Master's Thesis

Module 5

Master's Thesis - Research-focused specialization under faculty supervision of Dr. Sounny.

About the Master's Thesis

Module 5 (M5-MT) represents the culmination of the Master's program, where students undertake a significant individual research project or thesis. This module allows for deep specialization in a topic of interest, reaching EQF Level 7 through independent research and original thinking.

⏱️ Duration & Timeline

Full-time: 6 months (July 2026 – February 2027)

Part-time: Up to 12 months (requires approval)

Includes research time, report preparation, and final defense.

📊 Assessment Breakdown

  • 🔹 30% Progress during semester
  • 🔹 30% Quality of the report
  • 🔹 40% Quality of the defense

📦 Key Deliverables

Midterm Presentations

Three oral midterm presentations with your ISU supervisor and external mentor to review progress and methodology.

Thesis Report

A comprehensive research document of 60-100 pages (80 pages recommended) following APA standards and ISU formatting guidelines.

Final Defense

A 75-minute oral examination: 30-minute presentation, 30-minute faculty Q&A, and 15-minute public session.

📚 Master Thesis Project Proposals

The following research topics are available under the supervision of Dr. Sounny for MSc26. Each category contains specific thesis titles with abstracts, required backgrounds, and key references. Click any card to expand.

🌍 Climate Services, Earth Observation & Policy (TERRA)

Focusing on the intersection of satellite data, climate science, and EU policy through the Horizon Europe TERRA project.

1

Mapping the Climate Knowledge Landscape: A Copernicus-Based Framework for EU Policy Briefs

Copernicus C3S Policy Briefs EU Green Deal

Copernicus services generate vast quantities of climate-relevant data, yet translating these outputs into actionable policy documents remains a critical gap. This thesis develops a structured methodology for producing high-impact policy briefs from Copernicus Climate Change Service (C3S) data, validated through engagement with EU Green Deal stakeholders. The output will directly support the Horizon Europe TERRA project.

Required Background: Geography, Environmental Science, Political Science, or related field. Strong academic writing skills required.

Remote Work: Yes

Key References:
Copernicus C3S. (2023). ERA5 hourly data. [Data]
Weaver, C. P. et al. (2013). Improving the contribution of climate model information. WIREs Climate Change. [DOI: 10.1002/wcc.203]
European Commission. (2019). The European Green Deal. [Policy Link]
2

Building Digital Communities for Climate Resilience: A Socio-Technical Analysis of the TERRA Stakeholder Ecosystem

Stakeholder Engagement Communities of Practice

Climate services succeed only when end-users are actively engaged in their co-production. This thesis examines the design and governance of a digital stakeholder community within the Horizon Europe TERRA project, applying community-of-practice theory to assess how knowledge exchange, clustering of related EU initiatives, and workshop facilitation can sustain long-term ecosystem resilience.

Required Background: Social Sciences, Science & Technology Studies, or Policy. Strong communication and research skills.

Remote Work: Yes

Key References:
Wenger, E. (1998). Communities of Practice. [DOI]
Hewitt, C. et al. (2012). The Global Framework for Climate Services. Nature Climate Change. [DOI: 10.1038/nclimate1745]
Ostrom, E. (1990). Governing the Commons. [DOI]
3

Urban Heat Island Dynamics and Flood Risk Mapping Using GEE: A Multi-City Comparative Study

GEE Urban Heat Islands Flood Risk

Rapid urbanization is intensifying climate-related hazards. Using multi-temporal Sentinel-2 and Landsat imagery processed in Google Earth Engine, this thesis quantifies urban heat island intensity and flood risk across a set of European cities, producing interactive dashboards for municipal decision-makers. Theories will be tested on how landuse effects urban heat islands.

Required Background: Geography, Environmental Science, or Engineering. Google Earth Engine or Python skills preferred.

Remote Work: Yes

Key References:
Gorelick, N. et al. (2017). Google Earth Engine. RSE. [DOI]
Voogt, J. A., & Oke, T. R. (2003). Thermal remote sensing of urban climates. RSE. [DOI]
IPCC. (2022). AR6 WGII: Impacts, Adaptation and Vulnerability. [DOI]

🤖 GeoAI & Machine Learning for Earth Observation

Sit at the frontier of deep learning, computer vision, and geospatial science.

4

Rapid Damage Assessment After Natural Disasters Using CNNs and Sentinel-2 Imagery

Deep Learning CNNs Sentinel-2

Post-disaster response is constrained by delays in damage mapping. This thesis trains and evaluates CNN-based classifiers on pre/post-event Sentinel-2 satellite imagery to produce automated building damage maps within hours of a disaster event, benchmarked against established datasets (e.g., xBD).

Required Background: Computer Science, Data Science, or Geography. Python and deep learning experience required.

Remote Work: Yes

Key References:
Gupta, R. et al. (2019). Creating xBD. CVPR Workshops. [arXiv]
LeCun, Y. et al. (2015). Deep learning. Nature. [DOI]
Sublime, J. et al. (2019). Automatic damage mapping. Remote Sensing. [Scholar Link]
5

Foundation Models for Zero-Shot Land Cover Mapping: Evaluating SAM on Satellite Imagery

Foundation Models SAM / GeoSAM

Foundation models pre-trained on massive datasets are enabling zero-shot generalization in remote sensing. This thesis evaluates Meta's Segment Anything Model (SAM) and GeoSAM for land cover segmentation across diverse biomes, analyzing performance trade-offs against supervised baselines.

Required Background: Computer Science, Data Science, or Geography. Deep learning skills required (Python/PyTorch).

