JOURNAL GEOLOGICAL SOCIETY OF INDIA Vo1.50, Sept. 1997, pp. 283-288

A 2000-Year Palaeoflood Record from Sakarghat on Narmada, Central India VISHWAS S. W E 1 , SHEILAMISHRA~ AND VICTORR. BAKER3 'Department of Geography, University of Pune, Pune 411 007, 2Depanment of Archaeology, Deccan College, Pune 411 006, Qepartment of Hydrology and Water Resources, University of Arizona, Tucson AZ 85721, USA

Abstract: A continuous record of the largest Nannada floods over the last 2000 years was obtained from Sakarghat, on the Narmada River from a study of slackwater flood deposits. Two sequences of extreme floods date between ca. 400 and 1000 AD and post-1900 AD. The period, 400-1000 AD representing a period of less frequent but more extreme floods, has been dbcumented in the archaeological record as one of decline of human settlements.

INTRODZlCTION In recent years historical and palaeo-flood data hme become an important source of information for establishing the magnitudeand frequency of extreme floods that have occurred prior M the instrumental period (Baker, 1988). In contrast to modern gauge records the sedimentological records preserve evidence of extreme flood events on the time scale of hundreds q ~ thousands d of years. In appropriate geqmorphic settings, stratigraphic records of She largest floods are selectively presemed as slackwater flood deposits (SWD) in stable bedrock canyons. These sediments are fine-&rained sand and silt that fall rapidly out of swpension during large floods in protected ere flow velocity is markedly reduced er, 688). The upper layers of approximate the actual stage of the floQdpeak and provide evidence of only e x m e floods (Baker and Kochel, 1988). Tkh~tarymouths are the most common site for the accumulation of SWDs. The ages of tbe deposits are determined by radiometric dating of associated organic or archaeologicalmaterial. TheNarmada River between Handia and Omkareshwar is entrenched into resistant Proterozoicrocks of the Vindhyan Supergroup. The reach is highly suited to the accumulation and preservation of slack-waterdeposits. During

.

large monsoon floods SWDs are deposited at &gh-levels along channel margins and backflooded tributary mouths. The SWDs on the Narmada River were first reported by Baker (1988) and since then have been s W in some detail in Chahin Nala (Kale et al. 1994; Ely et al. 1996) where they date back to 1700 years. In this paper we report on the slackwater palaeoflood record at Sakarghat near Punasa, where a continuous2 Kyrs record is preserved. PUNASA GORGE: 3NTERESTING SWD SECTION

The channel of the Narmada River through the Punasa Gorge has an highly irregular bed topography with deep pools,rocky islands, rapids and scablands (Fig. 1). The majority of SWDs (thickness 3 to > 1 h ) occur as benches parallel to the channel or "gt the mouth of tributaries: An interesting SWDsection with multiple flood layers wzp found at Sakarghat, at the junction of an unnamed tributary and the Narmada Riverjust upstream of BajrikundNala (Fig. 1). The section was exposed by removal of material for brick making prior to our visit. The deposits form a large bench about 21 meters above the low water level. The section is signiticant because it reveals the inner core of

0016-762Y97-50-3-283/$ 1.00 O GEOL. SOC. INDIA

284

VISHWAS S. KALE AND OTHERS

FTg.1. Geomorphic map of the Punasa Gorge, showing the location of the Sakarghat palaeoflood site.

the SWD,where a continuous sequence of floods spanning the last 2000 years is seen in

one vertically-stacked sequence. Detailed observations on the exposedpart of the section

2000-YEAR PALAEOFLOOD RECORD, SAKARGHAT, CENTRAL INDIA

0

IRON SLAG IRON SLAG CHARCOAL CHARCOAL HEARTH-1

100

285

- 690545 YRS

BP (AD 1295 Cal YR)

CHARCOAL 200

30 0

CARBONATE FILAM ENTS DIFFUSED CHARCOAL

400

- l938f 53 YRS BP (?)

