This study was based on selected profiles from the PolandSPAN^® regional seismic survey located above the southwestern edge of the East European Craton in Poland (cf. Krzywiec et al., 2013, 2014). This survey consists of approximately 2200 km of onshore seismic reflection profiles acquired with high-end parameters such as broadband sweep (2–150 Hz), 960-channel symmetric spread, 12 km long offsets, long record lengths (12 s), tight station spacing (25 m), 2 ms sampling rate and high nominal fold (480). Nominal record length was 12 s, but in the field uncorrelated data down to 28 s were recorded. All the data were processed up to pre-stack time migration (PreSTM) and pre-stack depth migration (PreSDM). Additional details on acquisition and processing workflows can be found in Mężyk et al. (2019). The seismic data used in this study were obtained as fully processed and migrated PreSTM and PreSDM datasets, and no additional reprocessing was performed. Comparison of these acquisition parameters with parameters used in case of other deep seismic reflection onshore surveys illustrates well the rather unique character of the PolandSPAN^® survey. For example, ECORS (Étude de la Croûte Continentale et Océanique par Réflexion et Réfraction Sismiques) deep reflection survey acquired across the Pyrenees employed much larger station spacing of 50 or 80 m, longer (4 or 8 ms) sampling rate, and lower nominal fold (90–120) (Choukroune, 1989; Choukroune et al., 1990). POLCRUST deep regional seismic profile located in SE Poland was acquired with shorter offsets (maximum 10 485 m), larger station spacing (60 m), and lower nominal fold (175) (Malinowski et al., 2013, 2015). Such approach to data acquisition and processing of the PolandSPAN^® survey resulted in unparalleled imaging of the entire Phanerozoic sedimentary cover and crust, down to MOHO (Malinowski, 2016). So far, PolandSPAN^® data have been used in order to reinterpret the Teisseyre-Tornquist Zone (Mazur et al., 2015, 2016) and to decipher and interpret deep tectonic features within the crystalline basement (Mężyk et al., 2019, 2021), shed a new light on Ediacaran rifting of the cratonic edge (Krzywiec et al., 2018), helped to better constrain deeply buried Caledonian (Mazur et al., 2016) and Variscan (Krzywiec et al., 2017a, b; Kufrasa et al., 2020) orogenic fronts, and to reinterpret regional tectonics and depositional evolution along the cratonic edge during the Late Cretaceous inversion of the Polish Basin (Stachowska and Krzywiec, 2023). Recently, PolandSPAN^® profiles have been used to provide new insight on lower Carboniferous igneous intrusions within the Baltic and the Lublin basins in northern and southeastern Poland, respectively (Krzywiec et al., 2024; Poprawa et al., 2024).

For this study, three profiles of the PolandSPAN^® survey have been analysed: PL-5400 (Fig. 3), PL-1200 (Fig. 4) and part of PL-1100 (Fig. 5). They were calibrated using data from deep wells that drilled to the base of the Lower Paleozoic sedimentary cover or reached top of the crystalline basement (cf. Krzywiec et al., 2013, 2014; Mazur et al., 2016; Stachowska and Krzywiec, 2023).

Quality of seismic imaging directly depends on seismic resolution, which is determined by the relationship between dominant frequency and interval velocity (e.g. Faleide et al., 2021). The final resolution of seismic data is also influenced by various processing algorithms, including those used for migration of seismic data (Sheriff, 2002; Cartwright and Huuse, 2005). Seismic resolution decreases with depth which is a combined effect of two main factors: (i) general increase of velocity with depth due to mechanical and chemical compaction, and (ii) increased attenuation of higher frequencies with depth (Brown, 2011; Bjørlykke, 2015; Faleide et al., 2021).

The vertical seismic resolution, regarded as the limit of vertical separability of seismic events/features, is defined as quarter of a seismic wavelength used in a given seismic survey (λ/4; e.g. Brown, 2011), whereas precise limits of visibility or detectability can vary depending on the quality of the seismic data (Faleide et al., 2021; see also Brown, 2011, for more details). All of the above information was essential for conducting a detailed analysis of the vertical seismic resolution in areas where magmatic intrusions had been identified. The objective of analysis of vertical resolution was to determine the key parameters necessary for determining, or estimating, the thickness of igneous intrusions imaged within the basement of the Baltic Basin. Estimation of vertical seismic resolution was performed in two complementary steps: (i) advanced statistical tuning analysis using wavelets extracted from seismic data, and (ii) wedge-based forward modelling. The tuning analysis provides data-driven estimates of dominant frequency and vertical resolution, whereas wedge modelling illustrates the physical basis of tuning effects and waveform interference.

