1. Geological Setting
1.1. Alpine Structural Evolution
The alpine structural evolution of the Hellenides began during the Early-Middle Mesozoic. Discontinuous westward and eastward orogenic migration, characterized by intense collisional tectonics, related to successive subduction processes of the Tethyan oceanic basins, produced a complete stack of several nappes of Jurassic to Tertiary age. Extensional tectonic took place in between the nappes stacking processes causing tectonic denudation and crustal exhumation 
. Various palaeogeographic models have been proposed arguing about the tectonic evolution of Tethys and the emplacement of the ophiolites, most of which are extensively discussed by Robertson et al. 
. Towards the final stages of the alpine cycle, the region is marked by the convergence of the Eurasian and Nubian (African) plates, passing from the Mesozoic to the Tertiary, with the consequent closure of the intermediate oceans (Tethys Ocean) and the consequential back-arc extensional regime that led to the orogen collapse through large, normal, low-angle detachment faults often causing the exhumation of underlying units as metamorphic core complexes 
The main feature of the Middle-West Greek orogen is the successive tectonic nappes and slices, which are thrust on top of each other from east to west. The thrusting events started in the Middle-Late Jurassic with the obduction of the ophiolites onto the Pelagonian carbonate platform from one or more Tethys oceans 
. Paleogeographically, the Pelagonian zone is considered to be a large continental shelf, part of the Cimmerian continent, which was detached from Gondwana (i.e., Africa with other smaller continents; 
). During the Tertiary, the tectonic evolution continues with the thrust orientation toward the west-southwest and the thickening of the continental crust. The accumulation of successive nappes and slices, in combination with the development of subduction zones, is associated with the creation of two sub-parallel metamorphic zones of high pressure/low temperature (HP/LT) of Paleocene-Eocene and Oligocene-Miocene age. The extensional regime that followed the compressive tectonics led to the syn- and post-orogenic collapse with large, low angle, normal detachment faults, and the exhumation of the underlying zones in the form of tectonic windows in Olympos-Ossa, Cyclades, and Peloponnese-Crete areas 
. Extensional tectonics has played a crucial role from the Tertiary to the present day with extensive shear zones and detachment faults. The reactivation of the older reverse faults (inherited structures) as extensional structures (inverse tectonics) set a study case to be revised in correlation with the recent earthquakes.
1.2. Alpine Tectonostratigraphy
This area lies on the Pelagonian zone (or nappe) which consists of a Palaeozoic or older crystalline basement (gneiss, gneiss-schists and amphibolites) intruded by Upper Palaeozoic granitoids (Carboniferous age). It is overlain by a low metamorphic Permo-Triassic volcano-sedimentary series, composed by intercalations of marble layers, phyllites, sandstones and bimodal volcanic products. Triassic-Jurassic crystallised limestones and dolomites follow at the top of the Pelagonian stratigraphic column. Characteristic is the occurrence of the tectonically emplaced, initially during the Upper Jurassic, Neo-Tethyan ophiolite masses on the Pelagonian Triassic-Jurassic carbonate platform and secondary imbricated during the Cretaceous and Tertiary with the Pelagonian formations. Between the underlying Olympos-Ossa Triassic to Eocene-Oligocene platform carbonate formation and the continental block of the Pelagonian, High-pressure rocks (Blueschists of the Ampelakia Unit) are tectonically emplaced. They were metamorphosed in high pressure conditions during the Paleocene-Eocene, when they subducted under the Pelagonian block. Thus, after the Eocene, the Pelagonian formations were thrust upon the carbonate series probably of the External Hellenides, which were revealed afterwards as a tectonic window 
The study area is dominated by the Palaeozoic basement rocks and their overlain Triassic-Jurassic neritic carbonate nappe, which are tectonically emplaced on a continuous carbonate series (crystalline limestones and marbles) of Triassic-Jurassic and possibly Cretaceous age, also known as the carbonate series of Kranea, which is considered relatively autochthonous exhumed as the tectonic window of Kranea, equivalent to the Olympos-Ossa tectonic window more eastern of the study area 
1.3. Alpine Tectonics
The study area is characterised by a complex alpine tectonic deformation as a result of many tectonic events that have affected the described geological formations of the area 
