Abstract
The study of the ocean is challenging due to its complex nature, vastness and harsh environment that make its exhaustive sampling impractical. We present a set of eight multichannel seismic reflection sections acquired in the Southern and Southwestern margins of Portugal. These data were acquired between June and August 2001 by a seismic acquisition service company for energy resources exploration, and were carefully processed to imaging weak acoustic reflections in the water column originated at variations in ocean temperature and salinity (i.e., seismic oceanography). The data set provided here comprises a total length of 869.39 km, including the raw field records (i.e., shot gathers) for all seismic sections, in the time domain and with the relevant information for the water column, and the post-stack migrated seismic section in the depth domain. This comprehensive dataset provide a quasi-synoptic view of this complex oceanographic setting, the Gulf of Cadiz and Southwestern Iberian margin, and illustrate the potential of seismic oceanography data as complementary, and high spatial resolution data for oceanographic studies.
Subject terms: Ocean sciences, Geophysics
Background & Summary
Due to the vast and harsh nature of the ocean, its exhaustive sampling with direct observations at small spatial and temporal scales, and across multiple scales, is challenging. These sampling challenges result in a lack of adequate description of deep water oceanographic phenomena, particularly those occurring below 2000 m of water depth and at sub-mesoscale (i.e., 1–10 km)1–3. In conventional physical oceanographic data acquisition surveys, direct measurements of ocean properties are sparsely collected using various types of instrumentation, such as probes deployed from research vessels, attached to buoys and moorings, or installed in autonomous vehicles4. Typically, the data collected in these surveys fall over hydrographic data, such as water temperature, salinity and density, current velocity and direction, pressure and wave data, turbidity and light penetration, meteorological and acoustic data (i.e., ocean sound speed propagation). These measurements are often complemented by collecting water samples to assess the chemical composition of the ocean water.
These kinds of data acquisition surveys are valuable for observing and locally modelling the ocean in depth since they provide direct information about ocean water properties with high vertical resolution. However, their spatial distribution is highly limited (i.e., these measurements are sparsely spatially distributed), which limits the capability to obtain synoptic views of ocean processes and describe ocean processes across multiple scales. In summary, direct ocean measurements are scarce, deep ocean samples (i.e., below 200 m of water depth) are even scarcer, and these types of measurements have large horizontal distances from each other, which results in a considerable degree of uncertainty when spatially interpolating them. Predictions made in these areas exclusively with direct ocean measurements limit our capability to understand and forecast ocean dynamics at scales that are not directly observed.
Multichannel seismic reflection (MCS) data are commonly used to image the subsurface. However, their potential goes further than, subsurface imaging, by the identification of geo-energy prospects, groundwater characterisation, or engineering projects. Recently, multichannel seismic reflection data have been used to image the water ocean structure5–8 and spatially predict the ocean temperature and salinity distribution2,9–11. In seismic oceanography, acoustic waves formed by an acoustic source, typically an airgun towed by an acquisition vessel, propagates within the water column and are reflected back to the sea surface at contrasts in water masses with different temperature and salinity (i.e., interfaces with acoustic impedance contrasts). The reflected acoustic signal at the water column is detected by hydrophones towed behind the acquisition vessel (Fig. 1). In a typical seismic acquisition survey, data processing enhances information about the subsurface geology, discarding the information comprising the water column. In seismic oceanography, the data processing stage aims to enhance weak amplitude reflections happening in the water column and discard the information below the seafloor. The processed data allow imaging of the ocean sub-mesoscale (<10 km) along sections of hundreds of kilometres, representing a near-synoptic view of the ocean that can be used to interpret and spatially correlate ocean phenomena happening between direct measurements. Mainly due to energy industry efforts, multichannel seismic reflection data are present worldwide in most ocean basins, potentially providing useful information for poorly understood ocean processes.
Fig. 1.
Schematic representation of a typical MCS survey configuration with acoustic wavefields reflected at interfaces within the water column (i.e., seismic oceanography).