The extracted wavelet represents an effective statistical wavelet derived from final pre-stack time-migrated (PreSTM) seismic data provided by the PolandSPAN project, following complete regional processing and imaging procedures. It reflects the combined source signature and propagation effects rather than a true source wavelet. Wavelet estimation was performed using a frequency-matching approach, which computes a zero-phase wavelet whose spectrum matches that of the seismic data within the selected time window and does not impose a predefined waveform shape. No dominant multiple energy was observed within the analysed time windows. Wavelet extraction was performed separately for three polygons located in different parts of the intrusion in order to capture potential lateral variations in frequency content and tuning behaviour. The resulting dominant frequencies were similar, justifying the use of an average value in subsequent wedge modelling. The extracted wavelets were used primarily to estimate the dominant frequency content of the seismic data and were not directly used as input for forward or wedge modelling. Therefore, they should be treated as statistical approximations of the effective seismic bandwidth rather than representations of the local tuned response of the analysed intrusions.

The PolandSPAN data were acquired using Vibroseis sources, and therefore the recorded wavelet corresponds to a correlated Klauder-type wavelet. However, the present study is based on fully processed and migrated seismic data and does not aim to reproduce the original source signature. An effective statistical wavelet was first extracted from the pre-stack time-migrated data using a frequency-matching approach, yielding dominant frequencies of approximately 29–30 Hz. Based on this result, a zero-phase Ricker wavelet with an assumed dominant frequency of 29 Hz was finally adopted as a simplified approximation of the effective seismic bandwidth. The use of an idealized Ricker wavelet ensures that tuning effects observed in the modelling arise from interference within the modelled geological structures rather than from characteristics potentially present in the extracted wavelet. This approach is suitable for investigating tuning effects and thickness-dependent waveform interference, rather than for reproducing the detailed Vibroseis waveform or analysing absolute amplitudes.

Uncertainties in interpretation of seismic data have been well defined (Bond, 2015; Alcalde et al., 2017). They could be related (although not limited) to two major issues. One is complex geometry of geological structures and associated rapid lateral and/or vertical changes in velocities of seismic waves within particular geological complexes. Seismic imaging of steeply dipping strata is a common problem in fold-and-thrust settings that might result in incorrect structural interpretation of fold-and-thrust structures (e.g., Morse et al., 1991; Camerlo and Benson, 2006; Krzywiec et al., 2017a; Li and Mitra, 2020a,b). Significant geometrical disturbances caused by velocity pull-up effects are commonly visible on seismic data from areas characterized by salt tectonics and presence of thick high-velocity evaporites, (e.g., Pietsch et al., 2012; Marzec et al., 2013). Similar effect could be caused by thick high-velocity carbonate buildups (e.g., Słonka et al., 2025). Resolution of seismic data and the role of tuning effects are another important issues. For beds characterized by acoustic impedance different from acoustic impedance of rocks located below and above it, its thickness versus vertical seismic resolution determines possibility of interpreting base and top of that bed on seismic data. For thin beds the tuning effect (i.e. superposition of waves reflected from bed's base and top) might dominate recorded seismic wavefield, and as a result one reflector is imaged on stacked and migrated seismic data (cf. Widess, 1973). Very good example is provided by a very characteristic seismic horizon located at the base of the Miocene sedimentary infill of the Carpathian foreland basin that can be traced on seismic data for tens and hundreds of kilometers in Poland, Ukraine and Romania (cf. Krzywiec, 2001). It is related to a relatively thin (few tens of meters maximum) Badenian anhydrites – their small thickness prevents unequivocal detection of separate seismic horizons related to their top and base, and, as a result, a single composite seismic horizon is visible on most of available 2D and 3D seismic data. A similar situation could be associated with deep magmatic sills (e.g. Wrona et al., 2019). All these interpretation uncertainties could be assessed and at least partly mitigated by using seismic forward modelling techniques based on initial interpretation of seismic data. For deep structures beyond reach of wells, parameters such as velocities and densities must be indirectly estimated – of “guessed” – using e.g. literature, information from outcrops etc., and then tested by comparison between real and synthetic, i.e. modelled, seismic data. A high degree of similarity between these two datasets indicates that the final proposed geological model, including its geometry and petrophysical parameters, could be regarded as viable.

