Where is the iapetus suture

All first-order features listed above are robust in the 2-D models, in that they are found with varying subsets of the data and varying inversion parameters and varying starting models.

We undertook 3-D forward modelling to verify the robustness of the main features obtained using 2-D inversion. In comparison with 2-D, 3-D forward modelling is time-consuming because of the fine grid requirements for resolving conductivity contrasts.

To achieve convergence we selected a cell size of m in the vertical Z direction closest to the surface and increased it progressively with depth. The depth sections from the 2-D models obtained along the five profiles were used for building the initial forward 3-D conductivity model.

Forward solutions were derived in seven decade period bands, six discrete periods per decade, and were compared to original un-decomposed MT impedance responses. After trial-and-error forward models, the synthetic responses reproduce the main features of the observed data at all 39 sites; Figs 10 a and b illustrate this fit at six representative sites, one site from each profile. The data fits to the XY component curves and the YX phase curves are generally good, whereas the YX apparent resistivity curves exhibit a similar shape with lower magnitude Fig.

The regional features obtained with the 3-D forward model Fig. The final model is illustrated in Fig. To assess the validity and robustness of the 3-D forward model we performed two different types of 3-D inversion: a constrained 3-D inversion based on the final 3-D forward model shown in Fig.

The constrained inversion run using the 3-D forward model as the starting and prior model resulted in only some minor adjustments in the subsurface conductivity structure. Since the constrained inversion results cannot be regarded as an objective and unbiased assessment of the 3-D forward model, we therefore focus on the inversion study starting from homogeneous half space models.

However, as we will show, this study demonstrated that, to a large degree, the 3-D forward model represents an acceptable explanation of the observed data. In preparation for 3-D inversion, we synthesized the MT data from different instruments and processing methods to obtain an MT data set with 46 consistent periods.

The horizontal mesh was chosen to be consistent with the 3-D forward study; however tests with different discretization did not change the inversion results significantly. Since the regularization of the ModEM inversion scheme penalizes deviations from a prior model as part of the model update search, we needed at least a two-step inversion run to make sure the inversion model develops farther away from an homogeneous half-space.

The final model was re-started three times using the model of the previous inversion run as the starting model. Typically these runs complete because a moderate decrease in misfit is accompanied by a significant increase in model complexity. When restarting the inversion runs, significant improvements in data fit were again possible. The inversion was performed with isotropic model covariances of 0.

The data fit of the starting model half-space with ocean yielded an rms of 48, which was reduced within the course of the inversion procedure to a value of 1. In addition, we applied some resolution tests for certain conductivity features present and found to be robust in the 2-D inversion results that did not show up in the 3-D inversion.

This improved the data fit at stations — and 25 resulting in a final rms of 1. We will discuss these tests in more detail when comparing features below. Horizontal slices showing the electrical conductivity distribution of the 3-D inversion model at the same depth as in Fig. In the following we will compare the modelling and inversion results of three different approaches: 2-D inversion, the 3-D forward model derived from the 2-D models, and the 3-D inversion.

While it is rather well accepted that generally 2-D inversions can only be regarded as a first order approximation of the actual, usually more complex, subsurface conductivity structure, they are still superior in terms of possible discretization of the subsurface e.

Siripunvaraporn et al. Due to the much denser meshes feasible for 2-D inversion, these models typically result in more detailed conductivity images. Also, computational cost of 3-D inversions is prohibitively high for conducting hypothesis and resolution tests to the same extent possible in 2-D inversion.

It is beyond the scope of this paper to discuss the trade-off between structural details and artefacts introduced by the simplifying and not entirely satisfied 2-D assumptions, but pertinent to the work is that 2-D inversion models can be biased as off-profile features might be mapped as artificial conductivity features beneath the profiles e. Ledo et al. This implies that the mislocated features might have made their way also into the 3-D forward model, as it is, to a large extent, based on the 2-D results.

Therefore, the 3-D forward model cannot be regarded as completely independent of the 2-D results, but nevertheless it is able to account for the regional conductivity contrasts in Ireland, such as the surrounding oceans and that the modelled conductivity anomalies are not 2-D in a strict sense as changes can be observed between profiles.

Otherwise, 3-D inversion does not consequently result in superior, that is more realistic and veracious, conductivity models. Only the introduction of a priori information or rotation of the underlying coordinate system led to correct recovery of a pronounced regional 2-D structure.

