Tsunami records of the last 8000 years in the Andaman Island, India, from mega and large earthquakes: Insights on recurrence interval

As many as seven tsunamis from the past 8000 years are evidenced by sand sheets that rest on buried wetland soils at Badabalu, southern Andaman Island, along northern part of the fault rupture of the giant 2004 Aceh-Andaman earthquake. The uppermost of these deposits represents the 2004 tsunami. Underlying deposits likely correspond to historical tsunamis of 1881, 1762, and 1679 CE, and provide evidence for prehistoric tsunamis in 1300–1400 CE, in 2000–3000 and 3020–1780 BCE, and before 5600–5300 BCE. The sequence includes an unexplained hiatus of two or three millennia ending around 1400 CE, which could be attributed to accelerated erosion due to Relative Sea-Level (RSL) fall at ~3500 BP. A tsunami in 1300–1400, comparable to the one in 2004, was previously identified geologically on other Indian Ocean shores. The tsunamis assigned to 1679, 1762, and 1881, by contrast, were more nearly confined to the northeast Indian Ocean. Sources have not been determined for the three earliest of the inferred tsunamis. We suggest a recurrence of 420–750 years for mega-earthquakes having different source, and a shorter interval of 80–120 years for large magnitude earthquakes.

The ENE-WSW striking Badabalu beach along the south coast of Andaman is a famous tourist destination (Fig. 1b,c). The Badabalu area was severely affected by the 2004 tsunami (Sumatra-Andaman earthquake) and experienced a coseismic subsidence of ~40-45 cm ( Supplementary Fig. S1.1a-h). Google Earth images from 2004 to 2014 clearly exhibit the pre-and post-seismic changes in coastal geomorphology (Fig. 1c,d, Supplementary Figs. S1.1a-h and S1.2a,b).
According to the survivors, the Badabalu area experienced ~4 m high tsunami waves, with run-up up to 0.8-1.0 km from the coast. Coseismic subsidence resulted into a landward shifting of the beach by 35-50 m (Supplementary Figs. S1.1a-h and S1.2a,b). Landward migration of the beach also caused inundation of the area and formation of beach ridge, back-marsh inland, and dead forest along the coastline. Local residents artificially filled the area by 0.7-0.8 m, shifted their houses to the higher ground, and elevated the coastal roads to avoid inundation around residential area and agricultural fields during high-tides (Supplementary Figs. S1.2-S1.4).

Results
Stratigraphy. Based on the sedimentary structure, grain size, depositional and/or erosional contacts observed in litho-sections, 19 litho-units (a' , a-r, from top to bottom) were identified (Figs. 2b, 3 and 4; Supplementary Data S2; Figs. S2.1-S2. 4, Tables S2 and S5.1). Unit-a' is the youngest lithounit, represent present day beach-ridge facies. Unit-a is present-day peaty soil (humic), medium-fine sand observed from the backmarsh. Unit-b is 2004 tsunami yellowish medium-coarse sand. It is coarser and thicker near to the coast, and becomes thinner and finer towards inland (Fig. 1c, 2b,  Unit-i is structureless grayish fine silty-sand, with sharp to gradual contact with Unit-j. Considering the thickness of 55-110 cm and finer nature of the unit, we infer that deposition took place during basin-filling, and the area remained submerged for a longer span during inter-seismic period (Figs. 2b and 4c; Supplementary Figs. S2.3-S2.4). Unit-j is peaty soil with fine sand, shows gradual contact with the overlying unit and a sharp contact with the Unit-k. We infer that the area was at or above mean sea-level before subsidence. ( 4). The unit is thicker and coarser towards the ocean, and finer and thinner towards inland. It shows prominent inclined stratification with bi-directional structure, alternative layers of greyish medium-coarse sand with silt and fine gravel clasts containing broken shell and coral fragments, plant debris, and rip-up clasts of bedrock (Fig. 2b, 3d and 4a-d). We suggests that this unit was deposited by a tsunami triggered by an earthquake that caused coseismic subsidence as indicated by overlying finer Unit-k (Fig. 2b). Unit-m is peaty soil with silty-sand, which separates Unit-l from Unit-n with sharp contacts (Figs. 2b and 4a-d; Supplementary Figs. S2.2-S2.4). Considering the study area in proximity to the ocean, we infer that this peaty unit was formed due to land-level change (Fig. 2b). Unit-n is greyish medium-coarse sand, marked by inclined laminations, with thin layers of coarser fragments comprised of broken shells, and rip-up clasts in the upper portion. It shows sharp contact with underlying peat Unit-o, which formed at or above mean sea-level (Figs. 2b and 4b-d; Supplementary Figs. S2.3). We infer that this unit (Unit-n) was deposited by tsunami generated by local event along the Andaman segment (Figs. 2b and 4b-d; Supplementary Figs. S2.3). Unit-p, a thick grayish coarse sand, with corals clasts, broken shells and rock fragments, is exposed at a depth of ~2 m, and shows sharp contacts with Unit-o and Unit-q (Figs. 2b and 4c,d). This unit in some sections shows poor lamination, with scattered gravels observed in the upper, and the middle portions as well as inverse grading (Fig. 4b-d). The unit was deposited by a tsunami event (Fig. 2b). Unit-q is a peaty unit composed of greyish fine sand with scattered gravel fragments (Figs. 2b and 4b). It shows sharp contacts with the underlying and overlying units (Fig. 2b), and possibly formed at or above mean sea level. Unit-r is coarse sand with broken shell fragments, deposited by a tsunami event (Figs. 2b and 4d).
