Dataset of numerical simulations of landslide-generated tsunamis at Stromboli volcano

Description: This dataset supports the findings of the article "Sensitivity analysis and scaling of tsunamis generated by granular flows at Stromboli volcano: A numerical modeling approach" submitted to Natural Hazards and Earth System Sciences(NHESS). It contains the input parameters and the resulting tsunami waveforms for all scenarios simulated using the multilayer non-hydrostatic HySEA model and measured at 11 gauge positions around Stromboli island. File Content: The dataset is provided as a standalone SQLite3 database (database_all_with_ts.db). It contains a single table named simulations, with the unique identification number of each simulation, the parameters characterizing the granular flow source, and the waveform data. Field names are: id: Unique simulation identifier. xutm, yutm: Source horizontal position in WGS-84 UTM Zone 33N coordinates (which defines the initial elevation $q$). vol: Source volume ($V$) [m$^3$]. rho: Source density ($\rho$) [kg/m$^3$]. angle1_Pouliquen$-$angle3_Pouliquen: Source basal friction angles ($\delta_1 -\delta_3$) [degrees]. WLfriction: Water-landslide coupling coefficient ($m_f$) [m$^{-1}$]. azimuth: Transect direction azimuth of the source position ($D$) [degrees]. ts: Simulated tsunami time series at the virtual gauges. Time Series Structure (ts array): For each simulation, the ts field contains a 2D array (300 rows × 12 columns), representing the time series of the free-surface elevation (waveforms) recorded for 600 s every 2 s. Column 1: Time (in seconds). Columns 2 to 12: Wave heights (in meters) recorded at the 11 virtual gauges distributed around Stromboli island. The gauge locations correspond exactly to those mapped in Figure 1 of the main article and their coordinates are shown below: N Name X [m] Y [m] Depth [m a.s.l.] 0 Punta dei Corvi 516,789 4294442 -54 1 - 518,348 4295986 -46 2 - 516,788 4294437 -54 3 Punta Labronzo 518,427 4296006 -33 4 - 514,822 4299420 -1,474 5 - 514,871 4298650 -1,338 6 - 517,871 4295280 -98 7 - 521,152 4295020 -2 8 - 520,599 4295436 -9 9 Stromboli 521,020 4294513 -12 10 Ginostra 516,350 4293057 -15 How to read the data (Python): The time series (ts field) are stored as binary NumPy arrays. To properly extract them into a pandas DataFrame or NumPy array, you must register a custom SQLite converter in Python before executing your query. Here is a working snippet: import sqlite3import pandas as pdimport numpy as npimport io # Custom converter to read the binary numpy arraysdef convert_array(text): out = io.BytesIO(text) out.seek(0) return np.load(out) sqlite3.register_converter("array", convert_array) # Connect to the database enabling type detectionconn = sqlite3.connect('database_all_with_ts.db', detect_types=sqlite3.PARSE_DECLTYPES) # Example: Read the entire database into a pandas DataFramedf = pd.read_sql_query("SELECT * FROM simulations", conn)conn.close() # Now, for example, df['ts'][0] is a fully functional numpy array corresponding to the simulation id = 1.