Dr. Georg Pointner

PhD-Thesis: Geospatial Analysis of Features in Arctic Lake Ice Synthetic Aperture Radar Imagery (Oktober 2021)

The dissertation focuses on applications of Synthetic Aperture Radar data in combination with image processing techniques and geospatial analyses to monitor and study Arctic lake ice. The role of lake size and local phenomena for the classification of ground-fast and floating lake ice, and characteristic patterns of low backscatter (anomalies) in floating ice regimes of Arctic lakes and their potential relation to methane gas emissions are investigated. The results show that effects of improved methods for the classification of ground-fast and floating lake ice are most pronounced for the largest lakes and that anomalies are likely related to flooding of the lake ice surface through holes in the ice, which are in turn likely caused by methane seepage. The results further suggest that characteristic anomalies occur predominantly in Northwestern Siberia. The research contributes to a better understanding of the associated phenomena and is important with regard to climatological and ecological concerns.





Dr. Zahra Dabiri

PhD-Thesis: Investigation and Evaluation of Radar and Optical Earth Observation Data in an Object-based Image Analysis Framework (September 2021)

This dissertation focused on evaluating the importance of multiscale Earth Observation (EO) data for mapping and monitoring geographical features and assessing their changes in an object-based image analysis (OBIA) framework. The challenges associated with the integration of radar and optical data were studied for two main applications: (a) biodiversity, including forest mapping and tree species characterisation; and (b) geomorphological features, including landslides and their impacts on the surrounding environment and studying spatio-temporal dynamics of the ice-marginal lakes. Moreover, I assessed the applicability of the European's Union Earth observation Copernicus programme Sentinel-1 radar data for high-quality digital elevation models (DEMs) generation, and I evaluated the reliability of Sentinel-1 derived DEMs for landslide volume estimation. The results contribute to effective feature extraction from multiscale EO data and for the identification of environmental problems within the GIScience and GI*EO domain.



Dr. Omid Ghorbanzadeh

PhD-Thesis: Evaluation of knowledge-based and machine learning models for Earth Observation and Geoinformatics applications (June 2021)

The thesis contributes to critically evaluating the possibility and efficiency of using Artifical Intelligence (AI) methods for some Earth Observation (EO) and spatial modeling applications. It also shows the potential of combining human knowledge with AI methods to achieve more reliable results in both applications and various case studies of human-made and natural disasters. In this thesis, the EO application focuses on the accurate and timely assessment of disaster-affecting regions and approximation of the affecting population. The spatial modeling application aims to categorize and spatially delineate disaster-prone areas, which can be achieved by analyzing the event, characteristics, and influences of previous disasters, relating them to the current situation, and producing reliable predictions for the prospect support decision-making. Compared to previous approaches used for the same EO and spatial modeling applications, the innovative combining approach of human knowledge and AI methods significantly improved the results. 

Dr. Alessandro Crivellari

PhD-Thesis: Artificial Neural Networks for Human Mobility Analysis and Spatial-Temporal Activity Modeling (June2021)

The thesis focuses on advancing the research on human mobility by introducing artificial neural network (ANN) models in the context of trajectory analysis and spatial-temporal phenomena. The domain of natural language processing (NLP) inspired the designed extraction of movement patterns, motivating a conceptual analogy between mobility traces and language structure. As ANN-based NLP teaches machines how to read and write, our purpose aimed for intelligent systems that automatically analyze motion data and deliver the demanded information by learning hidden patterns in a purely data-driven manner, without requiring any manual feature extraction or human knowledge assistance. The proposed methodological direction opens to a variety of analytical tools that can potentially be successful in a wide range of applications related to trajectory modeling and spatial-temporal sequential phenomena, indicating a clear exploratory path for future research activities.


Dr. Anna Kovacs-Györi

PhD-Thesis: Defining and Assessing Urban Livability - A GIScience Perspective (February 2020)

This doctoral research investigated how urban livability assessment combined with geospatial analysis methodologies offer a way to reduce the complexity of the urban systems and livability itself by handling the characteristics and phenomena of cities systematically. The first objective was to provide a conceptual framework for defining and assessing livability. As a follow-up, various approaches were developed to grasp the complexity of the person-environment relationship in order to handle the mentioned challenges of livability assessment. All the examples are constructed in a way to represent either a specific segment or the overall assessment of livability by focusing on the person-environment relationship, and the perception of people, instead of merely statistical factors such as income levels or number of facilities.