CHARCoAt AT BOUNDARY 1 -HEARTH - 2

19 500

- 1661256 YRS

BP (AD 411 Cal YR) BURNT LAY ERlANGULAR PEBBLES l 9 2 3 + 6 8 YRS BP HEARTH -3 (AD 82 Cal YR)

-

60 0

-

700 0 -4

30

60 '1. SAND

90

Fi2.2. Stratigraphy of the Namiada River slackwater flood deposits at Sakarghat. 1 to 26 = flood units; Cal YR = Calibrated to calendar years.

were made after digging a step trench into the top 6-7 m. The sequence rests on compact silty-sandy deposits. The flood units fine upwards and the contrast between the fine top of one unit with the coarse bottom of the next unit is the major method of distinguishing between units. Occasionally charcoal, which floats and is the last to be deposited, marks the top of a flood unit. Human activity in between floods may be seen in the form of hearths, rubble

and iron slag. About 26 flood units (3 to 79 cm thick) were distinguished based on these criteria (Fig. 2). The units of 12 and 18 have undergone incipient pedogenesis, which has probably obliterated the distinction between sub-units. The site permits tracing of the individual flood deposits laterally, but the units are not completely horizontal. The lower units in particular are dipping upstream (of the

2000-YEAR PALAEOFLOOD RECORD, SAKARGHAT, CENTRAL INDIA

IRON SLAG IRON SLAG CHARCOAL CHARCOAL HEARTH-1

- 6 9 0 i 4 5 YRS

285

BP

(AD 1295 Cal YR)

CHARCOAL

CARBONATE FILAMENTS DIFFUSED CHARCOAL

- l 9 3 8 f 53 YRS BP(?)

CHARCOAL AT BOUNDARY HEARTH - 2

19

- 1661556 YRS BP (AD 411 Cal YR)

BURNT LAY ERlANGULAR PEBBLES -HEARTH -3 - 1923268 YRS BP (AD 82 Cal YR)

0

30

60 ,!* SAND

90

Flg.2. Stratigraphy of the Narn~adaRiver slackwater flood deposits at Sakarghat. 1 Calibrated to calel~daryears.

were made after digging a step trench into the top 6-7 m. The sequence rests on compact silty-sandy deposits. The flood units fine upwards and the contrast between the fine top of one unit with the coarse bottom of the next unit is the major method of distinguishing between units. Occasionally charcoal, which *floatsand is the last to be deposited, marks the top of a flood unit. Human activity in between tloods may be seen in the form of hearths, rubble

to

26 = flood units; Cal YR =

and iron slag. About 26 flood units (3 to 79 cm thick) were distinguished based on these criteria (Fig. 2). The units of 12 and 18 have undergone incipient pedogenesis, which has probably obliterated the distinction between sub-units. The site permits tracing of the individual flood deposits laterally, but the units are not completely horizontal. The lower units in particular are dipping upstream (of the

Table L Narmada Palaeoflood deposits at Sakarghat Series

Flood unit

Thickness in cm

Munsell colour

Percent sand

Remarks

I

1 2 3

0-3 4-7 9-21

lOYR 4t3 lOYR 5t3 lOYR 414

67.0 76.0 64.0

slag between unit 1 & 2 slag between unit 2 & 3 slag between unit 3 & 4

I1

4 5 6

21-3i 31-51 51-80

lOYR 3/3 lOYR 4/3 1OYR 4/3

65.5 69.0 73.5

charcoal at base charcoal at base health-1 at 80 (69Oi45 Yrs BP; Cal Yrs AD 1295) change in texture at 102 charcoal at base compact silty-sand silty-sand silty-sand coarse white sand; charcoal at 248

