Maria-Sotiria Frousiou

Maria-Sotiria Frousiou

Auto-generated profile, based on publicly available information

Maria-Sotiria Frousiou


Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 March 2026
Verified
Conference Paper
Open access
DPDimitrios Papadomarkakis
Dimitrios Papadomarkakis
MFMaria-Sotiria Frousiou
Maria-Sotiria Frousiou
ANAlexandros Notas
Alexandros Notas
Flooding remains among the most damaging hydro-meteorological hazards in the Mediterranean. On Euboea Island (Greece), steep terrain, ongoing land-use change, and highly connected transport corridors can intensify both flood occurrence and potential consequences. This study presents an integrated, decision-oriented framework that jointly maps (a) flood susceptibility and (b) flood impact potential (exposure), combining Google Earth Engine (GEE) for predictor generation with Python-based machine learning, explainability, and uncertainty analytics.A multi-source predictor database is assembled in GEE from satellite and ancillary datasets to represent key topographic, climatic, geological, and pedological controls on flooding. Terrain and morphometric predictors are derived from the ALOS 12.5 m DEM, including elevation, slope angle, plan and profile curvature, Topographic Wetness Index (TWI), and Topographic Position Index (TPI). Hydrologic connectivity is captured through distance to the river/stream network. Climatic forcing is represented using the Modified Fournier Index (MFI) from WorldClim v2.0 as a predictor variable for rainfall influence. Subsurface controls are incorporated via lithology (geological map) and topsoil texture (LUCAS database; sand, silt, and clay content), which modulate infiltration, storage, and runoff generation. Land-surface conditions affecting runoff are characterized using CORINE Land Cover 2018, reflecting vegetation cover and imperviousness patterns. In parallel, exposure is quantified using land-use intensity, building footprint/coverage metrics, and road-network descriptors (density, proximity, connectivity) to identify areas where flood impacts are likely to be most severe.Flood occurrence labels are derived from an event inventory, and spatially explicit sampling and partitioning are applied to reduce spatial autocorrelation and improve generalization. Susceptibility is modeled using tree-based ensembles (Random Forest and XGBoost), trained and evaluated in Python with spatial cross-validation and metrics capturing both discrimination and reliability (AUC, F1/TSS, Brier score, and calibration diagnostics). To explicitly communicate confidence and reveal spatial weaknesses, we generate uncertainty and entropy maps: (a) predictive uncertainty estimated from ensemble dispersion and calibrated probabilities, and (b) Shannon entropy of class probabilities to highlight ambiguous transition zones, data-sparse areas, and geomorphologically heterogeneous corridors. Explainability is delivered via SHAP (global and local), supported by interaction and partial dependence analyses to identify dominant controls and to attribute exposure hotspots to drivers such as building and road concentration.The resulting susceptibility, exposure/impact, and uncertainty–entropy maps provide transparent, decision-relevant information to support mitigation prioritization and strengthen trustworthy flood-risk screening on Euboea Island.Keywords: flood susceptibility; exposure; impact mapping; Google Earth Engine; Python; tree-based ensembles; uncertainty; predictive entropy; SHAP; explainable AI; Euboea; Greece