Synthesis of sedimentological (structures, grain size, lithology), geochemical (major and trace element abundances) and biological (foraminifera) data suggest that Units b, d, g, l, n, p and r were deposited by sudden high-energy wave events -tsunamis 3,18,19 (Figs. 2-6; Supplementary Data S2-S5). Clear discrimination between tsunami and storm deposits is difficult. However, most of the cyclones around Anadman start to develop at their initial stage and are not strong enough to affect the sedimentation pattern 3 . This rules-out the possibility of the identified deposits to be non-tsunami origin. Also, the lithounits identified in the exposed geoslicers and trenches show distinct sedimentological signatures like alternate layers of medium sand and silt or coarse sand, broken shell and coral fragments, poorly sorted sediments, normal to inverse grading, rip-up clasts, plant material, inclined stratification, bi-directional structures etc. Further, it is also argued that usually tsunami deposits show layers with bi-directional flow, i.e., towards landward and seaward directions, whereas, storm or cyclone deposits do not show such bi-directional flow 20,21 . Tsunami deposits usually show bi-modal distribution of grain size, whereas, storm deposits are well sorted 21 . In our study we found layers with bi-directional flow as well as bi-modal distribution of grain size. Hence, we conclude that the deposits from Badabalu are deposited by tsunami events.
Micro-fossil analysis. Quantitative analysis of foraminifera obtained from Units b, d, e, g, n and p indi-   Fig. 1d for location). Locations of geoslices and trenches for shallow stratigraphic record are marked along the profile. The area shows beach ridgeswale-beach ridge topography. The middle portion of the profile shows artificial fill by local residents to prevent inundation. LLT -lower low tide, MSL -mean sea level and HHT -higher high tide. (b) Geoslice and trench sections placed with respect to horizontal. The vertical scale represents depth from the surface. The distribution of all exposed sedimentary lithounits was correlated. Based on the sedimentary characteristics the exposed units were classified into total 19 units, from the youngest Unit-a' to the oldest Unit-r. In total seven tsunami deposits (including 2004 tsunami) were identified from the exposed stratigraphic sequence ranging in depth from 160-270 cm. Units b, d, g, l, n, p and r represents tsunami deposits marked by yellow colour. (2019) 9:18463 | https://doi.org/10.1038/s41598-019-54750-6 www.nature.com/scientificreports www.nature.com/scientificreports/ and underlying peaty soil (Unit-c) exposed in GS6. Unit-b is marked by inversely graded yellowish medium-coarse sand with broken shells. It also shows prominent laminations, and a sharp contact with underlying unit (Unit-c). (b) Close-up view of Units d, e and g exposed in GS7. Unit-d (paleo-tsunami) comprises coarse to medium sand with fine gravel clasts and broken shell fragments. It shows bi-directional structure with normal to inverse grading and a sharp contact with underlying peaty unit (Unit-e). Unit-g comprises coarse to medium sand with broken shell and coral fragments. It shows inverse grading and bi-directional structure.  www.nature.com/scientificreports www.nature.com/scientificreports/ Unit-p shows dominance of Ammonia beccarii -an intertidal species (Supplementary Fig. S3.3). Foraminifera assemblages from Units b, d, and g also show same kind of species and taphonomy. These units also show high percentage of abraded and fragmented foraminifera test (Supplementary Fig. S3.1). The peaty soil Unit-e with Elphidium discoidale suggests shallow intertidal-beach environment. The change in the environment from intertidal-beach to wetland is attributed to interseismic uplift. This further strengthen our interpretation that Units b, d, g, n and p were deposited by tsunami events, which transported and deposited forams from different depths (Supplementary Data S3).  