lOYR 3/3 70.0 lOYR 414 62.5 lOYR 4/3 58.5 lOYR 413 67.0 10YR 4/3 75.0 1OYR 414 88.0 ................................................................................ Major break ....................................................................................... 12a 248-298 1OYR 414 69.5 compact silty-sand 12b 298-327 lOYR 314 58.0 development of calaete filaments1 diffused charcoaVburnt earth 13 327-353 lOYR 414 75.0 loose sand; bioturbated some discon tinuous dark layers 1OYR 414 73.0 silty-sand 14 353-369 10YR 414 70.5 charcoal at 387 15 369-396 10YR 514 91.5 loose sand 16 398-405 76.0 silty-sand lOYR 414 17 405-415 ................................................................................ Major break ....................................................................................... lV 18 415-463 1OYR 4/3 71.5 carbonate filaments at top part of the unit 19 463-489 lOYR 4/3 74.5 charcoal at 18/19 unit (1938i53 Yrs BP?) 1OYR 414 63.5 unit covers hearth 20 489-510 lOYR 4t3 41.0 hearth-2 in unit (166146 Yrs BP; Cal 21a 510-523 Yrs AD 411) 10YR 414 76.5 sand 21b 523-537 lOYR 4/3 56.0 burnt layerlangular pebbles 22 537-562 lOYR 4t3 56.5 upward fining sequence 23 562-585 lOYR 4/3 50.0 hearth-3 in the unit (1923i68 Yrs BP; 24 585-601 Cal Yrs AD 82) 63.5 sand 25 601-615 lOYR 4/3 59.5 carbonate filaments at top 26 615-637 1OYR 414 7 8 9a 9b 10 11

80-102 102-125 125-140 140-166 166-217 217-248

m

Cal = Calibrated to calendar years using Stuiver and Reimer's computer program (1993). C-14 dates given by NSF AMS facility, University of Arizona, Tucson, USA.

tributary) and probably drape the preexisting surfwe of older deposits. The flood deposits are dominated by sand which varies from 40 to 90% with most units between 55 and 75% sand (Table I). The sediments are characterized by strongly fine-skewed, poorly sorted, fine sands and coarse silts (Mz = 2-4.5 $). Recent deposits (units 1 to 3) and flood unit 11 and 16 (Table I)

are coarsest indicating higher flood velocities (Kochel and Baker, 1988). The SWD units are generally structureless, implying rapid depositions from suspension (Kochel and Baker, 1988). l k o major breaks are recorded in the. section dter unit 18 and again after unit 12 (Table I). These units reveal evidence of incipient pedogenesis and indicate a time gap

between these and the overlying units. At these levels the pores in the flood deposits have been coated with calcium carbonate. I6 unit 18 this feature extends about 20 cm from the top, and in unit 12 it is about 50 cm thick. Based on the depth of the calcrete filament impregnation, the break after unit 12 appears to be the most prominent feature recorded in the section. Human activity is intense in flood units below unit 18, with two hearths and a burned layer, and is also present above unit 12 in the form of a hearth between units 6 and 7. The top three flood units are separated by iron slag. The sequence of coarser flood units between 12 and 18 have no trace of human activity. Dates have been obtained from the three hearths and from floating charcoal from the section (Table I). While the dates from the hearths are stratigraphicallyconsistent, the date from the floating charcoal occurring at unit 18/19 is older than either of the two dates obtained below it. We therefore, interpret this date as representing older charcoal incorporated into a younger flood unit. Two dates below flood unit 18 bracket the lower series of flood units to between the beginning of the Christian era and c a 400 AD. Unfortunately, we do not have any dates from the units between 18 and 12. The hearth above unit 7 has given us a date within the 13th century, placing the units to the Mediaeval period (ca. 1000-1400AD), while thedates from the lower unit (below 18) can be bracketed to the early Christian era (ca. 0-400 AD). In spite of the lack of dates from the units 12-18, logically they must belong to the intervening period which would be approximately 400-1000 AD.

MAJOR FLOOD SERIES The Sakarghat section, therefore, can be divided into four distinct series of flood units (Table I). The top three flood units which are separated by layers of iron slag form the first sequence, and probably are equivalent to the

coarse sand deposits seen in the Chahin Nala SWDs which date to the present century (Ely et al. 1996). The sequence of flood units above the prominent break represented by calcrete accumulation in unit 12 represents a series of similar flood deposits with no major breaks in between. The two calcrete impregnated horizons and the relatively coarser SWDs between unit 18 and 12 represent a contrast to both the upper flood sequence and the one below. The flood units below unit 18are similar to those above unit 12. Kochel and Baker (1988) have shown that the thickness and grain size of slackwater sediments is directly proportional to the flood magnitude at tributary mouths i.e. thicker and/ or coarser units represent high floods with higher flow velocities. Therefore, the units of uniform SWDs, below unit 18 and above unit 12, can be interpreted to represent periods of frequent, moderate floods. The sequence of coarser SWDs, between the two breaks in the sequence (Table I) can be interpreted as representing aperiod of less frequent, but more extreme floods between ca. 400 and 1000 AD. The topmost flood layers, similarly represent the reoccurrence of large floods after a break of 3-4 centuries (Ely et al. 1996).