13 March 2026
Verified
Conference Paper
Open access
ANAlexandros Notas
Alexandros Notas
MFMaria-Sotiria Frousiou
Maria-Sotiria Frousiou
DPDimitrios Papadomarkakis
Dimitrios Papadomarkakis
Wildfires in Mediterranean ecosystems are increasing in frequency, extent, and severity under the combined influence of climate change and human pressure. This trend is intensifying the need for operational hazard products that are not only accurate, but also transparent, auditable, and easy to justify to decision-makers. Here we present an open-access, fully reproducible remote-sensing workflow for (i) pre-fire danger mapping and (ii) post-fire burn severity assessment, explicitly designed around explainability rather than black-box prediction. The workflow is implemented in Google Earth Engine using only freely available data sources: Sentinel-2 and Landsat 8 optical imagery, ERA5-Land meteorological reanalysis, and OpenStreetMap ancillary layers.Post-fire impacts are standardized through NBR (Normalized Burn Ratio) and dNBR (difference Normalized Burn Ratio), converted into burn-severity classes using established USGS-style thresholds. Pre-fire danger is mapped using a physically interpretable, rule-based score derived from six binary, pixel-level indicators representing necessary conditions for elevated danger: (1) fuel availability (vegetation presence), (2) fuel dryness (SWIR-based moisture proxies), (3) heat (2 m temperature and/or LST), (4) atmospheric dryness (relative humidity), (5) wind speed, and (6) antecedent moisture deficit (recent precipitation/soil moisture). This structure provides built-in explainability, because each pixel’s class is directly traceable to the specific conditions that triggered it.We demonstrate the workflow through a comparative analysis across four major Greek wildfire contexts, Attica, Euboea, Rhodes, and Evros, spanning different seasons and synoptic regimes. Using consistent pre-fire (multi-week) and post-fire compositing windows, we quantify how danger conditions co-occur prior to ignition, assess concordance between high-danger classes and observed fire perimeters, and relate pre-fire signatures to subsequent dNBR patterns, including differences associated with fuel structure, topography, and human exposure (proxied by proximity to roads and settlements from OpenStreetMap).To move beyond qualitative map interpretation, we complement the rule-based danger score with two lightweight, fully explainable modeling layers that quantify driver effects and test cross-region generalization. First, we fit generalized additive models (GAMs) using continuous satellite- and reanalysis-derived predictors to recover nonlinear response curves and threshold-like behavior. Second, we use a hierarchical ordinal logistic regression in which baseline levels and selected driver effects can differ by region, enabling us to identify which driver–severity relationships are consistent across Mediterranean landscapes and which are site-specific. We keep the models fully interpretable by reporting GAM response curves and logistic-regression odds ratios (with uncertainty), so predicted danger can be directly linked to physical drivers rather than opaque feature-importance scores. We generate all satellite/reanalysis-derived layers and danger/severity maps in Google Earth Engine, then export pixel-level predictor and outcome samples to fit the GAM and hierarchical logistic models in open-source Python, enabling transparent estimation of driver effects with uncertainty. Finally, we evaluate transferability using leave-one-region-out validation to identify where learned driver–danger relationships remain robust under differing regimes and where localized recalibration may be required for operational deployment.Keywordswildfire danger; burn severity; Google Earth Engine; Sentinel-2; Landsat 8; ERA5-Land; dNBR; generalized additive models; hierarchical logistic regression; explainable AI; transparent hazard mapping; Mediterranean ecosystems

15 May 2025
Verified
Conference Paper
Open access
DPDimitrios Papadomarkakis
Dimitrios Papadomarkakis
MFMaria-Sotiria Frousiou
Maria-Sotiria Frousiou
In the past decades, the comprehension of major geodynamic processes has mostly been dominated by computational and numerical models, with researchers generally avoiding the usage of analytical methods. The main reason for the latter lies in the fact that geodynamic systems and processes can be very challenging, and sometimes even impossible, to model analytically due to their high complexity and unknown factor. However, with the proper assumptions, the processes can be simplified in a way that analytical approaches can be utilized to model the occurring phenomenon, without compromising accuracy and realism. Overall, a subject that has been studied by various researchers, and as a result a great number of computational models have been proposed in the last two decades, is the development of the Rayleigh-Taylor gravitational instability in the interface between the subducting plate and the flowing mantle. This instability is induced by the density contrast between the two aforementioned layers, and particularly the fact that a denser fluid, in this case the flowing mantle, overlies a lighter fluid, the subducting plate. It has been illustrated that overtime with the development of the instability, characteristic plume-like shapes are formed that enter the hot flowing mantle and at some point even detach completely from the subducting plate. These plumes are then subjected to high, or even ultra high, pressure and temperature conditions making them newly formed metamorphic rocks that at some point in time are likely to get exhumed. The initiation and early development of the above discussed phenomenon was modeled in the present work by using linearised Navier-Stokes equations for two viscous fluids, with different density and viscosity values. From this analytical approach a basic methodology is proposed, capable of estimating the required growth rate of the instability in its early stages and also the critical wavelength, after which the plume is considered to have been fully formed and probably even detached from the plate. Additionally, the introduced function for the amplitude of the instability was correlated with the detachment potential of the plume from the downgoing plate. Furthermore, the proposed model was applied to the subduction setting of the Mediterranean ridge, located south of the island of Crete. Lastly, macroscopic observations from the broader Hellenides region were employed, by mostly examining the existing literature, to ascertain whether any such metamorphic rocks had indeed surfaced, thus confirming their exhumation.