Seismic forward modelling was used to constrain the continuity, thickness, and lithology of inferred igneous intrusions identified on seismic data. The applied modelling methodology was based on a full-wave algorithm that accounts for multiple reflections and provides a more realistic representation of wave propagation in complex media than standard ray-tracing methods. In contrast to ray-tracing approaches, full-wave modelling simulates the complete seismic wavefield and therefore allows analysis of interference effects, reflection amplitudes, and tuning phenomena associated with thin geological bodies, such as igneous intrusions. The vertical incidence method was used to solve the wave equation by simulating the vertical propagation of P waves. This simple algorithm quickly evaluates the arrival time and amplitude of reflected waves, making it a very useful tool for seismic analysis. It is very effective methodology for studying the direct seismic response to lithology, and for verifying geological concepts when interpreting seismic data. The same approach has been recently tested and successfully used to study Upper Jurassic carbonate build-ups in southern Poland (Słonka et al., 2025). Similar forward modelling techniques have been widely applied to investigate also the seismic expression of igneous intrusions and related structures in seismic reflection data (e.g., Hansen et al., 2008; Magee et al., 2015; Eide et al., 2017; Wrona et al., 2019; Köpping et al., 2022).

For theoretical wave field calculations, we used a zero-offset variant, which simulates synthetic stacked and migrated seismic profiles. Seismic modelling involves simulating seismic wave propagation in two-dimensional, geologically representative numerical models. Each modelling scenario begins with the creation of seismic-geological model in depth domain that consists of seismo-geological layers i.e. polygons with assigned P-wave velocities and densities (cf. Fagin, 1991; Rabbel et al., 2018; Lecomte et al., 2015; Li and Mitra, 2020a, b; Słonka et al., 2025). Following the construction of various seismic-geological models and definition of the acquisition parameters, simulations were performed using the specified theoretical wavelet.

We performed a detailed analysis of the role of seismic tuning on analysed seismic data using statistical wavelets within the time windows that covered igneous intrusions visible on seismic data. This analysis was crucial for subsequent steps, including seismic forward modelling and final estimation of vertical seismic resolution within specified data intervals where intrusions are visible. Ultimately, this has provided an approximation of the intrusion thickness. The analysis was carried out for seismic profiles PL-5400 and PL-1200 (Figs. 3 and 4), where high-reflectivity seismic events are located shallower (approximately 3.5–4.5 s or 7–9 km, and 3–4 s or 6.5–8.5 km, respectively) and are better visible than on profile PL-1100 (Fig. 5), where analysed deep seismic features are located considerably deeper (approximately 6–7.5 s or 16–19 km).

Seismic tuning refers to a phenomenon where waves from closely spaced reflections interfere and this might significantly affects apparent amplitude and apparent thickness of the resulting seismic reflector(s). This effect limits the vertical resolution of seismic data, which is important in so-called thin-bed analysis as reflections from top and base of a layer (bed) thinner than a quarter of the seismic wavelength (λ/4) cannot be reliably resolved as separate seismic horizons on interpreted seismic data (Widess, 1973). For seismically thin beds (layers), seismic reflections from its top and base overlap and undergo interference, producing a single event of high amplitude. For geological layers characterized by thickness larger than λ/4, separate seismic reflections related to their tops and bases may be distinguished. The tuning thickness is the bed thickness for which two events related to top and base of the bed become indistinguishable on seismic data (e.g. Widess, 1973; Kallweit and Wood, 1982; Sheriff, 2002).