Here, we make available a unique data set composed eight two-dimensional multichannel seismic reflection sections distributed off SW Iberian Peninsula (i.e., the Gulf of Cadiz and the SW Portuguese margin). The data have industry quality, are long offset and were carefully processed with advanced seismic processing workflows to improve the ocean imaging quality. This data set aims to contribute to the generalisation of seismic oceanography in oceanographic studies, provide additional information to complement the existing direct measurements in this highly dynamic oceanographic region and bridge between two distant disciplines (i.e., oceanography and exploration geophysics). As we provide the raw and processed multichannel seismic reflection data, this data set also has a great potential for educational purposes.
Oceanographic Setting
The Gulf of Cadiz, located in the Southern part of the Iberian Peninsula, is a dynamic and complex oceanographic region in the Northern Atlantic. This oceanographic setting is characterised by complex water masses and structures formed by various interacting oceanographic processes. In summary, the warm and salty Mediterranean Water (MW) travels through the Strait of Gibraltar between 150 to 300 metres depth and mixes with North Atlantic Central Water12, mainly guided by bathymetry13, dropping to 800–1200 metres. The Mediterranean Undercurrent (MU) is divided into three cores following the seafloor; it is significant since it originates from the Mediterranean inflow/outflow through the Strait of Gibraltar, travels over the complex Portimão Canyon and reaches Cape St. Vincent, producing Mediterranean water eddies known as Meddies14–16, trapping in their cores the warm and salty water of the MW, that travel further into the North Atlantic. Along with Meddies, other oceanographic structures, such as meanders and filaments, are formed in the Portimão Canyon13.
The eastern end of the Azores Current has an important effect on the Gulf, contributing to eastward transports and zonal cyclonic recirculation caused by the Mediterranean Underflow’s entrainment of Atlantic Central Water17. Seasonal upwelling occurs throughout the summer, fuelled by stronger northerly winds, creating jets and filaments over the west coast of Iberia17. The sub-mesoscale ocean processes happening in this region and their interactions are captured in the data described in this work. The bathymetry of this region varies at a high rate, from very shallow on the shelf area to the deeper areas off the continental shelf break.
Methods
Seismic data
We processed a set of eight two-dimensional multichannel seismic reflection sections, acquired by TGS in the Portugal Deep 2000/2001 survey (Fig. 2), which target the imaging of geological prospects for potential oil and gas production. The acquisition survey occurred between May 2001 and March 2002 and was performed by Nan Hai 502 acquisition vessel, being the sections comprised in this work from June to August 2001. The data were acquired in the ocean using an airgun with a volume of 3680 cu.in, a 6000-meter active length streamer, with a 25 m group length (i.e., the distance between hydrophones). The recorded MCS data has 240 channels (i.e., traces) per shot record. Shot points are spaced by 37.50-meters, and the distance between the source and the first receiver (i.e., the near-offset) is 132 m. The data have a sampling rate of 2 ms and a total recording length of 12228 ms. The total length of the eight seismic sections is 869.39 km (Fig. 2). The individual, and detailed, characteristics of each seismic section are available in the data repository (10.5281/zenodo.14887000) under the Dataset description file.
Fig. 2.

Location of the two-dimensional multichannel seismic reflection sections off SW Portugal’s mainland. Bathymetry retrieved from General Bathymetric Chart of the Oceans21.
Seismic processing
Data processing is a critical step in seismic oceanography because the raw data collected during acquisition contains noise that can make extracting meaningful information about the ocean difficult. The field data were processed following a workflow designed to enhance the signal-to-noise ratio of the data, the weak reflection within the water column, preserve relative amplitudes and attenuate the effects of the direct arrival in the data (Fig. 3)2.
Fig. 3.
Seismic processing workflow applied to the two-dimensional MCS sections.
The seismic processing workflow differs from the conventional MCS processing by recovering and enhancing the reflections originating in the water column, while disregarding the information reflected at and below the seafloor. This represents a challenge not only due to the weak amplitudes of the reflected signal but also as a consequence of the direct wave energy. The direct wave is the acoustic signal that travels directly from the source to the receiver without being reflected or refracted within the water column. The direct wave is often the strongest and most easily identified signal in the seismic data and can often obscure the important reflections that contain information about the structure from the first few meters of the ocean.