Remote Work: Yes

Key References:
Kirillov, A. et al. (2023). Segment anything. ICCV. [DOI]
Mai, G. et al. (2022). Foundation models for GeoAI. ACM SIGSPATIAL. [DOI]
Shi, Q. et al. (2022). Building extraction with transformers. IEEE GRSL. [DOI]
6

Agentic GIS: Designing AI-Driven Autonomous Pipelines for Natural Language Spatial Analysis

LLMs Agentic AI

The integration of LLMs with geospatial tools promises to democratize spatial analysis. This thesis designs a "Agentic GIS" system capable of interpreting natural language queries and autonomously executing multi-step analyses on EO datasets. See sounny.github.io/geoaiagents/

Required Background: Computer Science, Geography, or Data Science. Programming skills (Python). Interest in AI/LLMs strongly preferred.

Remote Work: Yes

Key References:
Brown, T. et al. (2020). Language models are few-shot learners. NeurIPS. [Link]
Wang, L. et al. (2023). Survey on LLM-based autonomous agents. Frontiers of CS. [DOI]
Mai, G. et al. (2023). Opportunities of foundation models for GeoAI. [DOI]

📡 GNSS & Planetary Exploration

Frontier applications from autonomous rover navigation to Mars fluvial geomorphology.

7

Multi-Constellation GNSS Performance for Autonomous Rover Navigation

GNSS Rover Navigation

Reliable GNSS positioning in urban canyons, polar regions, or planetary analog sites is compromised by signal multipath. This thesis analyzes positioning performance of multi-constellation receivers and proposes fusion strategies with IMU and LiDAR for robust rover navigation.

Required Background: Electrical Engineering, Physics, or Aerospace Engineering. Programming skills preferred.

Remote Work: Yes

Key References:
Groves, P. D. (2013). Principles of GNSS, Inertial, and Multisensor Integrated Navigation Systems. [Link]
ESA. (2023). Galileo SIS Interface Control Document. [Doc]
8

Megafan Systems as Planetary Analogs: Reconstructing Paleofluvial Histories on Mars

Mars Hydrology MOLA / HiRISE

Ancient river systems on Mars record the planet's hydrological history. This thesis uses MOLA topographic data and HiRISE imagery to identify and characterize megafan structures on Mars, comparing their morphometric properties to terrestrial analogs.

Required Background: Geology, Physical Geography, or Planetary Sciences. Remote sensing or GIS skills a plus.

Remote Work: Yes

Key References:
Irwin, R. P. et al. (2005). Paleolake basins on Mars. EPSL. [Link]
Lefort, A. et al. (2009). Periglacial landforms in Utopia Planitia. JGR. [Link]
Zimbelman, J. R. et al. (2012). Hesperian Age for Western Medusae Fossae. Science. [Link]

🏗️ Smart Infrastructure, Agriculture & Workforce

Socio-technical applications of space data for sustainability and equity.

9

IoT-BIM Integration for Occupancy-Adaptive Energy Management in Space Habitats

BIM IoT Space Habitats

Future space habitats must minimize energy consumption while maintaining crew well-being. This thesis develops an IoT-BIM sensor framework using machine learning to dynamically optimize HVAC and lighting, benchmarked against NASA standards.

Required Background: Architecture, Engineering, or Computer Science. Interest in smart systems and sustainability.

Remote Work: Yes

Key References:
Eastman, C. et al. (2018). BIM Handbook. [Link]
Balaji, B. et al. (2016). Brick: Unified metadata schema. ACM BuildSys. [DOI]
NASA. (2021). Human Integration Design Handbook (HIDH). [Link]
10

Crop Yield Prediction in Sub-Saharan Africa Using Sentinel-2 NDVI Time-Series

Sentinel-2 Food Security Random Forest

Climate variability threatens agricultural productivity. This thesis applies a Random Forest regression model to Sentinel-2 NDVI time-series to predict seasonal crop yields in smallholder farming regions across Sub-Saharan Africa.

Required Background: Agriculture, Environmental Science, Geography, or Engineering. GIS or remote sensing skills preferred.

Remote Work: Yes

Key References:
Becker-Reshef, I. et al. (2010). Monitoring global croplands. Remote Sensing. [DOI]
FAO & ESA. (2018). Earth Observation for Agriculture. [Link]
Breiman, L. (2001). Random forests. Machine Learning. [DOI]
11

Spatial Feasibility Analysis of Terrestrial Renewable Energy Parks Using GIS

Renewable Energy Suitability Modeling

Identifying optimal sites for solar and wind installations requires balancing energy yield with ecological sensitivity. This thesis builds a multi-criteria GIS suitability model integrating satellite-derived resource data.

Required Background: Geography, Environmental Science, or Engineering. GIS skills required.

Remote Work: Yes

Key References:
Sánchez-Lozano, J. M. et al. (2013). GIS and MCDM for photovoltaic siting. RSER. [DOI]
CAMS. (2023). Solar Radiation Data Service. [Data]
IRENA. (2023). Renewable Power Generation Costs. [Report]
12

Mapping the Global Space Workforce: A Spatial Analysis of Opportunity Gaps

Space Policy STEM Equity

Participation in the space sector remains concentrated. This thesis uses georeferenced workforce data to construct an "opportunity gap" index and proposes policy interventions for equitable access to space careers.

Required Background: Social Sciences, Policy, or Data Science. Quantitative and qualitative research skills required.

Remote Work: Yes

Key References:
UNESCO. (2021). Women in Science: Fact Sheet No. 60.
OECD. (2022). Science, Technology and Innovation Outlook.
Space Foundation. (2023). The Space Report.