Geo-INQUIRE installation: HySEA - Earthquake and landslide generated tsunami simulations (TA2-531-3)

Hosting team: Prof. Jorge Macías Sánchez, Prof. Carlos Sánchez Linares (Universidad de Málaga (UMA), Spain) 

Principal investigator: Ran Novitsky Nof (Geological Survey of Israel, Israel)

Project title: Tsunami Risk Understanding and Simulation for Israeli Coastal Inundation

Project acronym: TRUSTIC

Project report ID: C1-TA2-531-3-1 (1^st Call)

Date of visit: 5 – 14 September 2025

Geo-INQUIRE Virtual Access:

Data/Products: A scientific paper based on this work is being prepared.

Project report:

The primary objective of the Tsunami Risk Understanding and Simulation for Israeli Coastal Inundation (TRUSTIC) initiative is to enhance Israel’s resilience and preparedness for tsunami hazards along its Mediterranean coast. This enhancement is achieved through the use of advanced, high-resolution wave propagation models and inundation maps generated with the Tsunami-HySEA model. In the current Transnational Access (TA), our focus is on evaluating the cost-effectiveness of using high-resolution Digital Elevation Models (DEMs) and bathymetry data. Additionally, we assess how variations in the width and morphology of the Israeli continental shelf influence potential tsunami hazards.

During his visit to UMA, Ran learned to use the Tsunami-HySEA model and to run it on the Leonardo supercomputer at CINECA. The first step of the workflow was to compile and harmonise bathymetry and onshore elevation datasets to build the model for the computational domain. A key challenge was integrating data from multiple sources with different spatial resolutions into a consistent set of nested, aligned submeshes suitable for high-performance computing (HPC) simulations with the Tsunami-HySEA code (Figure 1). The finest bathymetry data—0.5 m resolution Green LiDAR—covers a 1.5 × 4 km area from the shoreline to a depth of 15 m. In contrast, the coarsest dataset used is the 15-arcsecond GEBCO 2020 Grid (GEBCO Bathymetric Compilation Group 2020). For onshore elevations along the Israeli coast, 0.5 m LiDAR measurements were used.

The final grid model set-up consists of five nested grid levels, with the following resolutions:

• Level 1: 512 m/pixel
• Level 2: 128 m/pixel
• Level 3: 32 m/pixel
• Level 4: 8 m/pixel, with four submeshes
• Level 5: 1–2 m/pixel, consisting of one submesh covering the Maagan-Michael area

The tsunami source for the initial simulations is a simplified single fault based on the EFMS20 – CYCF001 parameters (Basili et al., 2024) (Figure 1).

To examine the sensitivity of the model to bathymetric resolution, we compared four simulation setups using time series extracted at 19 points (Figure 2):

1. High-resolution grid (2 m/pixel) with detailed bathymetry derived from Green LiDAR (0.5 m/pixel).
2. High-resolution grid (2 m/pixel) with coarse bathymetry based on interpolated 25 m/pixel data (Hall, 1993).
3. Low-resolution grid (8 m/pixel) with detailed bathymetry (same source as in setup 1).
4. Low-resolution grid (8 m/pixel) with coarse bathymetry (same source as in setup 2).

Comparison of model results

• High-resolution (2 m/pixel): detailed vs. coarse bathymetry (Models 1 and 2)

Time series show substantial differences in wave heights for points that become “onshore” in the interpolated coarse-bathymetry model (points 1, 2, 3, 20, 27, 28, 30). In some cases, the wave arrival time differs by ~30 seconds, and wave height differences reach up to ~10 cm (e.g., points 2, 3, 23) (Figure 3).

• Detailed bathymetry at different resolutions (Models 1 and 3)

When comparing the 2 m/pixel and 8 m/pixel detailed-bathymetry simulations, wave-height differences are negligible (Figure 4).

• Coarse vs. detailed bathymetry at low resolution (Models 3 and 4)

Results show similar behavior to the high-resolution comparison (Models 1 and 2), including comparable wave-height differences and shifts in shoreline classification (Figure 5).