Against this background, we show depth slices of the 3-D inversion result Fig. In general, areas with high or low conductivities correlate in a relative sense. This could explain why resistors are less resistive and conductors less conductive in the 3-D inversion result, or simply as a consequence of the Tikhonov smoothing regularization trade-off parameter employed in the inversion objective function.

At lower crustal depth levels both the forward Fig. Whereas in the 3-D forward model only higher resistivities were included in the southern part of the grid, the 3-D inversion model additionally infers more resistive structures in the northern part of Ireland outside the area of site coverage.

One of the major differences between 2-D and 3-D inversions is the parameters to be fitted: whereas 2-D inversions usually fit apparent resistivities and phases, 3-D inversions focus on the individual complex-numbered impedance tensor elements. Since they become very small in the case of conductive features, it can easily occur that such a structure, although the MT method is more sensitive to conductors, is missed. This might also be the reason why one of the pronounced deep conductivity anomalies UMC was initially not present in the initial 3-D inversion results.

Only after including this structure as a priori information, the restarted inversion run kept this feature by reducing the rms at the same time. Detailed inspection of stations — and 25 confirmed that the data fit to the long period data was improved with its presence. A detailed overview on the obtained data fit between the 3-D inversion model Fig.

The rms values are calculated for various subsets of data components and period ranges. The upper row of Fig. In the middle row, we present the rms values of both diagonal centre left and the individual off-diagonal impedances centre middle and centre right for the entire period range.

The lower row gives accumulated total rms values impedances and Hz for different period ranges. Only a few stations appear to have a significantly worse data fit than the average. In general, the total rms, based on the data fit for all individual impedances and vertical magnetic transfer functions, is below 2; there are a few sites with slightly higher yellow colours values.

The breakdown of total rms into its contributions from different impedance tensor components Fig. Maps of rms values for the 3-D inversion model Fig. The upper panel shows the total rms for the full impedance tensor and the vertical magnetic transfer functions left and separately for impedances middle and magnetic transfer functions right accumulated for all periods.

In the middle panel we present the rms values of both diagonal left and the individual off-diagonal impedances middle and right for the entire period range. The lower panel gives accumulated total rms values impedances and Hz for different period ranges. In general, larger areas for which the MT data is not explained by the 3-D inversion model are not obvious. Most of the prominent and labelled conductivity anomalies from the 2-D inversions see Fig. Profile I Fig. Conductivity anomaly C7 occurs in both the 2-D and 3-D inversions, although the higher conductivities dipping beneath R7 and R8 are not as similarly pronounced in the 3-D model as in the 2-D result.

Beneath site 02 both inversions sense a very shallow small scale high conductivity anomaly but which exhibits higher conductivities in the 3-D case. The detailed discussion and comparison to the 2-D sections of Fig. The surface expression of tectonic structures is indicated by arrows. Gray UK English spelling? In return, the 2-D inversion implies deep-reaching conductivities in this area.

Interestingly though, the NTL seems to define a southern limit to the shallow resistive block R5, which would be in contrast to profile I but more consistent with profiles II—IV. This is in contrast to the 2-D inversion model Fig. However, the general behaviour of deep structures conductor and resistor below is comparable again in the 3-D model, albeit with depth range differences. Based on the rms values see Fig. When considering the geological map Fig.

The shallow conductive zone C14 close to site 11 in the 3-D inversion seems to be connected north-eastward with C3 on profile IV. In the 2-D inversion model we can identify a similar feature further south. Given the distance between the stations along this profile, such shifts in position are not surprising though. Several zones of high conductivities C2-C4 in the upper crust appear to have their equivalents also in both inversions; however, C12, which does not occur in the 2-D inversion, is present in 3-D and shows a connection to C8 and C9 of profile V.

An additional conductor C11 is exhibited by the 3-D inversion that starts north of NTL and has a weaker expression up to the northern end of the profile and thus connects again to profile V. Profile V is by far the one with the most structure in 2-D.

The 2-D result Fig. The sediments that lie in the Midland Valley terrane are predominantly Carboniferous limestones and Devonian sandstones in the Carrick syncline stations — Philcox et al. Support for such a geological setting is provided by the COOLE seismic refraction profile that reveals a layer of P -wave seismic velocity 5.

Pertaining to profile V, the highly resistive layer feature R1; Figs 8 and 14 was interpreted previously by Rao et al. The feature is not sensed on the other profiles due to their limited northern extent. The two resistive blocks features R2; Figs 8 and 14 in the centre of the profile were interpreted by Rao et al. These two resistive blocks are also seen below the northernmost stations on profile IV in the 2-D inversion model, whereas both the forward and inversion 3-D models exhibit generally resistive shallow subsurfaces but with less pronounced anomalies.