20 January 2025
Verified
Conference Paper
Open access
MFMaria Sotiria Frousiou
Maria Sotiria Frousiou
Ioanna IliaIoanna Ilia
Ioanna Ilia
KKKasmas Kasmas
Kasmas Kasmas
IPIoanna Petropoulou
Ioanna Petropoulou
Landslide phenomena, acknowledged as significant geohazards affecting both human infrastructure and the natural environment, have been thesubject of intensive research aimed at pinpointing areas at risk of instability. This task involves the complex modelling of variables related to landslides, which requires both knowledge-based and data-driven methodologies. The challenge is heightened by the often intricate and obscure processes that trigger landslides be they natural or anthropogenic. Over the past two decades, the application of artificial intelligence, specifically machine learning algorithms, has brought a transformative approach to landslide susceptibility evaluations. These advanced methodologies, encompassing fuzzy logic, decision trees, artificial neural networks, ensemble methods, and evolutionary algorithms, have demonstrated notable accuracy and dependability. A significant recent development in this field is the incorporation of eXplainable AI (XAI) techniques into landslide susceptibility models. XAI tools, such as SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), offer a window into the previously opaque decision-making processes of AI models, thus demystifying the "black box" aspect of conventional AI systems.The primary aim of this study was to employ the XBoost algorithm and integrate SHAP methods for an in-depth landslide susceptibility assessment. The methodology was methodically divided into five distinct phases: (i)the creation of the inventory map, (ii)the selection, classification, and weighting of landslide-influencing variables, (iii)conducting multicollinearity analysis, (iv)applying and testing the developed model, and (v)evaluating the predictive performance of various models and analyzing the results.The computational work was performed using coding languages R and Python, while ArcGIS 10.5 was instrumental in compiling data and producing detailed landslide susceptibility maps. This study's efficiency was tested in the North-western Peloponnese region of Greece, known for its frequent landslide occurrences. Nine specific variables were considered: elevation, slope angle, aspect, plan and profile curvature, distance to faults, distance to river networks, lithology and hydrolithology cover and landslide locations, all contributing to the generation of training and test datasets. The Frequency Ratio method was applied to discern the correlation among these variables and assign weight values to each class. Multi-collinearity analysis further helped in identifying any collinearity among the variables.SHAP values were utilized to rank features according to their importance, offering a transparent view of variable contributions. The evaluation phase involved calculating the model's predictive power using metrics like classification accuracy, sensitivity, specificity, and the area under the success and predictive rate curves (AUC). This comprehensive approach combining XBoost and SHAP methods presents a refined model for understanding and predicting landslide susceptibility, aiming for more accurate and interpretable hazard assessments. The results highlight the high performance of the XBoost algorithm, in terms of accuracy, sensitivity, specificity and AUC values. SHAP method indicates that slope angle was the most important feature in this model for landslide susceptibility. Other features such as elevation, distance to river network, and lithology cover also contribute to the model's predictions, though to a lesser extent and with more mixed effects. Aspect, profile curvature, plan curvature, distance to fault, and hydrolithology cover appear to have a more moderate or minimal impact on the model’s predictions.