Timber Groeneveld Preparation
Air temperature estimation through thermal satellite imagery
Supervised by Azarakhsh Rafiee

Phuong Anh Ho Preparation
Urban Heat Island Modelling Using Deep Learning and Satellite Thermal Imagery
Supervised by Azarakhsh Rafiee; Joris Timmermans

Stepan Prikazchikov Preparation
Thesis topic to be announced

Danny Marx Kick-off
Applying Implicit Neural Representations to Landsat Thermal Infrared Imagery for Super-Resolution
Supervised by Azarakhsh Rafiee; Roderik Lindenbergh; Massimo Menenti

Faried Nurmohamed Kick-off
Improving public spaces through urban comfort metrics (thermal and social quality)
Supervised by Daniele Cannatella; Claudiu Forgaci

Kelly Schoonderwoerd Kick-off
Japan’s PLATEAU: from CityGML to CityJSON
Supervised by Hugo Ledoux; Hidemichi Baba

Mark van der Meer Kick-off
CityGit
Supervised by Hugo Ledoux; Hidemichi Baba

Juan Garcia Mezo Midterm
Semantic Interpretation of Change Detection at the Building Element Level
Supervised by Ken Arroyo Ohori; Gina Stavropoulou

This thesis proposal investigates semantic change detection in urban point clouds at the level of building elements, i.e. meaningful geometric components such as roof planes and roof installations. Binary change detection methods (e.g. C2C, M3C2) capture geometric differences between point clouds, but are sensitive to noise, varying point densities, and occlusions, and say little about what actually changed; existing semantic and object-based approaches describe changes only generically (e.g. new, demolished, taller, lower). The research bridges point-level and building-instance change detection by detecting and classifying changes at the element level and propagating this semantics to building instances, while accounting for the challenges of large-scale, multimodal point clouds.

Segher ter Braak Green-light
Downscaling ECOSTRESS Land Surface Temperature using Deep Learning
Supervised by Azarakhsh Rafiee; Roderik Lindenbergh

This thesis proposes a physics-guided deep learning framework to downscale ECOSTRESS land surface temperature (LST) data from ~70 m to 10 m resolution in heterogeneous Dutch landscapes. It integrates Sentinel‑2 imagery, land cover, and topography into a Swin Transformer‑based U‑Net architecture. The approach uses weakly supervised learning with physics‑informed loss functions—such as aggregation consistency and edge‑aware smoothness—to ensure physical plausibility and coherence across scales. The goal is to produce high‑resolution, physically consistent LST estimates suitable for detailed urban and agricultural analyses, addressing the limitations of traditional downscaling methods in complex environments.

Josephine Winterdal Green-light
Assessing Coastal Change and Vulnerability Using Remote Sensing: A Comparative Study of Aruba and Bonaire
Supervised by Azarakhsh Rafiee; Roderik Lindenbergh

This thesis proposal assesses coastal change and vulnerability along the coastlines of Aruba and Bonaire using multi-source satellite data. Coastlines are segmented and classified by coastal type, after which per-segment indicators are extracted — shoreline position from optical imagery (NDWI/MNDWI), vegetation buffers from NDVI, built-up intensity from Sentinel-1 SAR backscatter, and external exposure from coastline orientation, wave fetch and reef presence. Linking these indicators reveals how external exposure and development pressures shape coastal change, and a composite index produces relative vulnerability maps. Comparing PlanetScope (Aruba) and NSO (Bonaire) imagery additionally shows how data quality and island characteristics affect the results.