Detailed tuning analysis relies on tracking the tuning thickness based on the wavelet frequency content. To provide the necessary number of seismic trace samples for statistical analysis and to avoid distortion in the calculations, the wavelet extraction windows were set to 500 ms. The three study polygons, two on profile PL5400 and one on profile PL-1200 have been examined in order to determine the maximum tuning thickness and the dominant frequency of each wavelet extracted within the specified time windows. The results have been then compared, and the average value of the calculated dominant frequencies was used to determine the vertical seismic resolution in subsequent stages of our study.

Part of the PL-5400 seismic profile shown in Fig. 6A reveals the south-western part of complex system of strong-amplitude reflections related to inferred igneous intrusion (cf. Fig. 3). The wavelet was extracted from all seismic traces within the specified time window (marked by the black dotted rectangle) using a conventional statistical method. To ensure full comparability of the results, all analysed wavelets were 120 ms long and sampled at 2 ms. The calculated amplitude spectra were characterized by relatively low noise content (Fig. 6B). Figure 6C presents the results of the tuning analysis for the extracted wavelet, together with an estimation of the dominant frequency.

The tuning thickness chart (with all values in two-way travel time) consists of: (i) actual time thickness, which is the actual bed thickness expressed in time units; (ii) apparent time thickness, which represents the isochron time between the top of the layer (peak) and the base of the layer (trough), as measured from seismic data; (iii) a graph of the normalized peak-trough amplitude that shows the peak amplitude of the combined peak/trough pair that encompasses the layer. In the final tuning thickness charts, the blue curve is a cross plot of the apparent time thickness versus the actual time thickness. The apparent time thickness (marked by black dotted curve) tracks the actual time thickness until it reaches a point known as the tuning thickness. Below this point, however, it no longer tracks the actual time thickness; instead, it remains almost constant and slightly smaller than the tuning thickness value. The red curve shows the cross plot of normalized peak-to-trough amplitude against the actual time thickness. Starting at a value of 1 when the layer is relatively thick, it increases in amplitude as the bed thins until the maximum amplitude is reached at tuning thickness. This maximum amplitude is produced by the constructive interference of wavelets generated at the top and base of the studied layer. Below the tuning thickness, the curve decreases almost linearly as the layer continues to thin.

The seismic tuning analysis of the first polygon (part of PL-5400 profile, Fig. 6) revealed a maximum normalized peak-to-trough amplitude of approximately 1.4 at an actual time thickness of approximately 0.017 s. This indicates the tuning thickness point, which is marked by the thin black horizontal line in Fig. 6C. The two-way time obtained through analysis was multiplied by two to determine the period value (T). In this case, the multiplication yields a T value of 0.034 s. This indicates that 29 Hz is the most likely dominant frequency within the examined time window of analysed part of PL-5400 profile, as illustrated in Fig. 6C.

The second polygon, also extracted from PL-5400 profile, illustrates northeastern part of the inferred igneous intrusion imaged by this profile (Fig. 7).

The time window used for wavelet extraction is marked by a black dotted rectangle (Fig. 7A). Calculated amplitude spectrum (Fig. 7B) is slightly broader than for polygon 1. This indicates a slightly augmented frequency content in this segment of the PL-5400 profile in comparison to its southwestern counterpart. The tuning thickness chart (Fig. 7C) shows the results of the seismic tuning analysis that indicate that the maximum normalized peak-to-trough amplitude of approximately 1.3 was achieved at an actual time thickness of 0.0165 s. Thus, the calculated period value was approximately 0.0335 s. In this case, the dominant frequency within the examined time window of the second polygon is close to 30 Hz, slightly higher than in the first polygon.

The third polygon, based on part of the PL-1200 profile, imaged the most geometrically complex segment of the analysed inferred igneous intrusion (Fig. 8).

The wavelet extraction time window was more extensive than in previous polygons in order to capture all branches of strong reflectors associated to igneous intrusions, as shown by the black dotted rectangle in Fig. 8A. The reflection amplitudes are comparable to those depicted on polygon 1 from profile PL-5400, resulting in concise amplitude spectrum (Fig. 8B). A seismic tuning analysis based on the tuning thickness chart (Fig. 8C) demonstrates a maximum normalized peak-to-trough amplitude of approximately 1.4 at an actual time thickness of approximately 0.017 s, which is the tuning thickness point. The period value was found to be 0.034 s at that point, meaning that 29 Hz was the main frequency in the studied time window.