Dr. Martin Sudmanns

PhD-Thesis: Big Earth data disruption in Earth observation: A geospatial perspective of a transition (February 2020)

The dissertation investigated how the geospatial perspective can provide a unique asset and value to big Earth data technology, science, and applications. Big Earth data are the challenges, opportunities, and research questions associated with the management of large amount of Earth observation (EO) data, the development of new methods, and their application in a variety of (new) domains. Objectives of this work were to investigate and understand the spatio-temporal distribution of big Earth data, to develop an approach for managing spatio-temporal objects in EO data cubes, and to develop methods for semantic analyses in EO data. The results contribute to more effective information production from big EO data and make them better tangible in operational contexts and in variety of application domains.


Dr. Emmanuel Papadakis

PhD-Thesis: Bridging Place and Space in GIScience - Formalization and Integration of Place in Geographic Information Systems (October 2019)

This Thesis aimed to investigate whether the representation standards of a Geographic  Information System (GIS) can provide an adequate reflection of places, which depict the informal, subjective and vague geographical world as it exists in the human mind. This goal was achieved through three research objectives: a) conceptualization, b) formalization and c) integration of place in GIS. Initially, place was modelled as functional space bounded to human purposes, granting it intersubjective interpretation and spatial description. The second phase formalized place as a system of interconnected spatial objects that enable a particular functionality; this idea allows the digitization of the elusive concept of place and maintains a connection between spatial information and human purposes. Finally, the proposed framework Topos makes it possible to digitally represent, localize, delineated and in general integrate
instances of place within GIS.



Dr. Helena Bergstedt

PhD-Thesis: Spatial and temporal patterns of microwave scattering mechanisms across scales in cold terrestrial environments (September 2019)

This Thesis explored the spatial and temporal patterns of different mechanisms impeding the retrieval of freeze/thaw information from satellite based radar sensors and focused on permafrost regions. The results show that there are differences between freeze/thaw information based on datasets that have different spatial and temporal resolutions. More complex landscapes like alpine areas, lake rich regions and areas with dense forest were more sensitive to differences in temporal and spatial resolution. To better account for the spatial complexity of different cold landscapes, we propose to combine different satellite data sets (in this case from Sentinel-1 and Metop ASCAT) to be able to monitor the gradual freezing and thawing of the ground surface. Previous datasets for this purpose offered only yes/no freeze/thaw information. Describing the gradual nature of the freezing and thawing of landscapes in cold regions will allow researchers to improve the accuracy when monitoring freezing and thawing cycles.


Dr. Alina Ristea

PhD-Thesis: Integration and Evaluation of Social Media in Crime Prediction Models (August 2019)

This dissertation focused on geospatial predictive crime analysis through the exploration of the complex parameters of social media data and additional information, including demographic status and safety risk factors. The overall goal of this PhD research was to determine significant social media information and how to integrate it into spatial and spatiotemporal crime prediction models. The narrower focus was on crowd-based events, thus facilitating the exploration of three core elements: crime, social media and sporting events.


Dr. Francis Oloo

PhD-Thesis: Sensor-driven, spatially explicit agent-based models (February 2019)

This work contributes to the research on empirical spatial simulation by pursuing two objectives. The first objective aimed to develop methods for incorporating dynamic sensor-derived data into spatially explicit agent-based models. Secondly, this research aimed to explore the influence of data-driven approaches on the performance of spatially explicit agent-based models.


Dr. Gyula Kothencz

PhD-Thesis: Data and information development methodologies for green urban area planning and management (November 2018)

The dissertation proposed data and information development methodologies to support research and decision making concerned with green urban areas. The thesis achieved this goal by interpreting perceived green space characteristics, predicting their role on green space visitors’ well-being, and by developing a cost-effective urban vegetation height extraction work-flow utilising very high resolution (tri-)stereo images of the Pléiades satellite system. The developed methods and results are already applied in green space redevelopment in Hungary, and green space evaluation in Austria.