We obtained, directly from TGS, the raw seismic data as individual shot files in SEG-D format. The SEG-D files corresponding to a given section were merged into a single file and converted into SEG-Y format18. Next, we utilise the associated p190 files to generate header geometry and assign position coordinates to each seismic trace within the dataset. This step is crucial for ensuring accurate calculations in subsequent processing stages. Although the data are already prepared for the subsequent steps, we vertically crop it to 6000 ms to reduce computing time while retaining reflections from the water column and the first meters below the seabed, which aligns with this processing workflow’s objectives. These data are equivalent to the field records, but vertically cropped and with the acquisition geometry stored in the seismic trace headers.
The resulting data are made available in the data repository to allow reproduction of the seismic data processing sequence described below (10.5281/zenodo.15722256).
Afterwards, we visually identify and interactively remove erroneous traces to ensure the integrity of the seismic data.
To effectively remove the direct wave arrival, we use a combination of linear moveout, a median filter and adaptive subtraction2 to each shot gather. To horizontalise the direct wave, we apply a linear moveout correction with a constant velocity of 1522 m/s for all sections. This velocity value was achieved after trial-and-error during processing. Then, we apply a median filter, with 14 ms time length and considering seven consecutive traces, to keep exclusively the horizontal reflections (i.e., the part of the signal that corresponds to the direct wave). The results from the median filter step are subtracted from the linear moveout corrected shot gathers to remove the horizontal reflections related to the direct wave and keep shallow reflections happening at the water column. Finally, we apply the inverse of the linear moveout correction and proceed with the seismic signal processing.
The processing continues with CMP sorting, by gathering seismic traces that share a common reflection midpoint. This sorting step allows a better velocity model prediction. Alternatively, and depending on data quality, this step can be applied before or after the direct wave removal. The CMP gathers were filtered with frequency Butterworth filters to preserve frequencies between 10 and 70 Hz.
Velocity analysis was then performed based on semblance at regular intervals along the MCS section. The semblance was computed with a velocity starting at 1450 m/s incrementing at each 5 ms for 100 velocities. The resulting velocity models were used to apply a normal moveout correction and correct the travel times dependency with offset on acoustic wave propagation velocities. The corrected CMP gathers were then stacked to increase the signal to noise ratio of the image. After stacking, the seismic sections illustrate the vertical profile of the water column’s structure. However, to make these images comparable to direct measurements, it is necessary to convert the vertical scale from two-way travel time (TWT) to depth. Here, for time-to-depth conversion18 we used a constant water velocity of 1515 m/s19, applying (Eq. 1 we obtain the seismic sections represented in vertical and horizontal meters:
| 1 |
Data Records
The processed two-dimensional MCS data described in the previous section are freely available in the Zenodo repository under the Creative Commons Attribution 4.0 International (CC BY 4.0) license (10.5281/zenodo.15722256)20. The data set includes the shot gathers with in formation about the acquisition geometry stored in the seismic trace headers in the time domain and post-stacked, depth-converted MCS section in standard SEG-Y format for each seismic section. The file header contains Source X and Y coordinates, CDP X and Y, and Trace ID. Additionally, the data is accompanied by a PDF summary describing the location, main data characteristics and contact information.
Figure 4 illustrates a processed seismic oceanography section, representing a snapshot of the state of oceanic structures at the moment of recording.
Fig. 4.
Stacked, depth converted seismic section PD00603. For location, see Fig. 2.
Technical Validation
All processing steps applied during seismic processing are widely utilised in MCS data processing and are consistent with state-of-the-art methodologies. The quality and validity of each processing step was assessed by visually inspecting the data and trace/gathers’ amplitude spectra before and after applying a given step.
Velocity analysis and modelling are the most user-dependent processes and will highly influence the quality of the obtained image and positioning of the seismic reflections. The proper positioning of the seismic reflectors in depth could be at risk if velocity analysis is done wrong, which could result in miss interpretation issues for further analysis. We check the consistency of the processed data by assessing the match between the seafloor reflection and the region’s current bathymetry21. Figure 5 shows the bathymetry profile downloaded from General Bathymetric Chart of the Oceans (GEBCO, https://data.ceda.ac.uk/bodc/gebco/global/gebco_2024/ice_surface_elevation/netcdf) 2024 grid21 on top of the seismic section. It is evident the overlap of the seafloor reflection from the seismic section with the seafloor bathymetry. Besides, on the seismic section top, we can also identify a reflection free zone down to ~130 m correspondent to the near offset (i.e., the distance between the source and the first hydrophone in the streamer), seismic reflections are visible from that depth down.
