Geospatial Analysis Methods for Hydrological Risk Modeling

Climate change is increasing the frequency and intensity of extreme hydrological events, posing significant risks to alpine communities. This thesis explores an innovative approach to flood risk assessment in mountainous regions by combining high-resolution surface modeling with advanced hydrodynamic simulations. Using photogrammetry and LiDAR technology, detailed digital surface models were generated and integrated into a high-performance flood model to improve forecasting accuracy. The study focuses on Bardonecchia, a town in western Piedmont at the end of Val di Susa, which experienced a severe flood event in August 2023. Triggered by intense, localized rainfall, this event led to a destructive debris flow, highlighting the need for fast, reliable, and easily deployable flood prediction methods. To address this challenge, high-resolution digital terrain data provided by DigiSky was incorporated into the Basement software developed by ETH Zurich. This software’s 2D hydrodynamic modules enabled accurate water flow simulations by factoring in key parameters such as surface roughness, land cover classifications, and lithological data. Rainfall inputs were derived from Intensity-Duration-Frequency (IDF) curves calculated using probabilistic methods like the Gumbel distribution. These rainfall estimates, with varying return periods and spatial interpolation via kriging techniques, were processed using a MATLAB script to convert precipitation into effective runoff through the Soil Conservation Service-Curve Number (SCS-CN) method. Model validation was conducted by comparing simulated flood wave arrival times and peak discharge values with empirical formulas commonly used in engineering practice. The simulation, based on a 24-hour rainfall event totaling 180 mm, showed a peak intensity of 68.5 mm/h at the 12th hour. The model estimated a lag time of approximately 45 minutes and an average flow velocity of 6.5 m/s, closely aligning with observed hydrological behavior. Additional sensitivity analyses explored the influence of key variables such as Manning’s coefficient and rainfall inputs, assessing the model’s response under different conditions. By integrating high-resolution terrain data with advanced flood modeling techniques, this approach enhances flood forecasting capabilities in mountainous areas. Compared to traditional methods, it provides a more accurate representation of localized hydrological responses, improving flood risk assessments and optimizing infrastructure design. This methodology can help prevent both over-engineering and under-engineering, leading to more effective and adaptive flood mitigation strategies. To ensure its broader applicability, further validation is recommended in different alpine environments. The severe flooding event in Cogne, Aosta Valley, in June 2024, highlights the need for improved hydrodynamic modeling across diverse contexts. Applying this methodology to additional basins will refine its predictive capabilities and confirm its robustness, ultimately contributing to more effective flood risk management strategies for alpine communities.