Based on an analysis of all the studied polygons, the average dominant frequency was found to be 29 Hz. Assuming velocity of approximately 6000–6700 ms-1 characteristic for igneous intrusion (see also below), the vertical seismic resolution according to the λ/4 criterion (Widess, 1973) would be approximately 52–58 m. These results have then been used as the basis of the next stage of the vertical seismic resolution analysis based on wedge modelling, in which the empirically derived bandwidth was approximated by a zero-phase 29 Hz Ricker wavelet.

Section 4.1 presents a statistical tuning analysis of the observed seismic data aimed at estimating dominant frequency and vertical resolution using wavelets extracted by a frequency-matching approach. The relationship between apparent peak–trough separation and layer thickness is evaluated using tuning curves derived from the effective wavelets, without applying impedance-based forward modelling at this stage. At this stage, the analysis is purely empirical and based on observations. The physical relationship between tuning behaviour and layer thickness is addressed later using the wedge model (Sect. 4.3). The tuning behaviour illustrated by the wedge model (Sect. 4.3) provides a physical basis for the statistical tuning analysis presented here. The wedge model explicitly demonstrates how peak–trough separation and reflection amplitude vary with decreasing layer thickness.

The aim of 2D seismic forward modelling was to study the seismic response of inferred deep igneous intrusions characterized by three alternative lithologies: granite-granodiorite, basalt and dolerite.

Velocities and densities of the model layers have been based on data from Pasłęk IG-1 and Prabuty IG-1 deep research wells (Fig. 9). They are listed in Table 1.

Two-dimensional seismic-geological model based on the PL-5400 seismic profile was 160 km long and 15 km deep. Three different lithological scenarios for igneous intrusions have been tested, for which petrophysical parameters have been derived from the literature:

granite-granodiorite: velocity = 5950 ms-1; density = 2.67 gcm-3 (Dortman and Magid, 1969; Plewa and Plewa, 1992)

Conclusions regarding the viability of each lithological scenario were based on a comparison of modelled synthetic seismic data with real seismic data. The qualitative assessment of the synthetic seismic data relied on observations of the reflection amplitude resulting from acoustic impedance contrasts between the intrusion and the host rock, i.e. the surrounding Precambrian crystalline basement. As expected, models with higher acoustic impedance contrast produce stronger reflections; however, forward modelling provides quantitative constraints on the range of plausible lithologies consistent with the observed amplitudes. Because the dataset does not represent strictly amplitude-preserved data and reliable attenuation (Q) estimates were unavailable at these depths, the analysis focuses on relative amplitude behaviour rather than absolute amplitude calibration. The modelling is therefore not intended for absolute amplitude calibration, but rather for testing the consistency between the observed seismic response and the petrophysical models.

Figure 10 illustrates modelling results for the granite-granodiorite scenario, characterized by minimum contrast of P-wave velocities and bulk densities between the intrusion and the host rock (Fig. 10A). Due to the very weak impedance contrast, the calculated synthetic seismic response from the modelled igneous body is characterized by very low amplitudes, much lower than those observed in the real seismic data (Fig. 10B). It should be stressed here that the synthetic data are noise-free, whereas the real data contain seismic noise that can partly obscure the reflection amplitudes from the intrusion. Intrusion imaged by real seismic PSTM data (right panel in Fig. 10B), is, despite the noise, characterized by much higher amplitudes than synthetic data where intrusion is almost invisible (left panel in Fig. 10B). The observed difference in amplitude characteristics between the modelled synthetic and real seismic data led us to conclude that granite–granodiorite lithology for the inferred igneous intrusion should be regarded as highly unrealistic.

The results of the seismic modelling used to assess the second lithological scenario of rocks characterized by physical parameters comparable to basalts are presented in Fig. 11. In this case, we have also been dealing with a relatively low P-wave velocity contrast between the intrusion and the host rocks of the Precambrian basement, and the contrast is even lower for bulk density (Table 1). As a result, due to comparable acoustic impedance of the intrusion and the host rock, the intrusion is characterized by a low amplitude seismic response (although not as low as in the former case of granite-granodiorite scenario), different from what could be observed on real seismic data (Fig. 11B). Taking this into account, basaltic lithology has also been deemed as not viable.

