Evangelista et al. (2025) Non‐Linear Influence of Reservoir Initial Condition on Flood Reduction

Identification

• Journal: Journal of Flood Risk Management
• Year: 2025
• Date: 2025-10-20
• Authors: Giulia Evangelista, Miriam Bertola, Günter Blöschl, Pierluigi Claps
• DOI: 10.1111/jfr3.70142

Research Groups

• Department of Environment, Land and Infrastructure Engineering, Politecnico di Torino, Torino, Italy
• Institute of Hydraulic Engineering and Water Resources Management, Technische Universität Wien, Vienna, Austria
• Department of Civil, Chemical, Environmental, and Materials Engineering, Università di Bologna, Bologna, Italy

Short Summary

This study investigates the non-linear influence of initial reservoir storage on flood peak attenuation efficiency across approximately 250 large dams in Italy, revealing that dam performance significantly deteriorates with increasing flood return periods, especially when initial storage is high, and that commonly assumed full-reservoir conditions lead to overly conservative flood risk estimates.

Objective

• To investigate, for the first time, the non-linear effects of reduced initial storage on the relationship between flood peak attenuation efficiency and flood return period for approximately 250 large dams in Italy.

Study Configuration

• Spatial Scale: National scale, focusing on 247 large dams across Italy.
• Temporal Scale: Flood events with return periods ranging from 5 to 10,000 years; historical reservoir storage data from 1996 to 2023 (Sardinia) and 2006 to 2014 (Piedmont and Aosta Valley); historical flood timing data from 1960 to 2020.

Methodology and Data

• Models used:
  • Simplified hydrological model (Rational formula) for estimating inflow hydrographs.
  • Full hydraulic routing (continuity equation of lakes) for flood peak attenuation assessment.
  • Empirical classification criterion based on dam elevation and Köppen-Geiger climate classification for estimating initial storage levels.
• Data sources:
  • TILDE (The Italian Large Dams datasEt, Evangelista et al. 2025) for structural features of dams and upstream catchment attributes.
  • Italian Department of Dams for spillway geometries and daily historical observations of stored volumes (Piedmont and Aosta Valley).
  • River Basin Document Center (CEDOC) for monthly average historical observations of reservoir volumes (Sardinia).
  • Blöschl et al. (2017) methodology for average seasonal timing of floods from gauged sites across Italy.
  • Braca et al. (2024) for monthly precipitation data.

Main Results

• Most dams have an Initial Storage Level (ISL) between 0.5 and 0.9, with a peak around 0.7, suggesting a potential flood control storage of approximately 15% for all dams.
• Flood peak attenuation efficiency (η) is highly sensitive to the initial storage level and deteriorates sharply as flood return periods (T) increase, particularly when initial storage is high.
• Seven distinct classes of dams were identified based on their flood attenuation capacity relative to flood severity, highlighting non-linear and threshold effects.
• Doubling the incoming flood peak (e.g., by changing the runoff coefficient from 0.5 to 1) leads to a non-linear and larger decrease in attenuation efficiency as the system's capacity limits are approached.
• The commonly assumed full-reservoir condition (ISL = 1) yields overly conservative estimates: approximately 20% of the examined dams reach their maximum allowed water level for return periods of 100 years or less under this assumption combined with a runoff coefficient of 1.
• Dam classification into groups is primarily governed by the ratio of flood volume to available storage capacity (V1000/Vav) and the dimensionless ratio of the storage coefficient to the time of concentration (δ), rather than direct geographical location.
• A considerable number of dams (Group 7, 137 dams) do not produce outflow even for large floods, attributed to advantageous hydrological conditions, geometric characteristics, or significantly low ISLs at flood onset.

Contributions

• First national-scale analysis in Italy explicitly quantifying the non-linear effects of reduced initial storage levels on the relationship between flood peak attenuation efficiency and flood return period for a large sample of real reservoirs.
• Development of an empirical classification of dams based on their flood attenuation behavior (η-T curves), distinguishing between gradual decline and threshold effects.
• Quantification of the overly conservative nature of traditional full-reservoir assumptions in dam safety assessments, providing data-driven insights into actual operational conditions.
• Collection and analysis of historical reservoir storage data for a significant number of Italian dams, which is an original contribution given the typical scarcity of such data.
• Provides practical recommendations for optimizing reservoir operations under current and future climatic conditions by explicitly quantifying how reduced initial storage can enhance flood mitigation.

Funding

• European Union Next-GenerationEU (National Recovery and Resilience Plan—NRRP, Mission 4, Component 2, Investment 1.3-D.D. 1243 2/8/2022, PE0000005-Spoke TS 2)
• RETURN Extended Partnership

Citation

@article{Evangelista2025NonLinear,
  author = {Evangelista, Giulia and Bertola, Miriam and Blöschl, Günter and Claps, Pierluigi},
  title = {Non‐Linear Influence of Reservoir Initial Condition on Flood Reduction},
  journal = {Journal of Flood Risk Management},
  year = {2025},
  doi = {10.1111/jfr3.70142},
  url = {https://doi.org/10.1111/jfr3.70142}
}