Background & Summary

Active volcanoes continuously modify their morphology in response to eruptive activity and associated secondary phenomena such as landslides and wildfires.The reconstruction of the topography before and immediately after a volcanic event is essential for deriving qualitative and quantitative data on landscape changes and for volcanological purposes.

The surface of active volcanoes is generally reconstructed elaborating remote sensing data with different techniques. The main techniques and related data are: (i) Structure from Motion (SfM) applied to images acquired from Unoccupied Aircraft Systems (UAS) or crewed overflights; (ii) triangulation of elevation points acquired through Airborne Laser Scanning (ALS) and Terrestrial Laser Scanning (TLS); (iii) Stereo Photogrammetry applied to satellite imagery.

The above mentioned methodologies allow the reconstruction of topographies with a wide range of spatial coverage, spatiotemporal resolution and accuracy. The choice of the methodology mainly depends on two factors: the purpose for which the reconstructed topography is used and the cost-benefit assessment. Moreover, regarding geomorphological studies, it is also important to define the scale of analysis to select the most suitable methodology. Usually, for detailed geomorphological studies of active volcanoes, the mostly used methodologies involve data acquired by UASs, ALS and TLS1,2,3,4,5,6,7.

Stromboli Island is the northernmost volcano of the Aeolian Archipelago, located in the southern Tyrrhenian Sea, Italy (Fig. 1). It extends for about 12.58 km2 and represents the subaerial portion of a volcanic edifice, which rises ~3000 m above the seafloor8. At Stromboli Island, the volcanic activity occurs within the summit area, specifically in the crater terrace positioned at ~750 m a.s.l., and along the Sciara del Fuoco (Fig. 1). The volcano is characterised by an ordinary activity consisting of mild strombolian explosions (5–30 explosions per hour), ejecting lapilli and bombs up to a few hundred metres above the vent, and a few tens of metres beyond the edges of the crater terrace. These ejecta frequently roll down along the Sciara del Fuoco, a horseshoe-shaped depression involving most of the western flank. The mild activity is occasionally interrupted by energetic, impulsive explosions, lasting from tens of seconds to a few minutes, named major explosions and paroxysms, according to the magnitude and intensity of the event9,10. Concerning the paroxysmal activity, the most recent events occurred on July 11th 2024, August 28th 2019 and July 3th 2019. The two 2019 paroxysmal eruptions strongly changed the crater terrace morphology11, emitting a great amount of material, consisting of ash, lapilli, bombs and spatter clasts, which fell down the flanks and the summit of the volcano. The whole erupted mass is estimated in the order of 108 kg, according to Bisson et al.12 and references therein. Furthermore, the paroxysms generated not only pyroclastic flows into the Sciara del Fuoco, triggering small tsunamis at their sea entrance, but they also produced at least two gravity-induced flows13,14 and triggered wildfires, especially in the south-east and south-west sector of the island15,16.

Fig. 1
figure 1

Stromboli Island map. The Ground Control Points (GCPs) locations and the Airborne Lidar flight lines are visualized on a Google Earth image of the island. The insets at the top right of the figure show the location of Stromboli Island.

The period after the 2019 paroxysms, between August 28th 2019 and the end of July 2023, in addition to the typical Strombolian activity, was characterized by an increase in lava overflows, major explosions and Pyroclastic Density Currents17,18. In particular, in the considered period, 23 major explosions occurred, along with at least 30 overflows or lava flows (lasting up to 7 days) and 15 slope and/or rim failures generating Pyroclastic Density Currents. As a consequence, the morphology of both the crater terrace and the steep slopes of the Sciara del Fuoco was profoundly modified, with the generation of collapse scars, canyons (as occurred between October 9th and 15th 2022) and by the accumulation of lava flows and pyroclastic deposits.

Taking into account that the most recent high-resolution surface model of the entire island is dated 201219, and that the volcanic activity occurred between July 2019 and June 2023 generated relevant morphological changes both in the summit area and along the flanks, a dedicated Airborne Lidar survey to update the topography of the entire island was planned. Thanks to this survey, the Digital Surface Model (DSM) of Stromboli Island was updated to August 4th 2023, reproducing both natural and anthropic elements with a spatial resolution of 50 cm.

