**2.2. The 9663 BP event**

Varve 10,430 BP was originally interpreted in terms of a "drainage varve" [16], an old idea now substituted by the recording of the huge earthquake in varve 10,430 BP and its extensive

**Figure 6.** Red dots represent sites of recorded tsunamites. Green field represents the spatial distribution of firmly varve-dated tsunamites, seismites and liquefaction features with sites outside this field referring to sites dated by other

**Figure 7.** Paleogeography of the land–sea–ice distribution at the time of the big earthquake in varve-year 10,430 BP (modified from [3]). Red dot marks location of epicentre. Red double arrows refer to the Närke Straight opened by the tsunami wave so that marine water could enter the Baltic and turn it into the Yoldia Sea stage. Red cross marks the location of two sites recording an earthquake event with liquefaction and a tsunami event occurring 67 varves after

means.

120 Tsunami

deglaciation [17].

tsunami event [1, 3, 4]. The spatial distribution is given in **Figures 6** and **7**.

In the Hudiksvall area of central Sweden, there occurred a very large earthquake in the varveyear 9663 BP [1, 3, 4, 9, 12, 18, 19]. The paleogeography of this event is very well known (**Figure 9**). Sea level was in the order of +224–230 m, and only minor islands stack out of the sea in front of the ice margin. It is recorded by multiple criteria [1, 4, 9] as illustrated in **Figure 8**. It seems to represent one of the best documented paleoseismic events in the world [1, 2]. The intensity was estimated at XII and the magnitude as >8 [2, 9]. Bedrock fracturing is recorded in some 100 sites over an area of 50 × 50 km. Liquefaction is recorded at 12 separate sites covering an area of 80 × 40 km. At two sites, it is recorded in five separate phases thought to represent the main event and a sequence of aftershocks. The liquefaction event is directly tied to varve 9663 BP. The water depth at the epicentre is likely to have been in the order of 250 m (allowing for a tsunami wave with the same diameter). A tsunami wave is recorded at 14 different sites, including nine lakes where a total of 44 cores were taken. It is dated both by varves (at 9663 varve-years BP) and by radiocarbon (at about 9150 C14-years BP). The tsunami wave must have a height of at least 15 m [1, 3, 4, 9]. A turbidite in varve 9663 BP extends for 310 km along the coast [1, 19].

**Figure 8.** The 9663 BP paleoseismic event must have had a magnitude of *M* > 8. It is documented by multiple criteria [1, 3, 4, 9, 18].

**Figure 9.** Paleogeography of the Hudiksvall area at the high-magnitude earthquake in the varve-year 9663 BP (or ∼9150 C14-years BP). Explanations: blue: extension of the ice cap; brown: land areas; black winding strings: eskers; red line with dot: fault line and location of epicentre; red dots with numbers 1–14: sites where tsunamites of the 9663 BP event have been documented (from [1]).

**Figure 9** provides the paleogeography of the area in the varve-year 9663 BP and the location of the 14 sites where records of the tsunami event have been recorded in the near-field region [1]. The area is traversed by six separate drainage basins marked by the corresponding esker systems from the subglacial drainage. The turbidite of varve 9663 BP traverses the entire area and expands 310 km to the south.

The tsunami was probably composed of two to three waves judging from the cycles recorded (**Figure 10**). Site 2 (Lake Svartsjön) is the key site of 14 closely spaced cores, five C14-dates and diatom analysis [1, 3]. The site was deglaciated 25 years before the tsunami event. The highest coastline (HK in **Figure 10**) was closely determined at +31.3 m. At the time of the tsunami event, the Baltic level had fallen to +223.5 m, i.e., a drop of about 7.8 m in 25 years, providing a rate of relative uplift of about 312 mm/year. The five C14-dates all overlap at 9155-9135 C14-years BP. A C14-age of 9150 BP would correspond to an absolute age of 10,350 cal. years BP. The deviation between the absolute age and the varve age of 9663 BP records an error in the varve chronology of about 700 years [20].