Table S5.1). Tsunami and non-tsunami deposits (i.e., terrigenous) identified on the basis of sedimentological and microfossil proxies showed differences in geochemical signatures (Supplementary Data S3 and S5). In the upper section (<50 cm), the tsunami Units b, d, and g have distinctly lower abundances of Al 2 O 3 , Fe 2 O 3 , and K 2 O compared to those of the intermittent (non-tsunami) Units a, c, e, f, and h (Fig. 5). However, the tsunami layers (Units l, n, and p) in the bottom section (>150 cm) do not show lower abundances of Al 2 O 3 , Fe 2 O 3 , and K 2 O relative to the adjacent non-tsunami Units k, m, and q (Fig. 5). Interestingly, the tsunami deposits in the bottom section are thicker compared to those in the upper portion. It is possible that the some of the distinct geochemical signatures in these older tsunami deposits are disturbed due to prolonged burial and leaching 27 . On the other hand, all the tsunami Units b, d, g, l, n and p contain distinctly higher CaO and MnO than the other terrigenous Units a, c, e, f, h, i, j, k, m, o, and q. High abundance of CaO and MnO is a characteristic signature in tsunami deposits 27 . In general, the terrigenous units are also characterized by the higher contents of SiO 2 , TiO 2 , MgO and Na 2 O compared to the tsunami units (Fig. 5). The major oxides abundances further strengthen our interpretation that Units b, d, g, l, n and p are of marine origin, and deposited inland during tsunami events.

Dating (OSL and AMS).
Compared to the terrigenous units, the tsunami Units b, d, g, l, n and p are generally enriched in alkali elements, in particular Ca, Na, K, Sr and have higher Ca/Sr, Na/K, and Sr/Ba ratios (Supplementary Fig. S5.1; Table S5 Table S5.1). Cerium can exist in +3 or +4 oxidation states depending on the redox conditions. The insoluble Ce 4+ is prone to be adsorbed and sequestered by Mn-oxides and hydroxides under oxidizing environment and thus marine sediments rich in Fe-Mn exhibit positive Ce anomalies 29 . The range of Ce/Ce* values (1.0 to 2.55) in the sediments from tsunami layers confirm an origin in anoxic to suboxic environment. Thus, characteristic geochemical signatures in Units b, d, g, l, n and p further affirms their tsunamigenic origin.

Discussion and conclusions
The signatures of 2004 Sumatra-Andaman earthquake and tsunami were considered as a modern analogue to distinguish the role of local and distant source earthquakes towards the deposition of tsunami deposits. At Badabalu, we found relatively thicker and coarser deposits (Units l, n and p) as compared to Unit-b deposited by 2004 tsunami. The presences of thicker deposits could be attributed to the paleo-shoreline morphology. Possibly at the time of deposition the beach-ridge and associated back-marsh were located farther inland relative to the present coastline configuration, with deposition taking place in a swale or back-marsh area. Further, the coarser and thicker deposits could be related to tsunami events with much higher energy conditions, which was possible inclined stratification shows sharp contact with the overlying and underlying units (Units m and o). (d) The GS10 geoslice collected from back-marsh shows thin medium sand layer of 2004 tsunami sandwiched between peaty soils pre-(Unit-a) and post-(Unit-c) 2004 event with sharp contacts. Unit-l comprised of sandstone clasts (gravel) and coarse sand shows a gradual contact with the overlying Unit-j and a sharp contact with underlying Unit-m. Units n and p are paleo-tsunamis sharing the same contact. Unit-o is missing in this section. Unit-n is a medium sand unit showing inclined laminations, rip-up clasts and broken shell and coral fragments. Unit-p is medium-coarse sand with coarser fragments of coral clasts and broken shells with bi-directional structures. Unit-p shows sharp contacts with underlying Unit-q (peaty soil) and overlying paleo-tsunami (Unit-n). Unit-r comprise coarse sand with fine gravel clasts, occurs at a depth of ~270 cm. Refer Fig. 2a, www.nature.com/scientificreports www.nature.com/scientificreports/ by a major earthquake triggered along the Andaman-Arakan Segment, suggesting a local earthquake. This is well justified comparing the 2004 tsunami deposit at the same location. Therefore, we infer that the thicker and coarser units (viz. Units l, n and p) were deposited by the local-source earthquakes those occurred along the Andaman-Arakan Segment. Paleo-tsunami and paleoseismic events identified in the present study were correlated with the reported events from the areas adjoining Indian Ocean (Supplementary Table S1).