The third seismic modelling lithological scenario assumed petrophysical parameters typical for the doleritic intrusion (Brown and Kim, 2020). Figure 12 illustrates the assessment of the dolerite scenario completed using the same methodology as described before. In this case, significant contrast of P-wave velocities and substantial difference in bulk densities exceeding 5 % of the background values, resulted in a high-amplitude seismic image of modelled intrusion (left panel in Fig. 12B), comparable to amplitude characteristics of real seismic data (right panel in Fig. 12B). Taking this into account, and also the fact that numerous wells in the Baltic Basin, including Pasłęk IG-1 deep well located on profiles PL-5400 (Fig. 3) and PL-1200 (Fig. 4), have documented the lower Carboniferous dolerite intrusions, it was concluded that dolerite lithology is the most probable approximation for the inferred igneous intrusion imaged by PolandSPAN^® data within the basement of the Baltic Basin.

Apart from lithology, another important parameter that directly influences seismic imaging of inferred deep igneous intrusions visible on seismic data is the lateral variation of their thicknesses. In order to estimate dominant thickness and its lateral variations, and to analyze the seismic tuning and vertical resolution, a wedge model based on doleritic lithology was used (Fig. 13).

The geophysical parameters of the wedge that simulates igneous intrusion were consistent with the P-wave velocity and bulk density used for 2D seismic modelling described above and were 6700 ms-1 and 2.87 gcm-3, respectively (Fig. 12). The total length of the wedge is 1000 m, and its thickness changes from 0 to 200 m. The wedge model simulates zero-phase seismic data, and the resulting synthetic profile, shown in Fig. 13B, was computed using the full-wave modeling algorithm and Ricker wavelet with a dominant frequency of 29 Hz. Fig. 13C shows a graph that includes: (i) predicted normalized amplitude and (ii) apparent and actual wedge thicknesses, based on Dowdell's (2020) numerical algorithm. We estimated critical values, such as the onset of tuning thickness (λ/2) and tuning thickness (λ/4), in accordance with Widess's (1973) criteria, where λ is the dominant wavelength (Fig. 13C).

For intrusion 200 to 115 m thick, the reflections from the top and base of the wedge could be easily separated (Fig. 13B). As the wedge thins to λ/2 thickness, the tuning curve begins to turn upward as reflections from the top and base start to interfere. Further thinning of the wedge results in increased constructive interference until it is thinned to a thickness of λ/4 (the tuning thickness), at which point the interference reaches its maximum (Fig. 13C). The peak of the tuning curve indicates the minimum thickness of the intrusion that can be resolved seismically, which is approximately 57–58 m (Fig. 13B). Below the tuning thickness (λ/4), destructive interference between the top and base wedge reflections prevails, and this leads to a substantial decrease in amplitude (Fig. 13C). Consequently, the top and base reflections from the intrusion become indistinguishable, precluding more detailed thickness estimation of the igneous body from seismic data.

To verify the applicability of the wedge model and the adopted wavelet parameters to the observed seismic data, we performed a direct comparison between representative real and synthetic traces. A seismic trace was selected from profile PL-5400 (polygon 2; cf. Fig. 7) at a location where the intrusion is well imaged and characterized by strongly tuned reflection amplitude observed at its top (Fig. 14A), interpreted as being close to the estimated maximum tuning thickness. The corresponding synthetic trace was extracted from the wedge model at the same effective thickness (∼58–60 m, see Fig. 13).

Both traces were displayed using identical time windows (100 ms), consistent amplitude scaling, and variable-area wiggle format to ensure full comparability. The comparison (Fig. 14B, C, and D) shows a close correspondence between the observed and modelled waveforms in terms of peak–trough separation, relative amplitudes, and overall wavelet shape, supporting the validity of the adopted wavelet approximation and wedge-based thickness estimates.

Final step in this study was detailed interpretation of inferred igneous intrusions imaged by PolandSPAN^® data that used as a reference point results of seismic modelling studies described in previous chapters. It involved three main steps: (i) detailed analysis of the reflection amplitude on the pre-stack time migrated (PreSTM) seismic data, (ii) assessment of the intrusion thickness based on seismic tuning and the vertical resolution criteria provided by the wedge model analysis, (iii) interpretation of intrusion top and base and analysis of its overall geometry.