A predominant S1 schistosity is recognized in the broader area. It is SW-ward dip-directed at the west-southwestern part of the Kranea tectonic window (Study area) and NE-ward dip-directed at the north-northeastern part of the window. On the S1 planes, a L1 lineation is recognized in a consistent NE-SW to NNE-SSW orientation but NE-ward plunging at the north-northeastern part of the Kranea window and SW-ward plunging at its south-southwestern part. Both S1 and L1 are due to a deformation event (D1) associated with the SW-ward thrust of the Pelagonian nappe upon the Kranea carbonate series, as it is shown by shear criteria (e.g., S-C fabric, asymmetric boudins, mica fish, σ and δ clasts etc.). Isoclinal folds and sheath-folds also developed during D1, sup-parallel to parallel to the L1 stretching lineation after rotation and hence parallel to direction of the maximum extension (X-axis). During the last stages of the D1 deformation, extensional SB mylonitic shear zones were created and in the upper tectonic horizons low-angle normal faults, both dipping to the SW and NE. The presence of glaucophane (HP/LT conditions) on the SB planes related to a DSB deformational stage indicates that it initially developed at great depth, as the glaucophane is preserved rotated along the SB or S1 planes and parallel to the L1 lineation. The evolution of the D1 deformation in the extensional DSB stage resulted in the creation of extensive shear zones while the orogen emerges with the simultaneous tectonic escape of the accumulated nappes. The D1 is associated with a greenschist metamorphic facies (pressure up to 6–7 Kb and temperature around 400–480 °C). On the other hand, the HP/LT metamorphism and the development of the glaucophane characterize the period of the subduction of the geological formations below the Pelagonian continental block during the Paleocene-Eocene. An older, residual synmetamorphic deformation is also detected on the Pelagonian series, but strongly reworked by the already described Tertiary deformational events (compression vs. extension). It is possibly of Late Jurassic-Early Cretaceous age related to the first Alpine structural evolutionary stages of the Hellenides. Older Palaeozoic structures cannot be recognized in the studied area because they are totally disappeared by the younger alpine tectonics.
A compressive younger D2 deformational event followed, possibly during the Early Miocene, at very low P/T conditions, characterized by knick-folds or zones, shear zones and thrust faults. At this stage, the orogen has already emerged. The thrusting of the Pelagonian nappe and the HP/LT rocks of Paleocene-Eocene age on the underlain Kranea carbonate formation is assumed to have taken place during the Eocene-Oligocene. From the Late Oligocene to the Early Miocene, the orogen continued its emergence, while at the same time it collapsed (extensional event DSB). The following Neogene-Quaternary extensional tectonics, which is directly related to the alpine one, as will be discussed below, is described in the next section.
2. Post-Alpine Geology
The large Neogene basins in the central and northern Greek mainland are formed by the post-alpine extension of the Late Miocene that followed the final ENE-WSW-trending alpine compressional stage in the Early-Middle Miocene (Aquitanian-Langhian; e.g., 
), which is still active further to the west, in Epirus and the Ionian Sea 
. The post-orogenic collapse phase, following the convergence front of the Eurasian and African plates, led to the first NE-SW oriented extension. Given that this phase is responsible for the initial formation of most of the large basins, the major trend of the basins has a NW-SE direction, shaped by NW-SE-striking normal faults.
In a study area, three major basins are met, the intermontane Elassona-Domeniko-Amouri basin, and the much larger Eastern (also known as the Larissa plain) and Western (also known as the Karditsa plain) Thessalian basins (central mainland Greece). The former is the northwestern one and can be subdivided into two smaller sub-basins, the Elassona to the NE and the Domeniko-Amouri to the SW. These two sub-basins are connected through two narrow valleys. The Larissa plain is located to the east and is separated from the previous basin by a mountainous, marble-consisting mass; there is only a narrow passage that connects the Larissa plain with the Domeniko-Amouri sub-basin, i.e., the Titarissios River, which flows toward the Larissa plain. The West Thessalian basin is located south of the Domeniko-Amouri sub-basin and west of the Larissa plain. It is completely isolated from the former by the Zarkos mountains and connected with the latter via the Penios River, flowing toward the Larissa plain. All basins are separated by the alpine rocks of the Pelagonian zone, which are either schists and/or marbles.
Based on the oldest basinal deposits, all the aforementioned basins started to form in the Late Miocene, right after the last Early-Middle Miocene alpine compressional stress field.