**Figure 10.** A 5.5 km SW–NE profile of the topographic setting of sites 2–5 (from [1]).

**Figure 9.** Paleogeography of the Hudiksvall area at the high-magnitude earthquake in the varve-year 9663 BP (or ∼9150 C14-years BP). Explanations: blue: extension of the ice cap; brown: land areas; black winding strings: eskers; red line with dot: fault line and location of epicentre; red dots with numbers 1–14: sites where tsunamites of the 9663 BP

**Figure 9** provides the paleogeography of the area in the varve-year 9663 BP and the location of the 14 sites where records of the tsunami event have been recorded in the near-field region [1]. The area is traversed by six separate drainage basins marked by the corresponding esker systems from the subglacial drainage. The turbidite of varve 9663 BP traverses the entire area

The tsunami was probably composed of two to three waves judging from the cycles recorded (**Figure 10**). Site 2 (Lake Svartsjön) is the key site of 14 closely spaced cores, five C14-dates and diatom analysis [1, 3]. The site was deglaciated 25 years before the tsunami event. The highest coastline (HK in **Figure 10**) was closely determined at +31.3 m. At the time of the tsunami event, the Baltic level had fallen to +223.5 m, i.e., a drop of about 7.8 m in 25 years, providing a rate of relative uplift of about 312 mm/year. The five C14-dates all overlap at 9155-9135 C14-years BP. A C14-age of 9150 BP would correspond to an absolute age of 10,350 cal. years BP. The deviation between the absolute age and the varve age of 9663 BP records an error in the varve

event have been documented (from [1]).

122 Tsunami

and expands 310 km to the south.

chronology of about 700 years [20].

**Figure 11** provides a good and representative record of the stratigraphy and sequence of events in Lake Svartsjön (Site 2). After the deposition of 25 varves, a tsunami wave invades the lake basin by overflowing the sill to the east (about +228 m) and depositing a thick tsunamite spread over the entire lake basin. This calls for a minimum height of the tsunami wave of 6 m (cf. 15 m at Site 3).

**Figure 11.** Stratigraphy, dates, tsunami characteristics and diatom analyses of Core 3 in Site 2 [1, 3]. The tsunamites spans 60 cm and is composed of a fragmentary first cycle and a full main cycle of graded bedding fining upward (sand-silt-clay). The diatom content of the tsunamite represents a planktonic deep-water flora of the open Baltic Lake Ancylus stage [3].

The tsunamite has a thickness 50–60 cm and is composed of two (occasionally 3) depositional cycles of graded bedding. The diatom flora of the tsunamite represents planktonic deep-water species from the open Baltic environment of the Ancylus Lake stage. This is a useful characteristic of sandy beds deposited by tsunami waves from the sea (contrary to shallow-water beach erosion).

The Lake Källsjön records are important [1]. This lake was never a part of the Baltic. The sill in-between had an original level of about +236 m [3] (**Figure 12**). Five cores were obtained in the lake, all including a typical tsunamite of graded bedding, indicating a forceful overflow of the sill and ingression into the lake of the 9663 BP tsunami wave. The wave must be at least 12.5 m high and overflow the sill for 700 m. Therefore, we can set the tsunami height at 13 m or rather about 15 m. The full evidence of a major tsunami event comes from the content of diatoms in the tsunamite beds, viz., a typical planktonic diatom flora of the Baltic Lake Ancylus stage [1, 3]. Furthermore, in today's lake, a small fish, smelt, is living, which is considered to be a relict from the Lake Ancylus stage of the Baltic, washed into the lake by the 9663 BP tsunami wave.

**Figure 12.** Site 3, Lake Kjällsjön, was never a part of the Baltic because it was located above the highest Baltic limit (BL) and separated from it by a sill [1, 3]. At the 9663 BP event, the sill was overwashed and the lake was invaded by a wave from the Baltic. The tsunami height must have been at least 13 m, probably about 15 m high. From [3].