The present study from Badabalu revealed evidence of at least seven tsunami events in the last 8000 years (Figs. 2b, 6 and 7; Supplementary Data S4a-b; Tables S4.1 and 4.2). These events were bracketed based on their modelled calendar ages 22,23 (Tables 1 and 2; Supplementary Tables S4.1 Tables 1 and 2; Supplementary Table S1). We correlate this event with the 1881 (Mw 7.9) Car Nicobar earthquake, which was felt over much of India and parts of Burma as well as in the Bay of Bengal 30 . This was a local event that occurred along the Andaman segment, generated 0.8 m high tsunami, resulted in an uplift of 10-60 cm at Car Nicobar 30 , but did not have a widespread effect. Event-III (Unit-g) took place around CE 1747-1850, after CE 1649-1787 and before CE 1838-1833, and correlates with an earthquake of CE 1762 (Mw7.5) (Figs. 2b, 6 and 7; Tables 1 and 2; Supplementary Table S1). This event occurred along the Arakan Subduction Zone that caused uplift of ~3-7 m along the coasts of Ramree, Cheduba, and Foul Islands, located offshore of the Arakan coast of Myanmar 10 , and also generated a tsunami 31 . Signatures of liquefaction and tsunami deposit were also reported from Mitha-Khadi around Port Blair in Andaman Island 6 . However, no clear evidence of land-level change was found at the present study site. This suggest that Andaman was at the southern tip of this rupture. Event-IV (Unit-j) occurred after CE 1463-1581 and before CE 1747-1850. The event correlates with the historic earthquake of CE 1679 (Figs. 2b, 6 and 7), which was felt around Arakan (Burma), Bangladesh, Chennai, and areas adjoining Indian Ocean 32 . This event may also be correlated with: (i) the event occurred along the Andaman Segment, that accompanied land-subsidence during CE 1600 from Mitha-Khadi near Port Blair 6 , (ii) a tsunami event of CE 1640-1950 reported from Sumatra 8 , and/or (iii) with a tsunami event of CE 1530-1730 (380 ± 50 cal. BP) from Thailand 31 . Considering its wide-spread effect we suggest that this event was triggered along the Andaman Segment and was comparatively larger than CE 1881 (Event II), inflicting wider effect in the Indian Ocean. Event V (Unit-l) occurred after CE 1305-1420 and before CE 1510-1632 (Figs. 2b, 6 and 7), which correlates with the CE 1300-1400 tsunami reported from Phra Thong, Thailand 12,13,33 and CE 1290-1400 tsunami from Aceh, Indonesia 8 . This event may also be correlated with the CE 1120-1300 tsunami event reported from the west coast of Andaman 3 , and with turbidites found from Sumatra (T3: 630 ± 110 cal. BP; CE 1159-1480) 14 . Also, signatures of subsidence and tsunami deposit during CE 1040-1495 have been reported from the south Andaman, as well as an uplift from Hut Bay and north Andaman 11,12 . This was a mega earthquake sourced locally along the Andaman segment and resulted in a transoceanic tsunami 3 . Event VI (Unit-n) occurred after BCE 2660-2100 and before CE 428-600 (Figs. 2b, 6 and 7, Tables 1 and 2; Supplementary Table S1). Further considering the All elements show a comparatively higher concentration in the older tsunami deposits (Units l, n, and p). Whereas, all the adjacent non-tsunami deposits, i.e. Units a, c, e, f, h, i, j, k, m, and q show medium to high concentration of the elements like Si, Al, Fe, Ti, and Mg, and thus, are indicative of terrigenous deposits. Also, the overall depletion of these elements within Units b, d, g, l, n, and p characterize tsunamigenic origin of these deposits. The Ca and Mn enrichment is a characteristic signature indicating marine origin of these sediments. Though Na and K ions constitute a significant part of the seawater, their depletion in the tsunami deposits is suggestive of chemical alteration through ion-exchange due to leaching and prolonged burial. www.nature.com/scientificreports www.nature.com/scientificreports/ Figure 6. Modelled calendar ages and probability distributions of stratigraphy exposed in geoslices and trenches excavated along WNW-ESE transect at Badabalu. The ages are calculated and modeled using OxCal version 4.2.4 (Reimer et al. 22 ) and IntCal13 atmospheric curves (Reimer et al. 23 ). We infer at least eight events (earthquakes/tsunamis) those occurred in last 8000 yrs. Event I represents 2004 tsunami; Event II was around AD 1777-1883, could be correlated with AD 1881 earthquake and tsunami from Car Nicobar. Event III occurred during AD 1674-1821, could be correlated with AD 1762 earthquake/tsunami reported from Arakan Suduction Zone. Event IV was around AD 1485-1610, correlated with AD 1672 reported from Andaman Island. Event V was around AD 1325-1434, correlated with AD 1300-1400 earthquake and tsunami from Andaman, Thailand and Indonesia. Event VI was around BC 2480-2060, could be correlated with a tsunami reported from southeast Sri Lanka that occurred during BC 2000-3000. Event VII occurred during BC 2966-2286, correlated with tsunami event that occurred during BCE 2810-3200 reported from southeast Sri Lanka, and also with the event of BCE 2892-1895 reported from Indonesia. Event VIII occurred before BC 5600, correlated www.nature.com/scientificreports www.nature.com/scientificreports/ youngest age of detrital charcoal from Unit-n (BCE 2024-1885), we suggest that Event VI occurred after BCE 2024-1885. This event may be correlated with turbidites observed in a core (T26: 3720 ± 340; BCE 3095-1290) from Sumatra 14 , and also with the tsunami of BCE 2000-3000 reported from Peraliya, Sir Lanka 34,35 . We infer that Event VI was a local event produced by Andaman Segment, generated a tsunami that reached the eastern coast of India and Sri Lanka. Event VII (Unit-p) was after BCE 3086-2758 and before BCE 2661-2100, during BCE 2966-2286 (Figs. 2b, 6 and 7, Tables 1 and 2). Due to a wider age bracket it is difficult to correlate this event with a particular event reported from other adjoining areas in the Indian Ocean. Nevertheless, it may be correlated with the BCE 2810-3200 (4760-5150 cal. BP) tsunami reported from Karagan Lagoon, southeast Sri Lanka 34,35 , and also with turbidites reported from Sumatra (T27: 3900 ± 190 cal. BP; BCE 2892-1895) 14 . Since we do not have any lower limit bracketing Event VIII, we suggest that Event VIII (Unit-r) occurred before BCE 5612-5323 (Figs. 2b, 6 and 7, Tables 1 and 2; Supplementary Table S1). This event can be correlated with turbidite identified from Sumatra during (T43: 6600 ± 140 cal. BP; BCE 5786-5301) 14 , and with the BCE 5374-5579 (7324-7529 cal. BP) tsunami event reported from Indonesia 36 . Because we found this tsunami deposit in only one geoslice (Fig. 2b), it is difficult to ascertain if this was a local (Andaman segment) or a distant sourced event.
Based on the stratigraphic record, OSL and 14 C AMS ages, and modelled ages in OxCal, we observed a considerable depositional gap for almost 2000 year between 3700 and 1500 years BP (Figs. 2b, 6 and 7, Supplementary Data S4, Table S4.1-S4.2). This discontinuous stratigraphic record could be attributed to erosion due to one of the possibilities: (a) coseismic uplift or gradual uplift during inter-seismic period along the up-dip portion of the subducting plate or upper plate fault. But no upper plate fault from this region has been reported; (b) Relative Sea Level (RSL) fall which accelerated erosion of the stratigraphic sequence. The chronology of the beach ridges and reconstruction of complex pattern of shoreline progradation and erosion from Phra Thong, Thailand suggest a short episode of local erosion between 4000 and 3800 yr BP, could be attributed to climate change, impact of a tsunami or tropical cyclone 37 . Further, Brill et al. 37 also reported a signature of sea-level fall and shoreline progradation from Phra Thong with decreased rate of <1 m/year during 3300-3500 yr BP. Dura et al. 38 reported an incomplete record of subduction zone earthquakes in coastal stratigraphy from the coast of Sumatra, because the late Holocene (last 4 ka) RSL or sea-level fall facilitated erosion and restricting preservation of lithounits. Following these arguments, we suggest that erosion was accelerated due to RSL fall at ~3500 BP in Andaman and Nicobar Islands. The continuous stratigraphic sequence from 1500 years BP till present is attributed to a gradual RSL rise, which remained within the tidal frame of 1-2 m in the Andaman region.
