Friday, February 10, 2012

Yamuna And Sutlej Stopped Flowing Into Ghaggar / Sarasvati By Early Holocene

The Yamuna stopped flowing into the Ghaggar / Sarasvati and shifted course eastwards into the Ganga as early as around fifty thousand years ago. The Beas and the Sutlej stopped flowing into the Ghaggar / Sarasvati and joined the Indus before ten thousand years ago, several thousand years before the beginnings of the Harappan civilization.

That is the conclusion reached in a paper in Geology  (behind paywall) by Peter Clift and colleagues using U-Pb (Uranium - Lead) dating of zircon crystals from ancient channels and alluvium of the Ghaggar / Hakra river.

I had earlier suggested that the Ghaggar likely never had any glacial connection. That statement now needs to be modified to read that there is a strong possibility that the Yamuna flowed into the Ghaggar in the Pleistocene and the Sutlej until the early Holocene.

Detailed mapping of the region between the Indus and the Yamuna where the present day Ghaggar / Hakra flows  has revealed the presence of many dried up channels suggesting that a much larger Ghaggar river existed in the past. The Harappan urban civilization declined and was abandoned by around 2000 B.C to 1800 B.C. One of the main reasons given is a prolonged drying of the region which made agriculture unsustainable. Monsoonal strength over this region has fluctuated for the last ten's of thousands of years and evidence from many different sources indicates that this region began experiencing aridity by mid Holocene and this arid phase moved eastwards over time slowing making urban centers and agriculture unsustainable. The larger dried up channels point to this climate change.

Another explanation of many of these dried up channels is the hypothesis that large glacially fed rivers like the Yamuna and Sutlej once flowed into the Ghaggar. They shifted course away from the Ghaggar  in the late Holocene, dramatically reducing water supply to the Harappan urban centers and contributing along with climate change to their decline and eventual collapse.

How does one figure out whether ancient channels represent glacially fed rivers and match them to known rivers? Clift and colleagues use a conceptually simple yet technologically challenging geochemical technique to make this match between ancient channels, present day rivers and their source in the Himalayas.

The Himalayan mountains are made up of several geological terrains of different ages. The Karakorum and Kohistan terrain (Trans -Himalayas) are younger than 300 million years old, the Tethyan Himalayas are 300 -750 million years old, the Greater Himalayas are 750 - 1250 million years old and the Lesser Himalayas are 1500 -2300 million years old. The ages of these terrains have been well characterized over the decades using a number of different radioactive clocks.

For this study the U-Pb clock contained within zircon crystals was selected. Zircon (zirconium silicate) is a hard mineral which occurs in granites and metamorphic rocks which make up large portions of these terrains. It can survive physical attrition during long transport by streams. It is also chemically quite stable and there is less chance of chemical weathering leaching out U or Pb from the crystal. Another advantage is that there are two clocks inside zircon. One through the decay of U-235 to lead-207 and another through the decay of U-238 to lead-206. So there is an inbuilt crosscheck on your results.

Why not just use the types of minerals to fingerprint the source rocks? That method does work is certain contexts but geological terrains of different ages can contain similar suite of minerals making discrimination based on a unique signal difficult.

Below is a map of the study area.

Source: Clift 2012

Black dots represent trenches. Stars represent drill sites and white squares represent samples from modern rivers. Dotted lines are proposed courses of ancient Yamuna and Sutlej. U-Pb dates were calculated from zircons sampled from the Indus, Ghaggar,  Beas, Sutlej and the Yamuna. The idea is that the headwaters of each of these rivers would be eroding sediment predominantly from different geological  terrains and therefore if enough zircon crystals were analyzed it would be possible to identify a geological age signal that is unique to each river. For that though large number of crystals need to be dated for a statistically robust result.

For this study more than hundred grains per sample were analyzed using a scary sounding instrument- the Laser Ablation Inductively Coupled Plasma Mass Spectrometer.  It was found that individual rivers do cluster at different ages i.e. they contain a population of grains within a distinct age range and could be discriminated from each other. For example the Beas showed a major population of zircon grains of age 300 -750 million years ago implying that its headwaters drained mostly the Tethyan Himalayas. Similarly the  the Yamuna sample contained a major population which clustered around 1875 million years. The Sutlej has two populations clusters, one at 750 -1000 million years ago and another at around 1830 million years or so.