2.1. Post-Alpine Lithostratigraphy
The neotectonic activity, along with the local palaeoclimatic and palaeogeographic conditions, facilitated the formation and shaping of the Elassona-Domeniko-Amouri basin and its fill with the Neogene and Quaternary sediments. The Late Neogene (Pliocene) sediments are unconformably deposited upon the schist-crystalline basement of the Pelagonian zone, they are significantly developed, and they mostly outcrop along the margins of Domeniko-Amouri sub-basin. During a mine project seeking the lignite deposits in the Domeniko-Amouri sub-basin, a dense network of boreholes was drilled, revealing important tectono-stratigraphic information for the study area 
. The stratigraphic description of the findings includes (from older to younger):
Basal formation of the Upper Miocene unconformably lying over the alpine basement, with a maximum thickness of 55 m. It consists of mud, sandy and silty clay, with granule and gravel layers, representing alluvial fan deposits that filled the initial stage of the basin’s formation. mostly of fluvio-torrential origin.
The gray-green lignite-bearing formation of the Upper Miocene, consisting of lignite beds, mostly at the upper members, within clayey silts, sandy clays, sands, clays, and rarely silts. Fine-grained sand passes into medium- and coarse-grained toward the lower members. Conglomerate intercalations also occur, rarely solid. Fossils lack. The lithofacies represent a fluvial environment developed by a meandering river. Maximum thickness of 150 m.
The overlying clastic deposits separated into (i) a 50 m-thick, fine-clastic lower member of Upper Miocene, consisting of friable and rarely stiff siltstones, sandy clays, sands, with local intercalations of conglomerates, and (ii) a 30 m-thick, coarse-clastic, unconformably overlying upper member of Upper Pliocene, consisting of unconsolidated to loose breccio-conglomerate with sand and boulders.
The Pleistocene deposits divided into two members: (i) the Upper Pliocene? to Lower Pleistocene (Villafranchian), 70 m-thick, terrestrial deposits composed by khaki sandy clays with mud, and hard, solid, white limestone intercalations, and (ii) the Upper Pleistocene, 60 m-thick, terraces and talus cone deposits, consisting of brown mud with sandy gravels and conglomerates, occasionally in thin interbeds.
The Holocene eluvial deposits and river terraces. They can reach 25 m of thickness.
According to the IGME’s 
geological maps, the oldest post-alpine sediments within Eastern Thessalian basin are deposited unconformably upon the alpine basement in the Upper Miocene; they comprise of transgressive, polygenetic, compact conglomerates, passing upwards to thick-bedded, micro-brecciated limestone. Lacustrine deposits of marls and marly sandstones follow in unconformity, and above them fluvio-terrestrial formations of sandy-clay material and loam, both of Plio-Pleistocene age. Pleistocene fluvio-lacustrine deposits of clays and sands with variously thick intercalations of coarse-grained material occupy a large extent of the northern margin, and on the top, Holocene alluvial deposits have covered the greatest part of the plain, consisting of unconsolidated clays, sand, angular and rounded pebbles, and fluvio-torrential-lacustrine material.
The Western Thessalian basin is partially imprinted on the older molassic formations of the Meso-Hellenic Trough, a (piggy-back or pull-apart) basin developed during the last alpine stages, from the Mid-Late Eocene to the Mid-Late Miocene 
. As in the other basins, the older post-alpine sedimentation started in the Late Miocene with the transgressive, polygenetic, compact conglomerates, which was followed by terrestrial clastic deposits of the Pliocene, fluvio-lacustrine deposits of the Pleistocene, and the alluvial deposits of the Holocene 
River terraces along the local fluvial systems, and slope debris with talus cones along the margins, also occur in all basins.
2.2. Post-Alpine and Active Tectonical Setting
The post-orogenic collapse that followed the Eurasian-Nubian plates convergence led to the first NE-SW-trending extensional (σ3
) phase which lasted from Late Miocene until Pliocene 
. This phase was responsible for the formation of the large basins in Thessaly following the main alpine structures. The next extensional phase 
has been active since Middle Pleistocene and remains active until Today. It has a general N-S direction (roughly N10°E of the σ3
axis) and affects most of the broader Aegean region 
. The stress field is reflected in the crustal deformation of the area which is depicted by GPS/GNSS analyses. Indeed, the upper crust seems to be dominated by dilatation, but towards the western margin of the Western Thessalian basin, i.e., toward the Pindos mountain range, shear locally occurs with the presence of contraction 
The result of the active stress field is the creation of a new generation of normal faults preferably perpendicular to the extension (i.e., E-W direction, N100° to N110° systematically), which overprint their geomorphological imprint along the margins of the Thessalian basin. The faults of northern Thessaly have been thoroughly studied, especially since the 1990s, resulting in the recognition of many active faults. The ones that affect the area of interest are the Tyrnavos and Larissa faults.
