Comment on: “Interseismic Strain Accumulation near Lisbon (Portugal) from Space Geodesy” by Fonseca et al. (2021)

The paper by Fonseca et al. (2021), hereafter referred as FON21, published in Geophysical Research Letters2 make several conclusions that are not convincingly supported by the evidence of the data that is made available. In this comment we will address the following statements: 1) FON21 “provides new evidence of sinistral simple shear driven by a NNE-SSW first-order tectonic lineament; 2) “PSInSAR vertical velocities corroborate qualitatively the GNSS strain-rate field, showing uplift/subsidence where the GNSS data indicate contraction/extension”; 3) FON21 proposes “the presence of a small block to the W of Lisbon moving independently toward the SW with a relative velocity of 0.96 ± 0.20 mm/yr”; 4) FON21 shows “that the contribution of intraplate faults to the seismic hazard in the LMA is more important than currently assumed”. We conclude that more evidence needs to be collected to confirm or infirm FON21 statements and conclusions. For the moment the proposal of an autonomous crustal block moving with significant velocity in relation to the neighboring domain should be considered speculative and unproved.

differences on the choice of the best ground motion prediction equations let to the publication of PSHA results that differ strongly (e.g. Vilanova and Fonseca, 2007, Campos Costa et al., 2008, Mezcua et al., 2011, Woessner et al., 2015. These discrepancies may lead to confusion among the general public and policy makers and may lead to difficulties in the acceptance of building codes that are established from one of the studies.
In the following sections we will concentrate our analysis on the restricted area defined in FON21 figures 1a and 2a .

The rigid block identified by FON21
Investigating the GNSS derived velocities with respect to Eurasia, FON21 identified in figure 2a three locations, MAFR, PACO and IGP0, that moved with a velocity of 0.96 ± 0.20 mm/yr relative to the other 8 locations in the area. To reach this value FON21 discarded the CASC location (close to PACO) arguing that CASC is "pinned by the Sintra batholith that lays to the North. Later, FON21 present in figure 4b the tectonic model that could explain the block kinematics.
We begin the discussion by presenting in Fig. 1 the kinematic data relative to the FON21 reference block.  The southern and eastern rigid block limits are not provided in FON21 but they can be inferred from the GNSS sites. The southern limit lies probably along the E-W Tagus river or south of it. The eastern boundary must lie somewhere between the MAFR and ARRD sites. None of these presumed boundaries is explained by the FON21 tectonic model. If indeed the LTV faults represent a major tectonic feature, then site ARRD should move differently from the other reference sites and should move with a velocity close to the rigid block sites. The interpretation of the ARRD velocity in the light of the tectonic model proposed is not given in FON21.
If CASC is pinned to the Sintra batholith and it moves with the same velocity as the other reference sites, it means that CASC is probably moving with the basement velocity. Then the meaning and relevance of the velocities derived for MAFR, PACO and IGP0 may be questioned. Do they represent the velocity of a rigid block with crustal thickness or do they show a shallow feature that might be detached from the basement, as questioned by FON21? In this area the crustal thickness is ~26 ± 2 km (Dündar et al., 2016), approximately the length between the three sites that define the FON21 rigid block. As mentioned by FON21 the rigid block may extend further West, but no evidence is provided for that. If the rigid block proposed has crustal expression, then the aspect ratio is close to 1 x 1 x 1, which may be considered small and difficult to assess with sparse data.
We may also question the interpretation of CASC velocity in the light of the seismicity observed in the area and presented in FON21 figure 1b. If the Sintra batholith is resisting the block movement then a higher seismicity rate should be expected surrounding it, which is not observed.
The rigid block velocity FON21 associated three GNSS sites to the rigid block and discarded one site, very close to this set. Since we disputed in the previous section the reasons for discarding CASC, the procedure used by FON21 might be considered cherry picking. In this section we consider all possible sets of 3 close GNSS sites, without cherry picking, and compare their average velocity with the average velocity of the other 8 sites that are considered as reference.
The results are presented in Table 1 and compared the rigid block proposed in FON21.
We may conclude that, from the 9 possible combination of 3 close GNSS sites without cherry picking, 5 combinations show a significant velocity relative to the other 8 sites considered as reference, higher than 0.5 mm/yr. Using the same reasoning of FON21 one could argue that FVFI, VNOV and GRIB could define a rigid block with a relative velocity of 0.84 ± 0.07 mm/yr. Our conclusion is that the definition of the rigid block made by FON21 is somewhat arbitrary, guided by model considerations and not driven by the data. We may also infer that, if there is one rigid block in the LTV area, then its relative motion must stand out from the more or less random choice of sites. Then, the relative velocity attributed by FON21 for the proposed rigid block is clearly over-estimated. Testing the LTV alignment with GNSS We may expand the discussion provided in the previous section by assessing the relevance of the LTV alignment to explain the observed GNSS velocities. For this we consider the 11l GNSS sites separated into Western and Eastern blocks, as shown in Fig.  2. The Western side of the LTV alignment moves relative to the Eastern side with an average velocity of 0.60 ± 0.04 and an azimuth 198 ± 7. This velocity is smaller than the relative velocity of the rigid block defined by FON21 and may be considered only marginally relevant in the light of the considerations made in the previous section.
Looking into the GNSS velocities, we may test if a fictitious NW-SE tectonic alignment may also explain the observations (Fig. 3). Fig. 1.