FLOOD MAGNITUDE Hydraulic modelling of the discharges indicates that the highest SWDs (post-1900) in this reach were ernplaced by floods that were close to 60,000 m3/s (Kale and Mishra, 1994). Since at Sakarghatmodem flood depositi cover older SWDs, it is obvious that floqds observed in the present century were larger than the palaeofloods (Kale, et al. 1994).Extreme floods in this century were invariably associated with monsoon depressionsand cyclones originating over the Bay of Bengal (Kale and Mishra, 1994), it is therefore reasonable to infer that the SWDs provide evidence of individual intense monsoon storms during the historical period.

288

VISHWAS S. KALE AND OTHERS

DISCUSSION AND CONCIJJSION Recently, Sharma (1987), compiled the data from archaeological excavations in northern India. In the 140 excavated sites, he found that the period 300-1000 one of decay of urban features of settlements. Coinage almost disappears, honumental architecture declines, some of the settlementsare abandoned. Mate (1990) has also pointed out the scarcity of urban settlements in the same period in Maharashtra. Dhavalikar (1993) suggests that this urban decay could have been caused by a period of increased frequency of droughts. The coincidence of the dating for this period of reduced number and more extreme floods at Sakarghat with a period of archaeologically documented decline in human settlements is extremely interesting. In fact even at Sakarghat itself, the evidencefor human activity is greater in theperiod0-400 AD and after 1000AD. It is also interesting to note that some of the largest floods in the last 100years were associated with

was

E l Nifio conditions and periods of rainfall deficiency (Kale and Mishra, 1994). The Sakarghat section therefore does provide some support for the hypothesis that environmental factors might have been different in the period 400-1000 AD, and had an adverse impact on human societies. The above dficussion implies that slackwater sediments have great potential to complement the short-term gauge records and can provide an important input for hydrologic and climatic models aimed at predicting the behaviour of floods as well as the monsoons.

ACKNOWLEDGEMENTS This research was supported by the Department of Science and Technology, New Delhi Grant No. ESSlCAlA3-04/90. The authors thank S.N. Rajaguru, Yehouda Enzel and Lisa Ely for tlieir helpful comments.

References BAKER, V.R (1988). Paleofloodhydrology of tropicalrivers. Geomorphic and hydrologic aqpects of monsoon floods Proceedings, International Seminar on Hydrology of on Narmada and Tapi Rivers, central India. Extremes, Roorkee, pp. 1-10. BAKER, V.R. a n d K o m , KC.(1988): ~ o o sedimentatioz d K in bedrock fluvial system. In: Hood Geomorphology. V.R. Baker, R.C. Kochel, and P.C. Patton (Eds.). J ~ h n Wiley, New York, pp.123-137. M.K.(1993). The Second De-urbanization. K DHAVALIKAR, The Indian Historical Review, v.16, pp.211-217. ELY, L. L., ENZEL, Y., BAKER,V.R., KALE, YS. and MISHRA, S. (1996). Palaeoflood evidence of changes in the Wiley, New York, pp.357-376. MATE, MS. (1990). The clay feet? The Bulletin of the magnitude and frequency of monsoon floods on the Narmada River, central India. Bull. Geol. Soc. Am., Deccan College Research Institute, v.49, pp.243-250. v.108, pp.1134-1148. SHARMA, R.S. (1987). Urban decay inIndia(3WlOOOAD). KALE, V.S.,RY, LL., EN=, Y.and BAKER, V.R (3994). ~unshirambanoharlalPublishers, New Delhi. (Received: 13 February, 1997; Accepted: 22 March, 1997)

.