A final example of the 9663 BP tsunami comes from the harbour of Iggesund (site 14 in **Figure 9**), where we had five pits and a major trench cut [1, 4]. The area was free-melted in 9747 BP. Then, followed the deposition on 84 annual varves. In varve-year 9663 BP, the area was affected by a strong earthquake. The glacifluvial sand and silt was heavily liquefied, and the liquefied material vented to the surface where it spread laterally by mushrooming. Some lateral spreading also occurred along the surfaces of varve beddings. In pits 4 and 5, the intraclay sand layer had the character of a tsunami bed. In pit 4, we brushed off the sand covering the clay surface beneath. The surface exposed was traversed by furrows and wave patterns (**Figure 13b**) indicating that the tsunamites must have been deposited by a flow from the NE, which implied from the sea towards the coast. At that stage, we had not yet located the epicentre of the paleoseismic event. Later it was determined that the epicentre of the 9663 BP event must have been located 12 km to the NE (as marked in **Figure 9**). Consequently, our sedimentary records in the pits at Iggesund (**Figure 13**) are in full agreement with this location of the epicentre.

The tsunamite has a thickness 50–60 cm and is composed of two (occasionally 3) depositional cycles of graded bedding. The diatom flora of the tsunamite represents planktonic deep-water species from the open Baltic environment of the Ancylus Lake stage. This is a useful characteristic of sandy beds deposited by tsunami waves from the sea (contrary to shallow-water

The Lake Källsjön records are important [1]. This lake was never a part of the Baltic. The sill in-between had an original level of about +236 m [3] (**Figure 12**). Five cores were obtained in the lake, all including a typical tsunamite of graded bedding, indicating a forceful overflow of the sill and ingression into the lake of the 9663 BP tsunami wave. The wave must be at least 12.5 m high and overflow the sill for 700 m. Therefore, we can set the tsunami height at 13 m or rather about 15 m. The full evidence of a major tsunami event comes from the content of diatoms in the tsunamite beds, viz., a typical planktonic diatom flora of the Baltic Lake Ancylus stage [1, 3]. Furthermore, in today's lake, a small fish, smelt, is living, which is considered to be a relict from the Lake Ancylus stage of the Baltic, washed into the lake by the 9663 BP tsunami

**Figure 12.** Site 3, Lake Kjällsjön, was never a part of the Baltic because it was located above the highest Baltic limit (BL) and separated from it by a sill [1, 3]. At the 9663 BP event, the sill was overwashed and the lake was invaded by a wave

A final example of the 9663 BP tsunami comes from the harbour of Iggesund (site 14 in **Figure 9**), where we had five pits and a major trench cut [1, 4]. The area was free-melted in 9747 BP. Then, followed the deposition on 84 annual varves. In varve-year 9663 BP, the area was affected by a strong earthquake. The glacifluvial sand and silt was heavily liquefied, and the liquefied material vented to the surface where it spread laterally by mushrooming. Some lateral spreading also occurred along the surfaces of varve beddings. In pits 4 and 5, the intraclay sand layer had the character of a tsunami bed. In pit 4, we brushed off the sand covering

from the Baltic. The tsunami height must have been at least 13 m, probably about 15 m high. From [3].

beach erosion).

124 Tsunami

wave.

**Figure 13.** Stratigraphic records at site 14, Iggesund harbour [1, 4] recording liquefaction, venting and surface mushrooming of liquefied material, and the deposition of a tsunamite. The clay surface beneath the tsunamite in pit 4 is irregular (furrows and waves) in a manner indicative of a deposition from the NE (from right to left), i.e. from the sea outside.