It is difficult to discriminate electrically between sedimentary and volcanic rocks beneath this locality due to the high resistivity of the sedimentary Carboniferous limestones in this region. Sevastopulo inferred that these limestones had initially a high porosity, but that porosity was rapidly reduced with the deposition of micritic internal sediments and precipitation of marine fibrous cement.

The high resistivities may indicate low porosity or lack of fluid connectivity that may be consistent with various rock types in the Leinster terrane. One such possibility is Duncannon Group's thick sequences of intermediate to acid volcanic rocks, which crop out in the south of the Leinster terrane.

However, these are restricted to the shallow surface. The Inch Conglomerate may also support the interpretation of resistors R6 and R8. The Leinster granite is also known for its high resistivities in this part of the study region. The resistor R5 on Central terrane may be due to Silurian ferromagnesian rich greywackes.

As shown in Figs 8 and 14 , the high conductivity zone becomes shallower as one traverses from east to west across Ireland. From their MT study in south-western Scotland, Tauber et al.

The southernmost stations on profiles I and II show a subvertical highly resistive layer features R6 and R8.

This region lies on the Variscan orogenic belt that extends from Ireland into Great Britain and further eastwards Matte Similar resistive structures were also inferred in prior MT studies Brown et al. Sites 10 and 04 are located in the vicinity of a mapped fault zone, and Landes et al. The first-order and robust feature in the 2-D depth sections Fig. Additional geophysical data sensitive to other petrological-geophysical parameters also display this strong east—west asymmetry. Seismic refraction studies Jacob et al.

Those authors also noted the absence of the low velocity zone LVZ in the western part. Hauser et al. The silica richness in western Ireland is also corroborated by faster P -waves negative traveltime anomaly from the tomographic study by Wawerzinek et al.

This strongly supports our observation that the crust of western Ireland is highly resistive. The high conductivity in eastern Ireland may be due to the shale and sulphidic material within the sedimentary accretionary wedge at the Iapetus subduction zone.

As the Iapetus Ocean closed and sutured, in the east there was deposition of conducting sediments and subsequent subduction deep into the crust, whereas in the west either that material was not available for subduction, or it was not subducted deeply into the crust and has subsequently been eroded. In either circumstance, the tectonic processes varied along strike of the orogeny during Iapetus Ocean closure and continental suturing.

Finally, the recent surface wave anisotropy study of Polat et al. Intriguingly, the shaded Bouguer anomaly gravity map of Ireland Readman et al. High conductivity at crustal depths can be caused by many factors, such as saline water, partial melt, sulphides, iron oxides and graphite Frost et al.

It seems most likely that the high conductivity we observe is due to the occurrence of low-grade metamorphosed graphitic sediments in Palaeozoic suture zones. The conjugate fault system comprises strike-slip faults trending either generally north-west with a dextral sense of displacement, or generally east-north-east with sinistral displacement.

Though individually minor, these faults were to have a profound structural influence during subsequent episodes of extensional tectonism when their reactivation controlled Late Palaeozoic basin development and geometry. More immediately, in the transtensional tectonic regime pertaining during latest Silurian to Early Devonian times, strike-slip basins opened across the region and were filled with the clastic, terrestrial sediments of the Old Red Sandstone lithofacies.

The transtensional regime may also have been an important factor in the intrusion of the Early Devonian granite plutons. Terrane evolution of the paratectonic Caledonides of northern Britain.

Journal of the Geological Society of London, Vol. Clarkson, E, and Upton, B. Death of an Ocean — a Geological Borders Ballad. Edinburgh: Dunedin Academic Press. Parallel geological development in the Dunnage Zone of Newfoundland and the Lower Palaeozoic terranes of southern Scotland: an assessment. Atlas of palaeogeography and lithofacies. Geological Society of London Memoir, No.

Kelling, G. Southern Uplands geology: an historical perspective. Stone, P, and Merriman, R J. Abstract The Iapetus suture in Ireland and Britain is that line which separates Caledonian rocks of the Laurentian and Avalonian continents.

This content is PDF only. Please click on the PDF icon to access. First Page Preview. Close Modal. You do not currently have access to this article. You could not be signed in. Librarian Administrator Sign In. Buy This Article. Email alerts Article activity alert. Early publications alert.

New issue content alert. View Full GeoRef Record. Citing articles via Web of Science Archive Current Issue Early Publication. This site uses cookies.