Analyzed igneous intrusion are characterized by high amplitude seismic reflectors clearly visible within the Precambrian crystalline basement, which is typical for igneous intrusions imaged by seismic data (cf. Infante-Paez and Marfurt, 2018). The observed seismic reflections indicate that the analyzed intrusion is characterized by the following features: (i) partly lobate morphology, (ii) multiple, step-wise arrangements, and (iii) lateral thickness changes. This resembles lower crustal reflector system in the northern North Sea (Wrona et al., 2019), and seismic images of igneous intrusions from the North Sea basin (Hansen and Cartwright, 2006), from the offshore Norway (Cartwright and Hansen, 2006), and from the eastern Scandinavian Caledonides (Juhlin et al., 2016). Certain geometrical characteristics such as sill stepping etc. could be also compared to field analogues from other regions (e.g., Schofield et al., 2010; Stephens et al., 2017; Galland et al., 2019; Hoyer and Hastie, 2022; Lombardo et al., 2025; Williams et al., 2025; Galland et al., 2026; cf. Arachchige et al., 2022).

As shown in Figs. 15–17, the top of the intrusion is related to a positive amplitude reflection (peak), which is indicative of an increase in acoustic impedance. The top reflection demonstrates a moderate- to high-amplitude response, accompanied by periodic segments of amplification and dimming on variable length scales along the intrusion. These segments are indicative of lateral amplitude variations associated with thickness changes. Amplitude response of intrusion top attains its maximum value at the tuning thickness. This peak is estimated when the intrusion thickness reaches approximately 58 m, in accordance with the λ/4 criteria established by Widess (1973). As a result, reduced amplitude reflectors are expected when intrusion is thinner or thicker than this value.

The seismic image of the intrusion displays a close-to-tuned response over a substantial portion of the intrusion length (Fig. 15). The base of the intrusion is associated with negative amplitude, usually amplified but in some cases also dimmed, similarly to what could be observed for the top intrusion reflector. In areas where the inclination of the intrusion could be neglected, its base is often amplified, forming a specific amplified pair with the top reflector. Within the most of the analyzed area, it is possible to distinguish top and base of the intrusion, usually with some internal reflectivity between them. This suggests that the thickness of the intrusion is usually higher than the tuning thickness (λ/4). A reduction in reflection amplitude is often observed in regions where intrusion are thicker, so the amplitude changes as the distance between the top and base reflections increases (Fig. 15A).

The lateral margins of the intrusion can be identified by an abrupt cut-off of seismic amplitudes. In certain regions, an abrupt decrease of reflection amplitude may be indicative of intrusion lateral termination. In other cases, the zone of reduced amplitude corresponds to its tapering thickness. However, the precise position of the tip cannot be determined more accurately because its exact shape cannot be resolved seismically (cf. Cartwright et al., 2025). Another salient feature of the intrusion's seismic image is diminishing of amplitude accompanied by a polarity reversal (Fig. 15A). This is frequently observed in regions exhibiting substantial lateral variations in intrusion thickness, as illustrated in Fig. 15B. Recently, Cartwright et al. (2025) described similar seismic features characteristics for the Dogger Sill Complex in the Southern North Sea.

Figure 15 presents results of interpretation of the southwestern part of the intrusion imaged by the PL-5400 profile. The intrusion thickness estimation was based on the analysis of the time seismic section (PreSTM; Fig. 15A) and it was verified by quantitative observations based on the wedge model (Fig. 13). Because the amplitude response is directly related to intrusion thickness, the reflection amplitudes from the top and base of the intrusion could be analyzed using seismic tuning parameters and thickness values obtained from wedge model simulations. We then compared these results to the depth seismic section (PreSDM) and interpreted the top and base intrusion (Fig. 15B).