**Figure 14** provides the spatial distribution of the 9663 BP tsunami event in Sweden. The extension of this tsunami even will surely expand considerably with time to sites in Finland, Sweden and the Baltic coast to the SE, where there are numerous sites of proposed transgression deposits of the Ancylus Lake dated at the same age as the tsunami. Most probably, several of these sites must be reinterpreted in terms of tsunamites [1].

**Figure 14.** Red dots represent sites of recorded tsunamites. Green field represents the spatial distribution of firmly varve-dated tsunamites (also seismites and liquefaction features) with sites outside this field referring to sites dated by other means.

#### **2.3. The Late Holocene events**

The 17 tsunami events recorded in Sweden over the last 13,000 years (**Figure 1**, **Table 1**) are fairly regularly distributed over time. In the past 4000 years, there were as many as seven events observed [3, 21, 22], three of those are discussed below.

This includes a revision of the previously assigned age of the Hudiksvall event; now re-dated at 2900 BP, instead of 2000 BP. This makes the two events synchronous though recorded 160 km apart.

#### **2.4. The 2900 BP event recorded at two sites 160 km apart**

Site 1 refers to a well recorded at Skålbo, north of Hudiksvall [1, 3, 22, 23]. Here, violent methane-venting tectonics set up a tsunami wave recorded in five bogs at +8, +14, +18, +23 and +38 m [1, 3, 9, 22]. Lake Dellen, 25 km to the west and today at +37 m has a submerged peat dated at about 2000 BP [3]. This age was used to determine the age of the event. At 2000 BP, sea level was at +18 m, implying a wave height of 20 m [1]. Consequently, the run-up must have been at least 20 m in order to invade the +38 m bog.

**Figure 14** provides the spatial distribution of the 9663 BP tsunami event in Sweden. The extension of this tsunami even will surely expand considerably with time to sites in Finland, Sweden and the Baltic coast to the SE, where there are numerous sites of proposed transgression deposits of the Ancylus Lake dated at the same age as the tsunami. Most probably, several

**Figure 14.** Red dots represent sites of recorded tsunamites. Green field represents the spatial distribution of firmly varve-dated tsunamites (also seismites and liquefaction features) with sites outside this field referring to sites dated by

The 17 tsunami events recorded in Sweden over the last 13,000 years (**Figure 1**, **Table 1**) are fairly regularly distributed over time. In the past 4000 years, there were as many as seven events

This includes a revision of the previously assigned age of the Hudiksvall event; now re-dated at 2900 BP, instead of 2000 BP. This makes the two events synchronous though recorded 160

Site 1 refers to a well recorded at Skålbo, north of Hudiksvall [1, 3, 22, 23]. Here, violent methane-venting tectonics set up a tsunami wave recorded in five bogs at +8, +14, +18, +23 and +38 m [1, 3, 9, 22]. Lake Dellen, 25 km to the west and today at +37 m has a submerged peat dated at about 2000 BP [3]. This age was used to determine the age of the event. At 2000 BP,

of these sites must be reinterpreted in terms of tsunamites [1].

other means.

126 Tsunami

km apart.

**2.3. The Late Holocene events**

observed [3, 21, 22], three of those are discussed below.

**2.4. The 2900 BP event recorded at two sites 160 km apart**

Later, I come to question the age determination, however. First, the Lake Dellen pounding may have nothing to do with the tsunami event. Second, we find no records of the 2900 BP event in the Hudiksvall lakes and no records of the 2000 BP event in the Forsmark lakes. Third, subsequent C14-dates of the lake deposits provide ages older than 2000 BP.

Now, it seems quite clear that the Hudiksvall event should be synchronized [23] with the Forsmark tsunami event closely dated at 2900 BP.

At 2900 BP, sea level in the Hudiksvall area would have been at +26 m [23]. This implies a tsunami height of at least 12 m (to deposit a tsunamite in the +38 m bog). The tsunami wave had a trimming effect of the seabed (i.e., erosion) down to at least +8 or 18 m below sea level at that time.