Fig. 5.
Section PD00603. The blue line represents the bathymetry extracted from GEBCO18. For location, see Fig. 2.
We validate the processed MCS section by interpreting the spatial continuity of reflectors. If the data are corrupted one would expect no spatial continuity pattern of the seismic reflections (i.e., discontinuous seismic reflections). All seismic sections show structured reflections down to 2 km depth in the water column. Below this depth, the water column is mainly reflection-free until reaching the seabed. This agrees with the expected depth of the thermocline in this region. The reflection-free zone indicated a well-mixed water regime22.
Focusing on the area with structured reflections, from 130 m down to 2 km of water depth, we can distinguish an array of oceanographic structures23,24, such as Meddies8,25 (Fig. 5, between 50 km to 100 km on the horizontal and ~400 m to ~1500 m of depth) and in other sections such as PD00609 and PD00612, some horizontal and parallel reflections that remind of thermohaline staircases26,27 (Fig. 5, between 1 km to 35 km on the horizontal and ~1200 m to ~1700 m of depth) also identified in section PD00606A, also fronts and other oceanographic processes28–32 (Fig. 5, throughout 5 km to 45 km on the horizontal and up to ~1000 m of depth). We compare the seismic images jointly with the seafloor bathymetry and a numerical model from the E.U. Copernicus Marine Service (CMEMS) Atlantic-Iberian Biscay Irish- Ocean Physics Reanalysis accessed in March 12th 2025 (product ID IBI_MULTIYEAR_PHY_005_002, 10.48670/moi-00029). The information from CMEMS was used to correlate oceanographic differences of temperature and salinity with the seismic reflections. Figure 6 shows section PD00603 with temperature and salinity gradients overlaid on top. This visual inspection allows matching the correspondence between reflections from ~750 m depth and in the lenticular feature around 65 km to 105 km with the spatial distribution of the different water masses, characterised by changes in ocean temperature and salinity.
Fig. 6.
Section PD00603 and corresponding temperature (a) and salinity (b) numerical models from CMEMS (IBI_MULTIYEAR_PHY_005_002, 10.48670/moi-00029).
Not limited to structures listed before, we can also detect an interaction between water and sediments in shallower regions by analysing the seismic signatures of such reflections33,34, for example, section PD00715, which is evident in shallower waters and rough seafloor topography.
It was expected to see a continuity of reflections crossing to the other direction in perpendicular sections; for example, in Fig. 7, the reflections present in section PD00524A cross to PD00609 and PD00612; however, that is not evident in those sections as a result of the time difference between acquisition, with more than two weeks difference between PD00524A to the perpendicular sections.
Fig. 7.
Composite vertical seismic section that comprises the following two-dimensional MCS sections: PD00612, PD00524A and PD00609. For location, see Fig. 2.
Acknowledgements
The authors acknowledge TGS for the donation of the SEGD raw data. Schlumberger and Shearwater for the academic licenses of Petrel and Reveal, respectively. AFD and LA gratefully acknowledge the support of the CERENA through its project FCT-UIDB/04028/2025. AFD is funded by the Fundação para a Ciência e a Tecnologia (Portuguese Foundation for Science and Technology) through the project PRT/BD/154661/2023. LA acknowledges the support of the China-Portugal Joint Laboratory on AI for Climate and Energy and the National Key R&D Program of China (2024YFE0114000). RM was supported by JUNO—Robotic Exploration of Atlantic Waters project—Refa 2021/0008—from the Luso-American Development Foundation. The authors acknowledge the insightful comments of the Editors and the two anonymous reviewers that increased the quality and broden the scope of the original manuscript. The authors gratefully acknowledge the support of the FCT through project 2024.07606.IACDC/2024.
Author contributions
A.F.D. conducted data processing and analysis. L.A. and R.M. designed, and organised the work. A.F.D. prepared the manuscript. L.A. and R.M. reviewed the manuscript. All authors contributed to the final manuscript preparation and approved the submitted version.
Data availability
The processed two-dimensional MCS data described in the previous section are freely available in the Zenodo repository under the Creative Commons Attribution 4.0 International (CC BY 4.0) license (10.5281/zenodo.15722256)20.
