Scholarly Works
Research Output
Recent Publications
SeaScope: A Transparent and Reproducible LLM-Assisted Framework for Maritime Earth Observation Analysis
Publishing year: 2026
Journal article
Remote Sensing, 2026, Volume 18, Issue 17
Repository
Earth Observation (EO) analysis increasingly relies on large and heterogeneous satellite datasets, yet developing EO workflows often requires specialized expertise in data selection, geospatial programming, and cloud-based processing. Recent advances in Large Language Models (LLMs) offer new opportunities for natural-language interaction with EO systems, although challenges related to transparency, reproducibility, and domain-specific reasoning remain. This study presents SeaScope, an explainable AI framework that integrates LLMs, Retrieval-Augmented Generation (RAG), scientific knowledge retrieval, and Google Earth Engine (GEE) to transform natural-language requests into transparent and executable EO workflows. The framework combines knowledge retrieval, code generation, cloud execution, provenance tracking, and interactive visualization within a unified environment. A pilot implementation is demonstrated through maritime and coastal monitoring applications, including oil spill detection, vessel monitoring, water quality assessment, floating debris detection, and air quality analysis. Multiple state-of-the-art LLMs are evaluated under both RAG and non-RAG configurations using representative EO case studies. The results indicate substantial differences among model families and show that retrieval augmentation can significantly improve workflow generation quality and reliability for capable models, while providing more limited benefits for smaller models. The proposed framewor...
Robust scheduling policy learning for ore-flow blending in dynamic and uncertain environments
Publishing year: 2026
Journal article
Computers and Operations Research, 2026, Volume 194
Repository
Ore-flow blending is essentially a real-time scheduling decision problem where equipment failures, sudden weather changes, and time-varying road network exist during production. This problem faces core challenges including high-dimensional spatiotemporal dynamics, real-time decision response, and long-horizon decision dependencies. To overcome these challenges, this paper designs an end-to-end ore-flow blending and truck scheduling framework based on a spatio-temporal graph policy network. By encoding the spatial topology of the road network and the dynamic temporal sequences of the ore flow into a unified state representation, the proposed model achieves scalable learning across road networks of varying scales and structures. Furthermore, a two-stage training strategy combining local short-horizon pre-training and global long-horizon fine-tuning is proposed, effectively resolving the computational and memory consumption challenges in large-scale, long-horizon scenarios. Experimental results on multiple real-world mine cases and dynamic scenarios demonstrate that this method significantly outperforms state-of-the-art techniques in ore flow stability control and exhibits strong robustness and generalization capabilities under different road network structures, equipment scales, and dynamic events.
Optimized spectral indices for global vegetation and water mapping using Sentinel-2
Publishing year: 2026
Journal article
Scientific Reports, 2026
Repository
Pages 1-13
Reliable mapping of vegetation and surface water from satellite imagery remains challenging, as common spectral indices can saturate at high biomass, show limited sensitivity across ecosystems, and confuse targets with soil, shadows, or built-up surfaces. We present two indices, the Symbolic Regression Vegetation Index (SRVI) and the Symbolic Regression Water Index (SRWI), discovered with a data-driven symbolic regression framework applied to Sentinel-2 Level-2A reflectance and guided by ESA WorldCover labels. Expressions were evolved from physically interpretable building blocks using non-linear combinations of visible, NIR, and SWIR bands. Indices were derived on a spectrally complex Mediterranean site and evaluated on eleven independent regions spanning diverse biomes. Performance was assessed with the Jeffries–Matusita distance, averaged across months to account for phenology, and compared against established vegetation indices (NDVI, EVI, SAVI, MSAVI2, NDRE) and water indices (NDWI, MNDWI, AWEI, TCW, WI2015). SRVI improves separability between vegetation and non-vegetation and shows higher discrimination among vegetation types relative to all benchmarks. SRWI yields more consistent water delineation with reduced confusion with built-up and shadowed surfaces, outperforming standard alternatives on the same datasets. Results indicate that symbolic regression can produce compact, interpretable indices that generalise across regions and seasons, offering practical gains for...