The next step was to find out whether ancient sediments from dried up channels associated with the Ghaggar river contained zircon populations with age clusters that could be matched to modern rivers. If they did, then that would mean that that particular river was flowing in that channel in the past. And carbon-14  or Optical Stimulated Luminesence (OSL) dating of the sediment could tell when. So there are two kinds of dating techniques used here. The U-Pb technique dates the geological age of the rocks in the Himalayan headwaters of the rivers. The carbon-14 and OSL techniques date the timing of deposition of sediments in the various channels.

Channel and overbank sediments close to Harappan archaeological sites along the northern edge of the Thar desert were sampled. The sediments were dated using a combination of 14C which dates organic matter and OSL, a technique that is used to calculate the time since the sediment was last exposed to sunlight.

Analysis showed that ancient sediments did contain populations of zircons of distinct age ranges that could be matched to the Yamuna, Sutlej and the Beas implying that these rivers were flowing through these channels sometime in the past. Constrained by carbon-14 and OSL dates of the sediment, the patterns indicate that the Yamuna signal was lost around fifty thousand years ago implying a change of course of the Yamuna eastwards towards the Ganga at that time. The Beas and Sutlej signal was lost prior to ten thousand years ago, these two rivers migrating north-northwestwards and joining the Indus.

All this sampling is in Cholistan, the portion of the Ghaggar/Hakra in Pakistan. Some studies has mentioned ancient channels interpreted as the Sutlej joining the Ghaggar in Haryana in India, upstream of these sites. These channels were not sampled in this study. However, the site at Fort Abbas samples the main Ghaggar channel. Any signal of the Yamuna or Sutlej  joining upstream would have shown up here. The results indicate no signal from these rivers in the oldest sediments which are five thousand seven hundred years old or so.

If possible, these analysis could be extended in the future to sample more extensively the dried up large channels upstream in India too for a more expansive analysis. Over the last year some more studies (preliminary results presented at last year's AGU meeting) have shown an earlier episode of the Ghaggar drying up in the latest Pleistocene that coincides with a regional climate shift towards drier conditions. Did tectonically driven avulsion coincide with this climate change?

On the other hand, a paper in the GSA Bulletin by Sinha et. al. 2009. suggest a younger Yamuna, meaning that the Yamuna has been flowing in its present course for only the last 2-3 thousand years and may have been flowing westwards towards the Ghaggar before that. This study was based on detrital grains provenance studies and OSL dating of the Yamuna and Chambal sediments. I found the provenance analysis itself to be sound but the young Yamuna interpretation questionable, due to lack of sampling upstream of the cratonic rivers confluence and just one OSL date from the present day Yamuna channel.

So,  there are still answers to be found regarding the interplay of tectonics, climate and the Pleistocene and Holocene geography of these rivers.

Having said that, this is one of the more convincing studies of Ghaggar sediment and water provenance that I have come across so far.  Many earlier studies have suggested a much later river avulsion scenario wherein the Yamuna and the Sutlej change course as late as 2000 B.C. My impression was that these studies lacked the kind of hard evidence needed to pinpoint the dates of the deposition and the origin of the sediments. They relied mainly on the size of the channels without demonstrating physical continuity with the large glacial rivers,  or the presence of certain metamorphic grains and pebbles. Both these parameters are open to multiple interpretations and are not convincing of a high Himalayan source by themselves.

This study by Clift and colleagues along with some earlier isotope dating and provenance work on Indus and Ghaggar sediments by some of the authors and independent work on oxygen isotopes of ancient Ghaggar /Hakra channel water are strong results in favor of the Yamuna and Sutlej leaving the Ghaggar river thousands of years before the beginnings of the Harappan civilization.

The results also imply that these two rivers most likely stopped flowing into the Ghaggar thousands of years before the presence of Aryans in this region.

This is a touchy topic and it deals with whether the Aryans migrated into northwest India from central Asia around 1500 B.C. or so, bringing with them cultural habits and religious ideas that evolved into the Vedic Culture, or whether the Aryans were indigenous to this part since times immemorial. The Ghaggar has been equated with the river Sarasvati mentioned in the Rig-Veda by supporters of the indigenous Aryan theory.