The lower portion of the stratigraphic section comprising Units l to r reveals stacked sequence of peaty units (wetland soils; Units m, o and q) and tsunami deposits (Figs. 6 and 7). The Unit-k comprising fine-medium silty-sand suggests basin-fill under sub-tidal condition followed by a gradual uplift during inter-seismic period. The peaty soil (Unit-j) indicates the formation of wetland soil at or above mean sea-level. This was followed by a coseismic subsidence, and a long-term post-seismic subsidence is well justified by the presence of a 75 cm thick very fine sand (Unit-i). The presence of Unit-h (peaty soil) suggests that the area was at or above mean sea-level. We infer that the area emerged from deeper environment (sub-tidal) around this time.
The upper portion of the stratigraphic section, Unit-b marks the 2004 tsunami, Units d and g represents tsunamis during the recent historic time . Unit-f with fine sand suggests that the area was under the influence of sub-tidal environment, whereas Units e and c (peaty soils) indicate that the area was at or above mean sea-level. Possibly the area experienced subsidence during these earthquakes and recovered during post-seismic period, which eventually facilitated the formation of wetland soils and vegetation growth. The area remained submerged for substantially longer span during inter-seismic period as indicated by a thick fine silty-sand (Unit-i) (Figs. 6 and 7). A long-term inter-seismic subsidence implies a huge strain accumulation. However, couple of large magnitude earthquakes viz. CE 1881, with a rupture near Car Nicobar in the mid-segment of Andaman, and CE 1762, ruptured along the Arakan Subduction Zone, partially released the long-term accumulated strain after CE 1679 event. The CE 1679 event was a local event having its rupture along the Andaman Island. Hence, we conclude that the Andaman Segment has enough accumulated strain to trigger a mega-tsunamigenic subduction zone earthquake in near future. A 2000 years stratigraphic gap add to the uncertainty associated with the estimation of the recurrence of tsunamigenic earthquakes. However, 1500 years of continuous sequence suggests a recurrence of 420-750 years for a mega-earthquakes along subduction zone like the 660-880 CE 3 , 1300-1400 CE and the 2004 Sumatra Andaman earthquake having different source. A shorter interval of 80-120 years is inferred for the large earthquakes like 1679, 1762 and 1881 CE.

Methods
Google Earth images (pre and post 2004 earthquake) were used to identify the location that experienced land-level change and having a shoreline configuration with beach-ridge-swale topography, which are ideal for the preservation of tsunami deposits (Supplementary Figs. S1.1a-h and S1.2a,b). A detailed topographic survey using Total Station was conducted transverse to the shoreline along the WNW-ESE transect (Figs. 1d and 2a).
Four sediment samples were dated by Optical Stimulated Luminescence (OSL) dating technique at IIT Kanpur, and 18 samples were dated for 14 C (AMS) ages at Beta Analytic, USA, as well as at Inter University Accelerator Centre (IUAC), New Delhi (Supplementary Data S4; Figs. S4.1-S4.3). We collected sediment samples from the exposed trenches as well as geoslices obtained from Badabalu site (Figs. S2.1-S2.11; Table S4.2). For paleodose measurement, samples were treated with 1 N HCl for one hour followed by washing the sample at least www.nature.com/scientificreports www.nature.com/scientificreports/ three times with de-ionized water. It was followed by treatment with 30% H 2 O 2 until all the effervescence disappeared, and washed again with de-ionized water. This is done to get rid of carbonates and organic matter from the sediments. Dried samples were then sieved to obtain 90-212 um grain fractions of which only 90-125 um fraction size was used for further analysis. The quartz and feldspar were isolated with the help of Frantz magnetic separator with constant current of 1.50A. Then the isolated quartz was etched with 40% Hydrofluoric acid (HF) solution for 60 minutes to remove outer alpha skin and dissolve any leftover feldspar. The isolated quartz was then rinsed with HCL to get rid of any fluorite precipitate from HF acid. After drying, the sample was re-sieved to remove <90 um to acquire fine pure quartz grains. These grains were then mounted on 9.8 mm diameter stainless steel aliquots with the help of silicon spray. All the processing was carried out in the laboratory controlled red light environment. For the paleodose determination, Riso TL/OSL reader with an EMI 9635Q photomultiplier   Table S5.1). Samples processing and measurements were carried out at Beta Analytics, USA and IUAC, New Delhi. "Calibrated" or calendar ages were calculated using "CALIB rev 5.01" and calibration curves (IntCal04, Reimer et al. 22,23 ).