Fig. 3 -GNSS velocities relative to the rigid block formed by the stations in blue, north of a fictitious alignment (dashed red line). The red arrows show the velocities of the sites south of the alignment. The thick red arrow shows the average velocity between the red and blue sites. Other features as in
In this fictitious example, the sites SW of the alignment move relative to the NE side with an average velocity of 0.79 ± 0.07 and an azimuth 212 ± 6. This is a more relevant velocity than the one obtained with the LTV alignment, without cherry picking. Our conclusion is that the available GNSS data is not adequate to discuss tectonic models that are elaborated to explain the seismicity and deformation in the LTV region.

PSInSAR vertical velocities and the GNSS strain-rate field
To facilitate the discussion of the GNSS strain-rate field we show in Fig. 4 a restricted version of FON21 figure with the same geographical limits as the ones given in previous figures of this comment. We also add some relevant features, the location of the GNSS sites identifying those that belong to the rigid block defined in FON21 and we add also the tectonic model used by FON21 for the interpretation.

Fig. 4 -Strain-rates inferred from the GNSS horizontal velocities, modified from FON21. Blue corresponds to areas under compression, pink to areas under extension. Green circles show the GNSS sites used by FON21 to define a rigid block. The other GNSS sites are shown as white circles. Thick lines show the active tectonics model used by FON21 for the interpretation of the GNSS velocities
As mentioned in FON21 the whole domain where reliable strain-rates could be estimated is under extension, except for a small domain surrounding the Lisbon-Tagus bar area, North and South of the Tagus river. We may add that the LTV alignment or the tectonic model presented in FON21 do not show up on the strain-rate map. We remark also that the area where the rigid block proposed by FON21 is proposed in the area where the largest variation of strain-rate is found. Considering the 3 GNSS sites used by FON21, the strain regime changes N-S from extension to compression and the maximum compression axis rotates from NNW-SSE to NNE-SSW in the North-South direction. These observations do not favor the proposal of FON21 that a rigid block exists west of the LTV alignment.
As regards the PSInSAR vertical velocities we found the information presented in FON21 figure 3a difficult to read. Using the data provided we preferred to compute a representative grid with the average vertical velocity on a cell size ~1.1 x 0.8 km 2 (0.001º x 0.001º). We plot this grid on Fig. 5 with the same geographical limits used previously to facilitate comparing information. Comparing the strain compression and extension domains defined by GNSS (Fig. 4) with the vertical velocities in Fig. 5

Discussion on seismic hazard
The estimated relative velocity of 0.96 ± 0.20 mm/yr of a crustal block made by FON21, when converted to seismic moment release, is compared to the seismic moment release proposed by Woessner et al. (2015) for the areal source area that includes the LTV. These authors considered for PSHA a logic tree with 3 possible maximum magnitudes for this source area of 7.1, 7.4 and 7.6. Using 15 km as the seismogenic thickness, FON21 concludes that the paper estimates agrees well with the recurrence model with 7.4 as the maximum magnitude, suggesting also a strong seismic coupling for the area.
We have shown on the previous sections that the definition of a rigid seismic block and its relative velocity is largely uncertain and most probably over-estimated. The maximum magnitude used by Woessner et al. (2015) on the earthquake recurrence models exceeds the maximum historical event of 1531 by one degree which also suggests that the moment magnitude release is over-estimated.