My research activity focuses on the development of innovative methods for flood hazard assessment, combining geomorphology, remote sensing, and data-driven approaches.

The main objective is to provide accurate, scalable, and computationally efficient tools for the identification of flood-prone areas, particularly in data-scarce and ungauged basins, where traditional hydraulic modeling is often infeasible.

This research integrates:

  • Digital Elevation Models (DEMs) and terrain analysis
  • Geomorphic indices and classification methods
  • Remote sensing observations
  • Machine learning techniques

1. Geomorphic Approaches for Flood-Prone Areas Delineation

A central research theme concerns the use of geomorphology as a proxy for flood processes, exploiting the information embedded in landscape structure.

Floodplains are shaped over long time scales by hydrological extremes, making terrain morphology a key indicator of flood susceptibility. Based on this principle, several studies have proposed simplified approaches relying exclusively on DEM-derived features (Manfreda et al., 2015; Samela et al., 2017).    

Key contributions include:

  • Identification of relevant geomorphic descriptors controlling flood exposure
  • Development of linear binary classifiers based on terrain attributes
  • Demonstration of their applicability across different spatial scales and environments

These approaches allow the delineation of flood-prone areas with limited data requirements, providing a viable alternative to traditional hydraulic simulations, especially in ungauged regions (Samela et al., 2016).  

2. The Geomorphic Flood Index (GFI) and Its Evolution

Among geomorphic methods, the Geomorphic Flood Index (GFI) represents a key advancement.

The GFI combines:

  • Hydraulic scaling relationships
  • Topographic information derived from DEMs

to estimate the likelihood of flooding at each point of the landscape (Samela et al., 2017).  

Research developments have shown that:

  • GFI provides high predictive capability across large domains
  • It requires minimal input data and computational effort
  • It can be applied for large-scale and global flood mapping

Subsequent advancements have extended the method:

  • Inclusion of coastal processes and inter-basin water transfers (Albertini et al., 2021)  
  • Development of enhanced formulations (GFI 2.0) accounting for:
    • River confluences
    • Backwater effects
    • Complex floodplain interactions (Manfreda et al., 2025)  

These improvements significantly increase the robustness of geomorphic flood mapping in complex hydrological settings.

3. Flood Mapping in Data-Scarce and Ungauged Basins

A major challenge in hydrology is the prediction of floods in ungauged basins, where hydrological observations are limited or absent.

My research addresses this issue by:

  • Developing low-complexity models based on geomorphic information
  • Reducing dependency on hydrological and hydraulic input data
  • Enabling rapid flood hazard assessment at large scales

Studies demonstrate that geomorphic classifiers:

  • Can reliably reproduce flood-prone areas derived from hydraulic simulations
  • Are transferable across different climatic and geomorphological contexts
  • Provide cost-effective tools for risk assessment and planning (Samela et al., 2017).  

This framework is particularly relevant for:

  • Developing countries
  • Large-scale applications
  • Preliminary risk screening and decision support

4. Integration of Remote Sensing and Data Fusion

Remote sensing plays a crucial role in modern flood monitoring, offering synoptic and repeatable observations of flood events.

Research contributions include the development of data fusion frameworks integrating:

  • SAR imagery (e.g., COSMO-SkyMed)
  • Interferometric coherence
  • Ancillary geomorphic information

A Bayesian Network approach has been proposed to combine these heterogeneous data sources, improving flood detection accuracy and reducing classification errors (D’Addabbo et al., 2016).  

This approach allows:

  • Modeling of complex dependencies among variables
  • Explicit representation of uncertainty
  • Improved detection in heterogeneous landscapes

5. Machine Learning and Predictive Flood Modeling

Recent research explores the integration of machine learning (ML) techniques with geomorphic and remote sensing data.

Key developments include:

  • Application of Random Forest models for flood susceptibility mapping
  • Integration of multi-source datasets, including:
    • Satellite observations
    • Geomorphic indices (e.g., GFI)
    • Environmental predictors

Results indicate that:

  • A limited number of predictors is sufficient for robust modeling
  • The inclusion of geomorphic indices significantly improves performance
  • ML models provide high generalization capability across different regions (Saavedra et al., 2025).  

Additionally, predictive models based on catchment characteristics enable:

  • Estimation of envelope flood extents
  • Reduction of dependence on calibration datasets
  • Application to large-scale and ungauged domains (Tavares et al., 2020).  

6. Toward Scalable and Operational Flood Mapping

The integration of geomorphic methods, remote sensing, and machine learning provides a pathway toward operational flood mapping systems.

These approaches offer:

  • Scalability from local to continental scales
  • Reduced computational costs
  • Applicability in real-time or near real-time scenarios

They represent a paradigm shift from:

  • Data-intensive, physics-based models
    to
  • Data-driven, hybrid, and scalable frameworks

with strong implications for:

  • Disaster risk reduction
  • Climate change adaptation
  • Sustainable land-use planning

Concluding Remarks

This research contributes to the advancement of hydrology by:

  • Leveraging geomorphology as a fundamental driver of flood processes
  • Integrating remote sensing and multi-source data
  • Developing data-driven and scalable models

The resulting methodologies provide robust, efficient, and transferable tools for flood hazard assessment, supporting both scientific understanding and practical applications in risk management.

References

  • Manfreda, S., et al. (2015). Flood-prone areas assessment using linear binary classifiers.  
  • Samela, C., et al. (2016). DEM-based approaches for flood-prone areas delineation.  
  • D’Addabbo, A., et al. (2016). A Bayesian Network for flood detection combining SAR imagery and ancillary data.  
  • Samela, C., et al. (2017). Geomorphic classifiers for flood-prone areas delineation.  
  • Samela, C., et al. (2017). Flood susceptibility dataset for the continental U.S.  
  • Tavares, R., et al. (2020). Predictive modelling of envelope flood extents.  
  • Albertini, C., et al. (2021). Flood-prone areas in coastal regions using GFI.  
  • Saavedra, J., et al. (2025). Flood susceptibility using Random Forest and geomorphic features.  
  • Manfreda, S., et al. (2025). Geomorphic Flood Index 2.0.  

About the author

He is Full Professor of Hydrology and Hydraulic Constructions at the University of Naples Federico II. He is currently chair of the IAHS MOXXI working group. His research primarily centers on hydrological modeling and monitoring. Recognizing the challenges posed by the complexity and limitations of traditional hydrological observations, he actively explores advanced and alternative monitoring techniques, such as the utilization of Unmanned Aerial Systems (UAS) coupled with image processing.