As shown in Fig. 15A, the amplitude response indicates that interpreted intrusion thickness typically exceeds the maximum tuning amplitude (λ/4) and frequently approaches or surpasses the tuning onset (λ/2). The majority of amplified amplitudes are observed in the central part of the intrusion. This suggests that the intrusion thickness is significantly reduced in this area. According to the wedge model, the thickness ranges from 60 to 120 m, which corresponds to wavelength values between λ/4 and λ/2. In other segments, thickness of the intrusion is relatively larger, as evidenced by the partially separated reflectors from top and base of the intrusion. Reflection amplitude is only partially amplified, and internal (interfered) reflections are observed between the top and base horizons (the red and blue horizon picks in Fig. 15B). In these areas, the estimated intrusion thickness generally exceeds 120 m, locally exhibiting thicknesses of up to 200 meters or even more (Fig. 15B).

The seismic image of the entire intrusion is complex, with several amplitude cut-offs that are indicative of its lateral termination or tapering. Additionally, polarity reversals, another indicator of thickness changes, are clearly visible, highlighting areas of significant lateral thinning or thickening of the intrusion (Fig. 15A and B). The amplitude cut-off observed in the central part of the seismic profile in Fig. 15 indicates the presence of potential steps in the intrusion geometry, which likely manifest as abruptly diminished amplitude and sharply reduced thickness. These irregular seismic reflections resemble the characteristic “step-wise” geometry (cf. Kavanagh and Sparks, 2011; Muirhead et al., 2012; Burgess et al., 2017). However, due to the large depth ranges and the lack of well control, the origin of the observed geometries remains unclear and requires further study using various intrusion analogues.

A detailed interpretation of the northeastern part of the intrusion imaged by profile PL-5400 is presented in Fig. 16. It reveals that intrusion lateral thickness variations are significantly more pronounced than those depicted in Fig. 15. In many areas, the top and base of the intrusion can be clearly separated, with no evidence of amplitude amplification resulting from the seismic tuning effect which means that thickness exceeds the tuning onset (λ/2). In these cases, the estimated thickness of the intrusion significantly exceeds 120 m, reaching up to 200 m (Fig. 16A).

A polarity reversal indicates significant lateral thinning or thickening. Tapering of the intrusion, local lateral terminations, and gradual amplitude fading are clearly visible as an amplitude cut-offs in the seismic image. A significant portion of the intrusion displays amplified or dimmed reflection amplitude, indicating that its thickness may fall within the range of λ/4 to λ/2, and is considerably influenced by tuning (Fig. 16A). In certain areas, the intrusion thickness approaches or slightly exceeds the maximum tuning thickness value (λ/4), as evidenced by the strongly tuned amplitudes. In these segments, the intrusion thickness can be reduced to approximately 60 m (see Fig. 16A and B). On the other hand, the interpretation of the intrusions is locally unclear because instead of the expected amplitude cut-off associated with intrusion tapering, at some points it appears to represent clearly separated top and base intrusion reflections with medium amplitudes. This suggests the existence of an even intrusion body of higher thickness exceeding 200 m. Classic “step-wise” geometries are also visible in the seismic data (Fig. 16B).

Profile PL-1200 imaged most complex part of the studied igneous intrusion (Fig. 17). The thickness of the intrusion varies considerably laterally, with multiple polarity reversal points and discernible amplitude cut-offs associated with its tapering or lateral termination (Fig. 17A). Dimming or local amplitude amplification, particularly when approaching a value of λ/4, is indicative of significant variations in sill thickness. In some areas, the sill thickness varies between 60 and 120 m.

The top sill reflection is influenced by seismic tuning in broader areas, as indicated by the wedge model, where the sill thickness aligns with the λ/4–λ/2 criteria (Fig. 17A). On the other hand, a significant portion of the intrusion exhibits a reflection amplitude that far exceeds the tuning onset (λ/2), with thicknesses ranging from 120 to over 200 m (Fig. 17B). This is especially evident in the central part of the sill. Intrusion features here multiple internal “steps” and small branches from the main intrusive body. The interpretation of some of these features remains ambiguous, although the main sill branching could be clearly observed (Fig. 17B).

Despite the fact that thickness variations provide the most straightforward explanation for the observed lateral waveform changes, alternative factors – including compositional heterogeneity of the intrusion, internal layering, or variations in the surrounding host rock – may also contribute to the seismic response and therefore cannot be excluded, and should be further investigated in future integrated modelling and geophysical studies.