Methods

Airborne Lidar acquisition and DSM generation

The Lidar survey was carried out on August 4th, 2023, flying with a Piper PA 31 aircraft on the entire island of Stromboli, at an average elevation of 2500 m a.g.l. (above ground level), from 8.45 am to 11.13 am (local time). The 3D points acquisition was obtained by using the Leica Geosystems Terrain Mapper System, which integrates the photogrammetric camera Leica RCD30 (image sensor 10,320 × 7,752 pixels) with the Hyperion 2 Lidar sensor. This sensor, operating in the Near Infrared Region (λ = 1.064 μm), and programmed with a scan angle of 36° and a pulse repetition frequency of 900 kHz, acquired a total of 109,018,685 3D points, scanning the surface of the island through five flight lines (Fig. 1). Such planning assured 4 to 6 points per square meter, characterised by an instrumental vertical and horizontal error of ±5 cm (1 σ) and ±13 cm (1 σ), respectively. The points have been correctly geo-localised by using the latest generation Inertial Measurement Unit (IMU), capable of recording global navigation satellite system (GNSS)/Inertial measurements with high precision (https://leica-geosystems.com). The points were acquired in ellipsoid heights and geo-localised in WGS 84 (ETRF2000) UTM N33. Subsequently, the elevations were converted in ortho-metric heights (ITALGEO 2005 geoid), with the ConVERGO and Cartlab 3 software, by using the conversion grids (GK2) provided by Istituto Geografico Militare Italiano (IGMI). In addition to the x, y, and z coordinates, the Lidar sensor recorded a series of returns (in our case, 15 echoes), which allow the classification of different typologies of surfaces hit by the laser beam. The acquired data, stored in 23 LASer format (LAS) files, and containing the information related to geo-localization and return echoes, were elaborated on the TERRASOLID platform (www.terrasolid.com), by using dedicated filtering algorithms20 to eliminate outliers and classify the “ground” points from the “not ground” points. Subsequently, these two classes were merged (with a total of 71,298,241 points) and interpolated, to construct a Triangular Irregular Network (TIN) reproducing the elevation variation of the Stromboli surface (including barren soil, vegetation and anthropic features) through a vector model. Then, the vector model was transformed into a raster model (ESRI GRID) at a spatial resolution of 50 cm. The obtained model (Digital Surface Model - DSM) reproduces the topography of the entire island of Stromboli, reaching the maximum elevation (924.60 m a.s.l.) at Vancori (Fig. 2a,b). Figure 2a shows the elevation data of the island surface overlapped to the shaded relief of the DSM, while Fig. 2b distinguishes the different typology of features (ground, vegetation and anthropic), according to the elevation (see legend). Vegetation and anthropic features (“not ground” class) cover 29% of the entire island, except the western sector, dominated by the Sciara del Fuoco. In particular, the vegetation reaches 500 m a.s.l. in the west and north sectors, whereas it extends up 750 m a.s.l. in the southern portion. Anthtropic areas are located, for the most, below 100 m a.s.l., both at Ginostra and Stromboli villages (Fig. 2b). The ground/barren soil affects the entire summit of the volcano (above 700 m a.s.l.), a large portion of the east flank characterized by elevation ≥400 m a.s.l., and the Sciara del Fuoco. These three zones correspond to the areas more frequently affected, directly and/or indirectly, by volcanic activity. In fact, the Sciara del Fuoco is the result of several flank collapses, lava flows and minor lava overflows21,22,23, the summit area morphology is frequently modified due to the presence of the active crater terrace and, finally, the portion of the east flank at an elevation higher than 500 m a.s.l. is often affected by the fall down of ballistic products (lapilli, bombs and spatters), launched during the explosive activity, or by gravity-induced flows, as occurred during the paroxysm of July 3th 201912.

Fig. 2
figure 2

(a) Digital Surface Model of Stromboli Island. Elevation data overlaid on the shaded relief image. The contour lines every 10 meters of elevation are visualized in light brown colour. (b) Shaded Relief of Stromboli Island. Elevation data are visualised to distinguish ground, vegetation and anthropic elements. The contour lines every 10 meters in height are drawn in light brown colour.

Focusing on the summit area, the 2023 DSM shows significant morphological changes with respect to the 2012 one, especially in the two crater areas (north and south), as shown in Fig. 3. Overlapping the boundary of the 2023 crater areas on the morphological relief of 2012, the most evident morphological changes result: (i) a relevant enlargement of both crater areas, especially the south one, (ii) the formation and dismantling of volcanic cones within the crater areas, (iii) a significant accumulation (~30 m thickness) of pyroclastic products (mainly ash and lapilli) between the two crater areas.