Dr. Philip Glasner

PhD-Thesis: Spatial and temporal analysis of crime and the impact of rhythmic cycles to predict future crime locations: Studies from the city of Vienna, Austria  (September 2018)

In this thesis, crime data from Vienna, Austria, are used where predictive crime mapping has not been researched thus far. The results of this thesis suggest that there is considerable potential for law enforcement in the city of Vienna to employ concepts of predictive policing to forecast future crime events. Based on the findings, law enforcement agencies, not only in Austria, can optimize tactical and strategic planning of police resources in combating crime.


Dr. Bartosz Hawelka

PhD-Thesis: Collective Sensing and Human Mobility Analysis (September 2018)

The thesis focused on understanding the collective spatio-temporal behaviour of complex systems by analysing the vast amount of anonymous  digital traces that are continuously left behind while using digital networks. Data produced by those networks act as probes of collective behaviour. In the thesis they are called Collective Sensing Probes (CSPs). Three CSPs were analyzed that relate to human mobility, specifically to the time-space dimension of mobility, in particular telecom data, social media data and bank card transactions. The study explores the content and representativeness of the selected probes, as well as spatial (local, global) and temporal patterns (short-, long-term) that can be derived from the CSPs, also in combination with traditional data. The thesis also proposes a new algorithm to predict individual mobility traces based on the travel history of other users of the system.



Dr. Ovidiu Csillik

PhD-Thesis: Spatil-temporal object-based image analysis for land-use/land-cover mapping (July 2018)

The PhD thesis aimed to develop efficient, fast and accurate object-based image analysis (OBIA) workflows for single and time series satellite images for land-use/land-cover mapping, with a special focus on mapping agriculture. Superpixels proved efficient in terms of speed and accuracy in analyzing single single date very high-resolution remote sensing data. Adding time to the analysis, the thesis proposed an object-based dynamic time warping (DTW) approach to classifying crops based on their phenological pattern  derived from vegetation indices extracted from Sentinel-2 time series data. The first implementation of DTW into an OBIA framework was proposed. The research undertaken in this PhD thesis has a significant potential to enable automation and objectivity of OBIA workflows for time series data analysis.


  Dr. Willi Mann

PhD-Thesis: Efficient Techniques for Set Similarity Join Queries (March, 2018)

In my thesis I focused on techniques for set similarity join queries. The join is a fundamental operation in databases. A growing number of problems require joins with fuzzy predicates. A class of frequently applied fuzzy predicates is the similarity of two sets. Typical set similarity functions include Jaccard, Cosine, dice, and Overlap, and the Hamming distance. The set similarity join can be executed in offline settings where the input is known up front and online settings, like on streams of timestamped sets.



  Dr. Cornelia Schneider

PhD-Thesis: Methods for recording, processing and evaluating human movement data in the context of Active and Assisted Living (January, 2018)

In my PhD thesis, I aimed to contribute to the next generation of mobility assistance systems. For assisting frail or cognitively impaired people with mobility assistance systems, monitoring and analysing their outdoor movements is important. Questions such as where people typically stay (and for how long), where they have been or where they most likely are going to can be answered by means of movement analysis. Consequently, the aim of this dissertation was to address these questions by recording, processing and evaluating human movement data.



Dr. Pablo Cabrera-Barona

PhD-Thesis: A Multidisciplinary Spatial Framework for Health Inequalities Analysis with Emphasis on Deprivation and Healthcare Accessibility (July, 2017)

With my PhD research I provided a multidisciplinary spatial framework that uses different mixed-methods approaches to understand health inequalities. I developed different multidimensional indices: deprivation indices, healthcare accessibility indices, and an index of healthcare satisfaction.  These indices were related to different social variables and health outcomes. Additionally, I incorporated assessments of scale and geographical context implications of the developed indices. The results of my Thesis showed that the indices created facilitate the identification of social disadvantages and health-related inequalities, and that these indices can support the explanation of several health outcomes. Important evidence of neighborhood effects on health was also found.
My Thesis moved beyond the traditional GIS-based health geographies to a more integrated, holistic and pluralistic approach to understand inequalities and health.