**Figure 15.** The 2900 C14-years BP tsunami event in northern Uppland with respect to the rate of land uplift and shore displacement over the last 4000 years (the oblique line of ∼7 m uplift per millennia passing through dated anchor points marked by blue dots). At 2900 C14-years BP, the shore was at +20.7 m with land above (yellow) and sea (blue) below as marked on the right side of the diagram. The stars mark tsunami beds recorded and dated in offshore sediments (all falling sharply at the 2900 BP level), in coastal deposits and in lakes and bogs on land where the tsunami beds have eroded down into the older sediments. The graph provides evidence of a tsunami event that deposited typical tsunami beds over a vertical range from −20 to +6 m.

The Forsmark event (**Figure 15**) refers to a tsunami event very closely C14-dated at 2900 BP, which was recorded in seven lakebeds in northern Uppland, 160 km to the south of the Hudiksvall records [3, 9, 22]. The lakes investigated and dated form a staircase in elevation, viz., +5.5, +10.8, +16.0, +16.0, +22.2, +22.5 and +24.7 m. At the time of the tsunami event, sea level was at +20.7 m, implying that four lakes were located below and three above sea level at 2900 BP (**Figure 15**). In all the lakebeds, we found nice tsunamites of sandy layers in graded bedding. A lakebed at +27.4 m [24] shows no sign of a tsunami ingression. Therefore, the upper limit of the tsunami event is set at about 26.5 m, indicating a tsunami height in the order of 6 m (and a submarine trimming depth of about 20 m).

Several lakes and bogs in the region had been investigated before [24–26], but none of those studies recorded the occurrence of tsunamites despite the fact that they are quite clearly present in the lakes at +5.5, +10.8, +16.0 and +22.5 m [26], all later cored and dated by the author, as shown in **Figure 15**.

Synchronizing the events recorded at Skålbo and Forsmark (**Figure 16**) implies a fall in tsunami height of 6 m over a distance of 160 km, which seems quite reasonable.

**Figure 16.** Synchronization of the tsunami events recorded at Skålbo (Hudiksvall) and at Forsmark (northern Uppland) at 2900 BP [23]; elevation of dated cores (red dots), sea level at 2900 BP (dark blue lines). The tsunami wave height of at least 12 m in Skålbo decreases to at least 6 m in Forsmark, i.e., a decrease of 6 m in 160 km. The basal trimming seems to go down at least 18 m below sea level.

#### **2.5. The AD 1174 event**

In SW Sweden, there are evidence of a young paleoseismic event with local faulting of the Viking shoreline. This implies a faulting post-dating the formation of this shoreline, known as PTM-10, and dated at 1000-950 BP [1, 22, 27]. A fault offset of 1.1 m is recorded [1, 22].

It seems highly likely that this earthquake also set up a tsunami wave, which buried two Viking ships in the ancient harbour of Galtabäck [27]. A C14-date of the ship gave 1172 ±73 cal. years AD [28], which is quite close to a major historical earthquake event in 1174 as recorded in chronicles [29]. The covering and burying of the two ships in silt are indicative of a very rapid (instantaneous) process that took the ship owners by surprise, fully in line with the process of a tsunami event.

Therefore, it was proposed [27] that, indeed, the earthquake set up a tsunami and that the event refers to the earthquake in 1174 mentioned in the chronicles [29], which corresponds to 776 cal. years BP.

Having established this, it is possible to re-evaluate a turbidite record from the Gullmar Fjord [30] 230 km to the NW in terms of a tsunamite (**Figure 17**).

**Figure 17.** The time/depth relations of core 9004 [30] and the position of the "turbidite" here revised in terms of a tsunamite of the 1174 AD earthquake event.

An inferred age of the "turbidite" in the Gullmar Fjord core of 1174 AD fits perfectly well with the time/depth curve of [30]. I therefore, quite confidently, re-evaluate this layer as a tsunamite of the 1174 paleoseismic event [27].
