The third research objective was examined in Chapter 5 with the aim to identify recurring governance opportunities for wildfire across different socio-ecological contexts and burning regimes. A targeted literature review of recently published high-level wildfire reports almost consistently revealed five wildfire challenges. This initial framing was then tracked back to analyze the case of Cyprus through policy analysis and qualitative data collection during a three-week internship at the Department of Forests during the 2023 fire season. Findings revealed well-established processes in the governance systems, but also deficits in wildfire thinking and action to anticipate predicted trends in wildfire regimes. Considering the size of the country, which is smaller than most regions in countries like Spain, Italy or Germany, the case served to study the potential for change in acting on and thinking of wildfire in a range of flammable landscapes and at local to national levels.

Collaboration is an integral part of disaster operations but often fails due to the poor preparation and relationship building. Based on previous research, this research argued that interagency collaboration is not fully understood due to different perceptions of what collaboration entails and requires. This research argues that through collaboration could lead to better wildfire management practices. Building upon Reason's Swiss cheese model, the thesis categorizes barriers hindering collaboration into individual, technology, social, and organizational dimensions, aligning with the Human Technology and Organization (HTO) perspective. Recognizing the interactive nature of these factors, the study distinguishes between latent conditions and active failures, emphasizing the contextual dependence of barrier manifestations on local circumstances. Drawing upon a robust theoretical framework on collaboration (Hutter, 2016; Francesch- Huidobro, 2015; DeHoog, 2015; Hu, Knox, & Kapucu, 2014; Kapucu & Hu, 2014; Noran, 2014), study conducts a holistic case study involving personnel from wildfire management agencies, NGOs, and fire departments in Cyprus. Employing qualitative methods, systematic interviews discover implicit norms shaping exchange and collaboration within the context of disaster management (DeHoog; Hu, et al.; Kapucu & Hu; Noran, 2014). The findings underscore the crucial role of leadership traits, political influence, authority dynamics, effective communication, and relationship building in optimizing the efficiency and effectiveness of wildfire management practices. The research advocates for practical recommendations, emphasizing the need for consistent educational and training requirements for emergency managers. Furthermore, the thesis proposes an integrated approach as a viable strategy to diminish or eliminate collaboration barriers, positing its potential for practical application. The study concludes by encouraging future research to deepen the systematic understanding of additional factors or circumstances impacting collaboration in wildfire management. This work aims to contribute valuable insights for practitioners, policymakers, and scholars committed to advancing the science of collaborative emergency management.

In this thesis I present the publicly available open-source spectral energy distribution (SED) fitting code SMART (Spectral energy distributions Markov chain Analysis with Radiative Transfer models). Implementing a Bayesian Markov chain Monte Carlo (MCMC) method, SMART fits the ultraviolet to millimetre SEDs exclusively with radiative transfer models that currently constitute four types of pre-computed libraries, which describe the starburst, active galactic nucleus (AGN) torus, host galaxy and polar dust components. These libraries are part of the collection of radiative transfer models named CYprus models for Galaxies and their NUclear Spectra (CYGNUS), but also include other publicly available libraries of AGN torus models. An important novelty of SMART is that, although it fits exclusively with radiative transfer models, it takes comparable time to popular energy balance methods to run. Here we describe the key features of SMART and test it by fitting the multi-wavelength SEDs of the 42 local ultraluminous infrared galaxies (ULIRGs) that constitute the HERschel Ultraluminous Infrared Galaxy Survey (HERUS) sample. The Spitzer spectroscopy data of the HERUS ULIRGs are included in the fitting at a spectral resolution, which is matched to that of the radiative transfer models. We also present other results that highlight the performance and versatility of SMART. We compare the main features of SMART with those of other popular SED fitting codes. SMART promises to be a useful tool for studying galaxy evolution in the James Webb Space Telescope (JWST) era.

We present a methodological framework for studying galaxy evolution by utilizing Graph Theory and network analysis tools. We study the evolutionary processes of ultraluminous infrared galaxies (ULIRGs) and quasars and their underlying physical mechanisms, such as star formation and active galactic nucleus (AGN) activity, through the application of graph theoretical analysis tools. We extract, process and analyse mid-infrared spectra of local (z < 0.4) and high-redshift (0.3 < z < 3.0) ULIRGs and quasars between 5 − 38μm through internally developed Python routines, in order to generate relational networks of ULIRGs and quasars (similarity graphs) based on the similarity of their midinfrared spectra. We examine and compare similarity graphs generated using both linear and non-linear supervised classification methods. We also demonstrate how graph clustering algorithms and network analysis tools can be utilized as unsupervised classification techniques under a unified framework for extracting direct and indirect relations between various galaxy properties and evolutionary stages, which provides an alternative methodology to previous works used for classification in galaxy evolution. Furthermore, our methodology compares the output of several graph clustering algorithms in order to identify the best-performing graph theoretical tools for studying galaxy evolution. Additionally, we extract and compare physical features from the mid-IR spectra of ULIRGs and quasars, such as the polycyclic aromatic hydrocarbons (PAHs) emission and silicate depth absorption/emission features, as indicators for the presence of star-forming regions and obscuring dust respectively, in order to determine the underlying physical mechanisms of each evolutionary stage of ULIRGs. We also perform a detailed comparison between the results of the low-redshift (local) and high-redshift samples of ULIRGs and quasars, in order to investigate the evolution of ULIRGs throughout the history of the Universe. Our analysis identifies five types of ULIRGs based on the physical features of their mid-IR spectra, which is quite consistent with the well-established fork classification diagram by providing a higher level classification scheme. The results of our graph theoretical analysis support the evolutionary paradigm of the merger scenario for ULIRGs and showcase noteworthy distinctions in the distribution of ULIRGs and quasars at different evolutionary stages across different redshift ranges.