The Tyrnavos fault
is a typical normal E-W- to WNW-ESE-striking, N-dipping structure considered as active, similar to the many others in the broader area 
. It is clearly seen cutting the Mesozoic marble and it crosses diagonally the Titarissios valley.
It also shows typical morphological characteristics affecting younger sediments. It enters the Larissa plain from the west as proven from geophysical and palaeoseismological investigations near the basinal margin 
. Its surficial length is estimated to 10–12 km, although its basinal length remains unknown. Further details for this fault can be found in the GreDaSS 
The Larissa fault
is a sub-parallel (striking more to the WNW-ESE) normal fault that crosses most of the Larissa plain with an inferred length of ca. 20 km reaching the alpine basement to the west 
. Once again, further details for this fault can be found in the GreDaSS 
. However, based on the 1:50,000 scale geological map of IGME 
, the alpine basement is fractured by a fault system of the same direction, parallel to the passage that hosts the Penios River, which seems to reach the marginal fault of Titarissios valley, implying a possible connection.
The earthquake activity in northern Thessaly lacked any significant earthquakes before the 2021 sequence. Strong to major historically and instrumentally recorded events 
mostly occurred in the Eastern and Western Thessalian basins with the last one occurring in 1941. The very recent activity (since 2011, when the completeness of the earthquake catalogue 
was more improved) shows a large gap in Eastern Thessalian Basin, although large tectonically active faults occur, and a small, clustered, diachronic activity in the epicentral area. The Western Thessalian Basin shows much more frequent activity during those last 10 years.
The Domeniko-Amouri Basin Tectonic Setting
The previously mentioned mine project for the lignite deposits in the Domeniko-Amouri sub-basin with the borehole network 
revealed tectonic structures that are buried under the Holocene alluvial deposits. After correlating the borehole logs, vertically displaced (downthrown) horizons and the basin’s base were observed indicating the presence of normal Faults. Given that the layer succession is not always cut up to the top, the detected faults are interpreted of having variable ages: sometimes they displace only the deeper formations of the Late Miocene, characterized as “Neogene” faults, whereas other times they reach and displace the base of the Quaternary (Villafranchian), characterized as “post-Neogene” faults 
. The latter, which are of the interest, have mainly a NW-SE direction forming bookshelf or graben-style patterns. Along the southwestern margin of Domeniko-Amouri sub-basin, a fault of this direction, dipping to the NE, demonstrates a downthrow of several tens of metres.
4. The 2021 Seismic Sequence
In March 2021, two strong earthquakes were recorded close to the towns of Tyrnavos and Elassona: the first one occurred on March 3 with a magnitude of Mw
6.3; the second one occurred one day later, on March 4, few kilometres WNW of the first one, with a magnitude of Mw
6.0. Both events and the rich aftershock activity that followed them were shallow, with depths rarely exceeding 12–15 km 
. According to Michas et al. 
, “the aftershock sequence exhibited scaling properties consistent with well-known empirical relationships for aftershocks. In particular, the frequency of aftershock magnitudes follows the Gutenberg-Richter scaling law for a b
-value of 0.90 ± 0.03”. Published focal mechanisms from various institutes revealed NW-SE-striking normal faulting, and hence, NE-SW oriented extension (σ3
), which is quite different to the ca. N-S estimated from neotectonic investigations in previous studies. This faulting direction is also delineated by the horizontal distribution of the aftershock sequence 
The two strong shocks with the few kilometres distance between them strongly suggest the occurrence and reactivation of two adjacent and roughly aligned fault segments. This scenario can be also supported by the spatiotemporal evolution of the sequence. Indeed, the aftershocks that followed the first main event, and exactly until the second main event, were constrained to the west as far as the second mainshock lies. Right after the second main event, most of the aftershock activity migrates toward the WNW on a parallel direction, delineating the neighbouring sub-parallel segment. The overall picture shows a WNW-ESE- to NW-SE-trending horizontal distribution of the hypocentres extending from the western margin of the Larissa plain to the east, until the eastern slope of the Antichasia Mountains (toward the Domeniko-Amouri basin) to the West.
4.1. Ground Deformation Phenomena
The earthquake sequence produced a wide range of ground deformation phenomena, including (primary and secondary) coseismic ruptures, liquefaction, rockfalls, and landslides.