Code availability
No custom code was generated in this study.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Fang, Y., Chen, D., Deng, Y. & Huang, H. Analysis of influencing factors on displacement sampling performance of deep-sea gas-tight sampler. Front. Mar. Sci.9, 1075446 (2022). [Google Scholar]
- 2.Azevedo, L., Matias, L., Turco, F., Tromm, R. & Peliz, Á. Geostatistical Seismic Inversion for Temperature and Salinity in the Madeira Abyssal Plain. Front. Mar. Sci.8, 685007 (2021). [Google Scholar]
- 3.Pascual, A. et al. A Multiplatform Experiment to Unravel Meso- and Submesoscale Processes in an Intense Front (AlborEx). Front. Mar. Sci. 4 (2017).
- 4.Thomson, R. E. & Emery, W. J. Data Acquisition and Recording. in Data Analysis Methods in Physical Oceanography 1–186, 10.1016/B978-0-12-387782-6.00001-6 (Elsevier, 2014).
- 5.Holbrook, W. S., Páramo, P., Pearse, S. & Schmitt, R. W. Thermohaline Fine Structure in an Oceanographic Front from Seismic Reflection Profiling. Science301, 821–824 (2003). [DOI] [PubMed] [Google Scholar]
- 6.Carbonell, R. et al. Geophysical Oceanography? a new tool to understand the thermal structure and dynamics of oceans. in Eos, Transactions, American Geophysical Union vol. 87 OS16G-03 [Abstracts] (Honolulu, Hawaii, 2006).
- 7.Ruddick, B., Song, H., Dong, C. & Pinheiro, L. Water Column Seismic Images as Maps of Temperature Gradient. Oceanography22, 192–205 (2009). [Google Scholar]
- 8.Pinheiro, L. M. et al. Detailed 2-D imaging of the Mediterranean outflow and meddies off W Iberia from multichannel seismic data. J. Mar. Syst.79, 89–100 (2010). [Google Scholar]
- 9.Tang, Q., Hobbs, R., Zheng, C., Biescas, B. & Caiado, C. Markov Chain Monte Carlo inversion of temperature and salinity structure of an internal solitary wave packet from marine seismic data. J. Geophys. Res. Oceans121, 3692–3709 (2016). [Google Scholar]
- 10.Azevedo, L. et al. Geostatistical Inversion of Seismic Oceanography Data for Ocean Salinity and Temperature Models. Math. Geosci.50, 477–489 (2018). [Google Scholar]
- 11.Jun, H., Cho, Y. & Noh, J. Trans-dimensional Markov chain Monte Carlo inversion of sound speed and temperature: Application to Yellow Sea multichannel seismic data. J. Mar. Syst.197, 103180 (2019). [Google Scholar]
- 12.Ambar, I. & Howe, M. R. Observations of the Mediterranean outflow—I mixing in the Mediterranean outflow. Deep Sea Res. Part Oceanogr. Res. Pap.26, 535–554 (1979). [Google Scholar]
- 13.Serra, N. & Ambar, I. Eddy generation in the Mediterranean undercurrent. Deep Sea Res. Part II Top. Stud. Oceanogr.49, 4225–4243 (2002). [Google Scholar]
- 14.McDowell, S. E. & Rossby, H. T. Mediterranean Water: An Intense Mesoscale Eddy off the Bahamas. Science202, 1085–1087 (1978). [DOI] [PubMed] [Google Scholar]
- 15.Richardson, P. L., McCartney, M. S. & Maillard, C. A search for meddies in historical data. Dyn. Atmospheres Oceans15, 241–265 (1991). [Google Scholar]
- 16.Drillet, Y., Bourdallé‐Badie, R., Siefridt, L. & Le Provost, C. Meddies in the Mercator North Atlantic and Mediterranean Sea eddy‐resolving model. J. Geophys. Res. Oceans110, 2003JC002170 (2005). [Google Scholar]
- 17.Peliz, A., Dubert, J., Marchesiello, P. & Teles‐Machado, A. Surface circulation in the Gulf of Cadiz: Model and mean flow structure. J. Geophys. Res. Oceans112, 2007JC004159 (2007). [Google Scholar]
- 18.Yilmaz, Ö. Seismic Data Analysis: Processing, Inversion, and Interpretation of Seismic Data10.1190/1.9781560801580 (Society of Exploration Geophysicists, 2001).