Dr. Ivan Tomljenovic

PhD-Thesis: Transferable and Robust Building Extraction from Airborne Laser Scanning Data using Object-Based Image Analysis Paradigm (October, 2016)

In my PhD thesis I investigated use of Object-Based Image Analysis for the Airborne Laser Scanning Point Cloud data. The main goal of my research was to extract building outlines in and automated and robust way. Initially, a overview of an existing building extraction approaches has been made,  followed by the generation of a  prototype single-click solution in order to obtain initial spatial data in an automated manner. Obtained results were tested for their appropriateness in two separate cases: building type identification using random forest statistical approach and ISPRS Benchmarking. In both cases the results provided satisfying response (above 90%). The approach was additionally improved between the two use case scenarios. Final work provided additional analysis of the stability of the developed solution. The core result provided a rule-based solution for an automated building outline extraction from the ALS data based on the application of OBIA methodology.



  Dr. Christian Neuwirth

PhD-Thesis: Evolution of space in System Dynamics simulations (May, 2016)

In my PhD thesis I studied the influence of Spatial Structural Feedbacks (SSF) on the behavior of complex systems and developed a tool for simulating SSF. Effects of SSF can be observed in a multitude of system types, ranging from natural to anthropogenic and from microscopic to global scales. They all have one thing in common: Processes (e.g. the flow of water) reshape spatial structure (e.g. the earth’s topography), which in turn affects the process (e.g. the flow of water). Among other things, results indicate a balancing effect of SSF. Neglecting those effects in computer-based simulations of real-world phenomena causes an overestimation of a system’s self-enforcing capacities, runaway behavior and exponential growth or decline.



Dr. Peter Ranacher

PhD-Thesis: GPS movement analysis: measurements, similarities and patterns (March, 2016)

In my PhD thesis I aimed to quantify the influence of error when recoding movement with a GPS. GPS movement data are affected by two types of error, measurement error and interpolation error. Both errors can significantly influence knowledge extraction from movement data. However, appropriate methods to assess the influence of error on real-world movement data are still lacking. This thesis introduces such measures for both measurement and interpolation error. The second aim was to develop a model to study the energy-efficiency of cars in an urban road network. This model analyses a car's movement patterns recordedwith the GPS and estimates how energy-efficiently the car is moving.



Dr. Ourania Kounadi

PhD-Thesis: Geospatial Privacy Framework for Confidential Discrete Data with Emphasis on Spatial Crime Analysis & Visualisation (August, 2015)

In my PhD thesis I aimed to understand the risks, trends, and opinions on location privacy. Existing protection methods for confidential discrete location data as well as published warnings about disclosure risk were reviewed. Furthermore, the public’s perception on location privacy in the particular context of crime mapping was collected through interviews and further analyzed. Also a basis was created to achieve the second aim of this thesis, which is to offer cartographic display guidelines when crime data are presented on public maps or disseminated among stakeholders. The results are scientifically innovative and highly relevant for the work of Law enforcement agencies like the Bundeskriminalamt.



Dr. Günther Prasicek

PhD Thesis: Morphometric analysis of alpine topography conditioned by tectonic uplift and glaciation (August, 2015)

In my PhD thesis I developed geomorphometric tools to extract information about the distribution and persistence of glacial topography from digital terrain models. The distinct geometry of U-shaped glacial and V-shaped fluvial valleys facilitates the automated identification of glacial valley cross sections and the quantification of glacial imprint. A multi-scale curvature approach is used to fit the analysis scale to valley width and to subsequently extract valley cross-sections. Power-laws are fitted to the extracted transects to quantify valley shape. This procedure allows for the automated, efficient and objective analysis of extended mountain areas.

The results of a worldwide analysis of prominent mountain ranges indicate that the lifespan of glacial topography in Earth’s most rapidly uplifting mountain ranges is on the order of one interglacial period, preventing the development of a cumulative glacial signal over multiple glacial cycles. In contrast, in most alpine landscapes more than 100 kyr are required for the transformation from glacial back to fluvial topography and glacial landforms have not or have only partially been erased during the current interglacial.