D1.18 Impact Assessment Report for RP5
Publishing year: 2026
OtherProduct
Repository
This Deliverable presents the impact assessment of the EXCELSIOR project for Reporting Period 5 (RP5), covering the period from 1 April 2025 to 30 September 2026 (Months 67-84). The assessment is based on the impact monitoring framework established through Deliverables D1.12 and D1.13, and evaluates the progress achieved against the approved Key Performance Indicators (KPIs) across the domains of participating organisations, scientific community and society.
RP5 represents a particularly productive period in the evolution of the ERATOSTHENES Centre of Excellence (ECoE). During RP5, the Centre continued to strengthen its research infrastructure, scientific capacity, international visibility, stakeholder ecosystem, and long-term sustainability.
Major achievements included the full operationalisation of the Satellite Data Acquisition Station (DAS), the establishment of the EarthView Cyprus joint venture with CYTA, the progression of the Ground-Based Remote Sensing Station (GBS) from infrastructure operation towards full scientific exploitation and EarthCARE calibration/validation, the launch of Space BIC by ERATOSTHENES CoE, and the strengthening of ECoE's role within the European and international space ecosystem following Cyprus' accession as an ESA Associate Member. RP5 was also marked by the signing of the contract for the construction of ECoE’s new headquarters, which is expected to be concluded by April 2027. Among several other highlights, the approval of a national Perfo...
Cyprus Earth Observation Data Cube towards water resource management in the context of climate crisis
Publishing year: 2026
Journal article
Big Earth Data, 2026
Repository
The amount of Earth observation data is getting larger day by day. This rapid evolution of the field and the exponential growth of data generation mediums urges the need for modern ways of fast analysis and processing of satellite images. Earth Observation Data Cubes enhance the way Earth data are distributed, handled, stored, and analyzed by providers and users. In this study the development of the National Earth Observation Data Cube of Cyprus is described and the importance of such software-based infrastructure is demonstrated through a water resource monitoring use case in the context of climate crisis in the Mediterranean region. Moreover, this study exhibits the capabilities of semantic classification in Earth observation, introducing a spatially generalized approach for easier environmental monitoring. Such developments can enhance the preservation of the environment and its protection from future natural and human disasters.
Ambulance Routing Optimization Problem with Patient Triage Priority
Publishing year: 2026
Journal article
International Journal of Intelligent Transportation Systems Research, 2026
Repository
Emergency medical services (EMS) face critical challenges in efficiently allocating and routing ambulances to minimize response times, particularly in scenarios where high-risk patients must be prioritized under limited resources. This study investigates an ambulance routing problem (ARP) with patient triage priority, where the objective is to jointly assign ambulances to patients and determine their routes so as to minimize total travel time, defined as the sum of the ambulance-to-patient and patient-to-hospital travel times. A mixed-integer linear programming (MILP) model is proposed that integrates assignment and routing decisions, while capturing realistic EMS operational constraints such as patient prioritization and heterogeneous travel times. The proposed MILP is formulated as a multi-criteria optimization problem that prioritizes the transportation service of red-code patients while minimizing total travel time. To address the computational complexity of solving large-scale ARP instances, a metaheuristic based on Tabu Search (TS) is developed, incorporating a set of problem-specific local search operators. To evaluate the effectiveness of the MILP model and TS algorithm, a new set of benchmark instances is generated for the ARP with triage-based patient priority. Computational experiments demonstrate that the MILP model solves small-scale instances (up to 40 patients) optimally in less than one minute, but requires up to 32 minutes to solve large-scale instances (up t...
104
Unique publications
54.8%
Open access (57 publications; 16 without information)
51.9%
With other institutions (54 publications; 22 without information)
H-Index Reports
| Index type | Scopus | Google Scholar | Web of Science |
|---|---|---|---|
| H-index | 23 | 31 | 0 |
| H5-index | 6 | 9 | 0 |
| H10-index | 15 | 18 | 0 |
Publications per year and source
No description yet.
- Title
- Senior Researcher