- 19.Millero, F. J. & Kubinski, T. Speed of sound in seawater as a function of temperature and salinity at one atmosphere. J. Acoust. Soc. Am.57, 312–319 (1975). [Google Scholar]
- 20.Duarte, A. F., Azevedo, L. & Mendes, R. Dataset for seismic oceanography in the Gulf of Cadiz. Zenodo10.5281/zenodo.15722256 (2025). [DOI] [PMC free article] [PubMed]
- 21.GEBCO Bathymetric Compilation Group 2024. The GEBCO_2024 Grid - a continuous terrain model of the global oceans and land. NERC EDS British Oceanographic Data Centre NOC. 10.5285/1C44CE99-0A0D-5F4F-E063-7086ABC0EA0F (2024).
- 22.Fusco, G., Artale, V., Cotroneo, Y. & Sannino, G. Thermohaline variability of Mediterranean Water in the Gulf of Cadiz, 1948–1999. Deep Sea Res. Part Oceanogr. Res. Pap.55, 1624–1638 (2008). [Google Scholar]
- 23.Dickinson, A. & Gunn, K. L. The Next Decade of Seismic Oceanography: Possibilities, Challenges and Solutions. Front. Mar. Sci.9, 736693 (2022). [Google Scholar]
- 24.Song, H. et al. Progress and prospects of seismic oceanography. Deep Sea Res. Part Oceanogr. Res. Pap.177, 103631 (2021). [Google Scholar]
- 25.Biescas, B. et al. Imaging meddy finestructure using multichannel seismic reflection data. Geophys. Res. Lett.35, 2008GL033971 (2008). [Google Scholar]
- 26.Fer, I., Nandi, P., Holbrook, W. S., Schmitt, R. W. & Páramo, P. Seismic imaging of a thermohaline staircase in the western tropical North Atlantic. Ocean Sci.6, 621–631 (2010). [Google Scholar]
- 27.Duarte, A. F., Peliz, Á., Matias, L., Mendes, R. & Azevedo, L. Acoustic imaging of stable double diffusion in the Madeira abyssal plain. Sci. Rep.14, 8273 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Holbrook, W. S. & Fer, I. Ocean internal wave spectra inferred from seismic reflection transects. Geophys. Res. Lett.32, 2005GL023733 (2005). [Google Scholar]
- 29.Tsuji, T. et al. Two‐dimensional mapping of fine structures in the Kuroshio Current using seismic reflection data. Geophys. Res. Lett.32, 2005GL023095 (2005). [Google Scholar]
- 30.Klaeschen, D., Hobbs, R. W., Krahmann, G., Papenberg, C. & Vsemirnova, E. Estimating movement of reflectors in the water column using seismic oceanography. Geophys. Res. Lett.36, 2009GL038973 (2009). [Google Scholar]
- 31.Sheen, K. L., White, N. J. & Hobbs, R. W. Estimating mixing rates from seismic images of oceanic structure. Geophys. Res. Lett.36, 2009GL040106 (2009). [Google Scholar]
- 32.Krahmann, G., Brandt, P., Klaeschen, D. & Reston, T. Mid‐depth internal wave energy off the Iberian Peninsula estimated from seismic reflection data. J. Geophys. Res. Oceans113, 2007JC004678 (2008). [Google Scholar]
- 33.Vsemirnova, E. A., Hobbs, R. W. & Hosegood, P. Mapping turbidity layers using seismic oceanography methods. Ocean Sci.8, 11–18 (2012). [Google Scholar]
- 34.Chen, J.-X. et al. Geophysical analysis of abnormal seismic (oceanography) reflection characteristics of oceanic bottom boundary layer. Chin. J. Geophys. Chin.2148, 2161, 10.6038/cjg20160620 (2016). [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Citations
- Duarte, A. F., Azevedo, L. & Mendes, R. Dataset for seismic oceanography in the Gulf of Cadiz. Zenodo10.5281/zenodo.15722256 (2025). [DOI] [PMC free article] [PubMed]
Data Availability Statement
The processed two-dimensional MCS data described in the previous section are freely available in the Zenodo repository under the Creative Commons Attribution 4.0 International (CC BY 4.0) license (10.5281/zenodo.15722256)20.
No custom code was generated in this study.