Dr. Thomas J. Lampoltshammer

PhD-Thesis: Natural Language-based Modelling, Processing, and Interaction in Geographic Information Systems (July, 2015)

The focus of my thesis is the evaluation of the applicability of natural language-based approaches and methods within the following three major functional dimensions of geographic information systems, namely modelling, processing, and the interaction of geo-referenced information. The presented research work contributes to the research area of semantic knowledge engineering using ontologies within geoinformatics applications. It is demonstrated how ontologies can be used to facilitate the modelling of expert knowledge for an efficient model-based classification of real-world entities. In addition, how ontologies can be employed as a means of natural language-based proxies to interact with expert systems is demonstrated.

My research further analyses implicit and explicit geo-referenced information in the form of social media data and demonstrates how these data sets can be used to discover tangible and intangible infrastructure within social structures. Furthermore, it is elaborated how natural language-based processing methods and tools can foster the functionalities of geographic information systems such as geocoding or reverse geocoding.





Dr. Eva Haslauer

PhD-Thesis: GIS-based backcasting: An innovative method for parameterisation of sustainable spatial planning and resource management (July, 2015)

In my Doctoral Thesis I developed a GIS-based, spatially explicit backcasting model. Backcasting was conceptually developed in the 1970s as a planning method for electricity supply and demand and to support sustainable decisions in the energy sector. At this time it was called “backwards looking analysis”. Backcasting, as a decision support model, starts from an desired future goal and works backwards until the present state is reached. The future goal is often a vision describing a desired future scenario. During the backwards development milestones are set. They are small interim scenarios along the inverse way between the present state and the future scenario which is usually

20 to 50 years ahead. The milestones shall reveal how to achieve the desired future scenario step by step. The backcasting model is implemented with Python and applied to a case study of urban sprawl in Salzburg, Austria. The results of the model show a back-casting of land use classes from a future state back to the present, in 10 year time steps. This can be utilized to counteract urban sprawl and its negative consequences by supporting spatial planners to derive long-term strategies for sustainable spatial developments in the case study area.



Dr. Sebastian d'Oleire-Oltmanns

PhD-Thesis: Gully mapping on multiple scales based on UAV and satellite data (December, 2014)

In my dissertation I present methods for gully mapping on two different scales. The small scale level is based on Small Format Aerial Photographs (SFAP) acquired with an Unmanned Aerial System (UAS). The large scale level is based on high resolution optical satellite data (QuickBird-2).

In addition to the traditional field investigations carried out in geomorphology, remotely sensed data acquired at different scales covers various possibilities to analyze a specific study site, the broader spatial context around this study site (e.g. catchment level) and finally enables area-wide mapping of gully-affected features. Several workflows were developed and applied: a workflow for photogrammetric processing of SFAP data, the concept of object-based image analysis (OBIA) was applied at two different (spatial) scale levels for the detection of gullies and gully-affected areas and further analysis of the DSM provided estimations on the total soil loss for a specific gully as well as for the catchment level.



Dr. Michael Hagenlocher

PhD-Thesis: Integrated spatial indicators for modeling, exploring and visualizing vulnerability to vector-borne diseases (December, 2014)

My dissertation presents novel concepts, methods and tools for the spatial assessment of social vulnerability to vector-borne diseases (VBDs) at different spatial scales. The focus was placed on modeling spatial variations in vulnerability of the population to two mosquito-borne diseases, malaria and dengue. Study areas include Santiago de Cali, Colombia (local scale) for dengue, as well as Tanzania (national scale) and the five member states of the East African Community (regional scale) for malaria. To evaluate the robustness of the results in regards to changes in the input parameters, a GIS tool for sensitivity analysis is presented which was used to assess the impact of indicator selection and indicator weights. A web-based tool is proposed which enables users to visualize and explore the results in an interactive manner. By indicating vulnerability ‘hotspots’ and simultaneously highlighting the contribution of the single vulnerability indicators for the respective region, the tool enables decision makers to prioritize intervention areas and to target appropriate interventions. It is therefore anticipated that the concepts, methods and tools proposed in this dissertation can make a valuable contribution to reducing prevailing vulnerabilities and strengthening resilience of the population to VBDs in the study areas, and act as a role model for future assessments.



