Data-driven flood hazard zonation of Italy.

Area of Science:

• Hydrology and Water Resources Engineering
• Geomorphology
• Environmental Science

Background:

• Accurate river flood inundation mapping is crucial for hazard assessment and mitigation.
• Existing physically-based models can be data-intensive and computationally expensive.
• Data-driven approaches offer a promising alternative for flood delineation.

Purpose of the Study:

• To introduce and validate Flood-SHE, a novel data-driven statistical procedure for delineating river flood inundation areas.
• To assess the accuracy and applicability of Flood-SHE across diverse physiographical settings in Italy.
• To evaluate the predictive capability of Flood-SHE-derived inundation zones for flood risk assessment.

Main Methods:

• Application of Flood-SHE across 23 Italian River Basin Authorities (RBAs).
• Utilized existing flood zoning data as the dependent variable and six hydro-morphometric variables from a 10m x 10m DEM as covariates.
• Trained and validated a large number of statistical models (2208 trained, 3072 validated) for different covariate combinations and return periods.

Main Results:

• Flood-SHE accurately delineated potentially inundated areas, closely matching existing flood zonings derived from physically-based models.
• The data-driven method delineated larger inundation areas compared to physically-based models, with variations depending on input data quality.
• Analysis confirmed that Flood-SHE inundation zones are effective predictors of flood risk to human populations.

Conclusions:

• A small set of hydro-morphometric terrain variables is sufficient for accurate inundation zoning in various settings.
• Flood-SHE demonstrates potential for application in areas lacking traditional hydrological modeling-based flood zoning.
• The developed data-driven inundation zonings can guide decisions on where improved flood hazard assessments are needed.