Coseismic ground ruptures abounded in the Domeniko-Amouri basin and the Titarissios valley. Almost persistently, they remained in a NW-SE direction, following the shape of the valley, parallel to the strike of the main faulting. They usually crossed farmed fields and fluvial deposits near the Titarissios riverbanks, as well as manmade constructions, such as asphalt roads and local irrigation network channels made of reenforced concrete. In fact, whatever the constructions direction was, the ruptures continued almost unhindered. However, not all of these ruptures are considered primary. Taking into account the alluvial-covered ‘Post-Neogene’ faults found by Dimitriou and Giakoupis 
Extensive liquefaction phenomena were manifested not only in the Domeniko-Amouri basin and the Titarissios valley, but in Penios valley (Peniada and Zarko) as well 
. They were mostly continuous and aligned along ground fissures.
Secondary gravitational effects other than liquefaction, such as rockfalls and landslides, were observed throughout the area. Their distribution did not follow a specific pattern, but was rather dependant on the local site conditions, i.e., rock formation attributes and slope angle. Quite common was lateral spreading observed next to riverbanks, such as in Damasi, Mesochori, and other nearby areas.
The extent and intensity of ground deformation are revealed from InSAR (Sentinel-1) images 
. Utilizing a pair of images just before and after the first mainshock, the uplift and subsidence were estimated, also delineating the surficial location of the seismic fault in the alpine basement of the Zarko Mt. Similar results have been published by other research teams as well 
. The results of Yang et al. 
, based on InSAR analysis and relocated aftershocks, suggest that at least four unmapped low-angle normal faults were activated. They also suggest that afterslip propagated along a shallow steep fault while coseismic slip occurred on a deep, moderately dipping fault, revealing a ramp-flat structure.
4.2. The Seismic Sources
The determination of the seismic sources that produced the two mainshocks has been the main issue, as well as in many already published research. The principal evidence on which all research teams based their models are (i) the hypocentral locations, especially of the two mainshocks, (ii) the co-seismic ground ruptures, and (iii) the InSAR images 
. Some researchers focused only on the Domeniko-Amouri basin and the Titarissios valley where most of the ground ruptures occurred 
. However, few paid attention to the fresh slickensides southwards, within the alpine gneiss-schist basement 
. Taking into account both sets of primary coseismic ruptures, the InSAR fringes, and the geometry suggested by all published focal mechanisms, two roughly parallel faults were ruptured during the first strong event: the southern one, the Zarkos fault
, accommodated the main slip, whereas the northern one, the Titarissios fault
, was partially ruptured generating the eastmost ground fissures along the Titarissios valley. The rupture of the northwestern adjacent fault segment, i.e., the one that produced the next day’s (March 4) strong shock probably produced the westmost ground fissures between the villages of Vlachogianni and Amouri and is probably related to the blind ‘post-Neogene’ faults of Dimitriou and Giakoupis 
There is no doubt that the two mainshocks were produced by two adjacent and aligned faults or fault segments, with a small strike difference. The ground ruptures in the Domeniko-Amouri basin and along the Titarissios valley strongly support the surficial expression of the seismic faults, but they do not coincide to InSAR results. This means that the rupture of the first major event bifurcated along two strands, (the northern) one reaching the Titarissios valley, and the other (southern one) cropping out in the alpine basement on Zarkos Mountains. A corresponding situation failed to be observed for the second major event. The detachment hypothesis is also referred by Koukouvelas et al. 
, noting the complex geometry caused by the possible occurrence of a low-angle, detachment fault.
The study of the static stress change patterns due to fault slip is crucial in order to understand the effect of the seismic source on the surrounding receiver-faults; the study is achieved with the Coulomb static stress modelling. Plenty of the published works about the currently studied earthquake sequence present stress transfer models which diversify due to the input parameters used in each case (e.g., fault location and geometry) 
. Chatzipetros et al. 
calculated the Coulomb static stress changes at a depth of 8 km for receiver faults similar with the seismic source. Their source fault model was based on the published focal mechanisms and the hypocentral spatial distribution of the first mainshock (March 3). Their results show bilateral stress load along fault strike which can explain the triggering of the adjacent fault (NW-ern ‘segment’ of the Zarkos fault zone) that produced the second mainshock on March 4. In the study of Kassaras et al. 
, several stress models clearly show the different stress patterns of each one of the main events corresponding to the relocated aftershocks cloud.