Dr. Mariana Belgiu

PhD Thesis: Formal ontologies for extracting information from remotely sensed data (June, 2014)

My PhD thesis is a contribution to bridging the semantic gap between low-level information extracted from remotely sensed data and high-level user concepts. It proposes a classification protocol which combines data-driven segmentation approach with formal ontologies of target classes. Ontologies have been successfully applied to classify urban building types, land cover classes defined by Food and Agriculture Organization of the United Nations (FAO/UN) and land cover classes defined by the Austrian Environmental Agency within the Land Information System of Austria (LISA) from remotely sensed data. Land cover and building type ontologies are extended with descriptive information detectable in remotely sensed data by using image interpretation keys, supervised classifiers such as Random Forest, and by using common sense knowledge gained so far for mapping urban buildings and land cover classes in urban/suburban environments. The developed ontologies can be easily integrated into other knowledge infrastructures dedicated to information sharing and integration in a distributed way.



Dr. Bakhtiar Feizizadeh

PhD-Thesis: Uncertainty, Sensitivity and Fuzzy Sets in GIS Multicriteria Decision Analysis (May, 2014)

My main research interests include methodological issues in geoinformatics, integration of GIS and Remote Sensing for land use/cover monitoring, land suitability analysis, geohazard and risk assessment, in particular landslide mapping. I have practiced developing methodology for GIS based Multicriteria Decision Anlysis (GIS-MCDA), modelling sensitivity and uncertainty in GIS environment, Object Based Image Analysis (OBIA) and thermal Remote Sensing image processing. I have teaching experiences in both Remote Sensing and GIS.



Dr. Clemens Eisank

PhD-Thesis: An Object-Based Workflow for Integrating Spatial Scale and Semantics to Derive Landforms from Digital Elevation Models (June, 2013)

In my PhD thesis I proposed a two-phased semi-automated workflow that allows for more objective mapping of landforms in Object-Based Image Analysis (OBIA) based on Digital Elevation Models (DEMs). The workflow integrates methods for detecting ‘characteristic scales’ of segmentation-derived terrain object patterns (Phase 1), as well as explicit representations of the common sense geomorphological landform knowledge (Phase 2). These knowledge models support the selection of representative operational features for semantics-based classification of landforms at detected scales. The workflow was successfully applied to derive two distinct types of glacial landforms from DEMs: cirques and drumlins. A third test on extracting gullies in aerial photographs showed that the workflow can potentially be transferred to non-glacial landforms, and to other data types with various spatial resolutions. The proposed workflow ensures that landform mapping is conducted at representative terrain object scales and that objective geomorphological landform knowledge is integrated. In comparison to previous approaches landforms are mapped more objectively, and thus it is expected that the workflow may become a standard in geomorphometry.



Dr. Günther Sagl

PhD-Thesis: Spatio‐Temporal Analyses of Environmental and Social In Situ Sensor Data (December, 2012)

In my dissertation I investigated dynamics and relationships of some environmental and social phenomena on the basis of such sensor data using interdisciplinary analysis methods linked with GIScience. Specifically, I enhanced a service‐oriented geo‐analysis workflow that facilitates near real‐time environmental monitoring and time-critical decision support; I analysed collective human behaviour patterns derived from user‐generated data in mobile networks or social media to better understand the “pulse” of dynamic urban systems; I developed the novel “context‐aware analysis approach”, which comprises several (statistical) analysis methods, to explore spatial, temporal and periodical relationships between the weather and collective human activity – thereby providing insights into environment‐human interface aspects. The outcomes of my PhD research can therefore provide a basis for further interdisciplinary investigation towards the development of an adaptive framework for real‐time monitoring and modelling of environment-human feedback loops.

Currently I am working as a PostDoc Researcher in the GIScience Research at the University of Heidelberg, where I am involved in several fascinating and challenging projects somehow related to my PhD topic.