Fig. 3
figure 3

Morphology of the Stromboli summit area dated 2023 (on the left) and 2012 (on the right). The red dotted lines indicate the boundary of the 2023 crater areas.

Data Records

The data record24 consists of a Digital Surface Model at a spatial resolution of 50 cm. The model was obtained by elaborating, on the ESRI platform (ArcGIS Map 10.5), the .las files acquired during the Lidar survey. Details of the Lidar survey are described in the Method section. The Digital Surface Model of Stromboli island is geocoded into WGS 84 (ETRF2000) UTM N33 Coordinate System. It is stored in GeoTIFF file format within the repository https://doi.org/10.13127/stromboli/airborne-lidar-2023. The model is shared under the CC BY 4.0 use license.

Technical Validation

The validation process of the DSM has been performed by using Ground Control Points (GCPs) measured during two distinct Global Navigation Satellite System - Real Time Kinematics (GNSS-RTK) field campaigns, carried out in July 2024, and retrieved from the available permanent stations of the GNSS monitoring network of Stromboli, managed by INGV-OE. A total of 41 GCPs were acquired, using anthropic and natural features considered stable in space and in time, and well identifiable both in the field and on the Digital Surface Model. The spatial distribution of the GCPs is shown in Fig. 1. It was planned to best cover the entire island, with the exception of inaccessible areas, such as Sciara del Fuoco and several unstable/steep zones.

One field campaign was carried out by using a Stonex S990A receiver, with differential corrections sent by the GNSS Sicilia, while the second one by using a Trimble R6 L1/L2 receiver, with differential corrections from the Spektra Network Real Time Kinematics (NRTK) local area network. Instead, the GNSS monitoring network consists of 4 permanent stations, with 3 of them operating during the Airborne Lidar survey: San Vincenzo station (SVIN), equipped with antenna and receiver Leica GR30/AR20 + LEIM, Punta Lena (SPLN) and Timpone del Fuoco (STDF) stations, equipped with antenna and receiver Leica GR25/AR20. The planimetric coordinates (X,Y) of the GCPs were acquired in WGS 84 (ETRF2000) UTM N33 geodetic cartographic system. The altimetric coordinates (Z), acquired in ellipsoid values, were subsequently transformed into orthometric values by using the ConVERGO software. The coordinates of the 41 GCPs (X,Y, Z) are reported in Table 1, as the site description and the horizontal and vertical errors calculated by the receiver during the acquisitions of each GCP.

Table 1 GCPs acquired by GNSS-RTK field campaigns and the permanent stations of the Stromboli GNSS monitoring network.

The vertical accuracy of the DSM was obtained as the root-mean-square-error (RMSE) of the residuals (ZL − ZGPS). Residuals represent the difference in Z between the elevation stored in the DSM cell (ZL) that contains the GCP, and the elevation value of the GCP measured on the field (ZGPS). If the GCP falls at a distance < 13 cm from the cell boundary, the DSM elevation value (ZL) is assumed to be the value stored in the adjacent cell. This distance of 13 cm derives from the instrumental horizontal error of the Lidar sensor at 1 σ. The residuals calculated for the 41 GCPs range from 0.001 m to 0.312 m as shown in Table 2.

Table 2 Comparison in elevation between the Lidar and the GCPs.

The vertical accuracy of the DSM results in 8 cm and such value is consistent with the instrumental vertical error (±5 cm) of the Lidar sensor used in the acquisition survey. In addition, to evaluate the reliability of the obtained accuracy, the Residual and the Total Error (Table 2) were compared for each GCP. In detail, the Total Error is calculated as the sum of the instrumental errors of GPS, Lidar and the uncertainty of ±0.5 cm referred to field measurements. The DSM is considered validated when the Residual is minor than the Total Error. This is true for 90% of the acquired GCPs, as shown in the plot of Fig. 4, confirming the high quality of the DSM.

Fig. 4
figure 4

Residuals (ZL − ZGPS) between DSM and GPS elevations and the sum of the vertical instrumental errors (Total Error) referred to each GCP (Table 1). The Y axis reports the value of the Total Error and Residuals expressed in meters (Z).