It is described as a mighty river flowing from the high Himalayas. This in turn has been interpreted to mean that the Sarasvati would have been flowing out of glaciers for it to be mighty, and therefore the Aryans must have been present in northwest India before 2500 - 3000 B.C., because there is evidence that since then the Ghaggar became a smaller river.  This argument is then taken further to claim that the Aryans built the Harappan civilization. In this scenario the glacial rivers change course only after 2500 - 2000 B.C or so and the resulting water crisis forces the "Vedic" Harappans to disperse eastwards towards the Gangetic basin.

I have stressed that this attempt to link a hypothesis of a mighty Sarasvati to the presence of Aryans is misguided and one that has caused harm to the public understanding of the topic and to what constitutes good science. Many geologists and archaeologists accepted the validity of a glacial Sarasvati without critically weighing the evidence. Taking their cue, in web forums and books, supporters of a glacial Sarasvati have popularized the hypothesis of a late river avulsion and often presented it as irrefutable evidence favoring the indigenous Aryan theory.

I have commented on this earlier in Pragati and on my blog (here and here ) and suggested that evidence at that time did not support a late avulsion and further that this issue of the timing of Aryan presence in this region doesn't really depend on  glacial rivers flowing into the Ghaggar. Rivers can be mythologized and worshiped whether they are big or small. The Aryans could just as well have considered holy a Siwalik fed river and exaggerated its size in their hymns.

Linguistics, cultural evidence, archaeology and perhaps in the future higher resolution genetic data are better placed to answer the question of Aryan origins. Unfortunately, during this saga of the search for the glacial Sarasvati some very questionable geological scenarios have been put forth as being definitive. Perhaps the time has come for supporters of that theory to at the very least start being more cautious when promoting the late river avulsion theory.

Meanwhile, whether there were glacial rivers connecting the Ghaggar during Harappan times is not just an interesting geological question but has implications in understanding  Harappan water use and agriculture patterns. Head over to Dorian Fuller's (one of the authors of this study) blog The Archaeobotanist for some comments.

Clift, P., Carter, A., Giosan, L., Durcan, J., Duller, G., Macklin, M., Alizai, A., Tabrez, A., Danish, M., VanLaningham, S., & Fuller, D. (2012). U-Pb zircon dating evidence for a Pleistocene Sarasvati River and capture of the Yamuna River Geology DOI: 10.1130/G32840.1

Also see additional posts on this topic - 1) New Geomorphological Work on Ghaggar
2) K.S. Valdiya on the glacial Saraswati in Current Science



  1. Interesting to find that the hypotheses on Saraswati continues to shift courses as well.

    My comment relates to a point you raise in your paper: "...Aryans must have been present in northwest India before 2500 - 3000 B.C., ..."

    Yes, some believe in that theory. Two points I would like to make are: One, the Mohenjodaro-Harappan civilizations are much older than the Vedas. This indigenous civilization need not be Vedic in origin. Two, that such a vast civilization flourished MUST point to the existence of a large perennial source of water in that region without which a sustained civilization could not have emerged for thousands of years before drying up of the region forced migration out of that region. Isn't that a fairly logical hypothesis?

  2. Viswag- the sentence you quote is part of a para where I describe the indigenous Aryan theory for the readers benefit.

    I agree with you that Harappan civilization is older than Vedic civilization and I have pointed out that in this post and in previous posts that the climate was wetter in the mid Holocene and that would have meant a year round source of water in the Ghaggar.

  3. Suvrat Kher I am in the process of researching the manner in which linguistics has muddied the picture in the same way the way you feel that non geologists have misinterpreted geologic data. Since you are a geologist and not a Vedic scholar, would you be able to comment on the assertion that the location of the Ghaggar river corresponds to the location of the mythical Saraswati. The issue of "rain-fed" or "glacial melt" is unrelated to this.

    The "Aryan Invasion" has long since been disproved by recent genetics papers so that issue can be left out as a reason for controversy.


  4. Shiv- the location of the Ghaggar does seem to be a reasonable match for descriptions of the location of a Saraswati river in ancient literature. Though that in my opinion does not answer the question of Aryan origins.

  5. Suvrat,

    I am no expert in geology nor in linguistics...I just want to say, your analysis is good but isn't there a problem that world's oldest civilization had no rich literature whereas Aryan nomads had one of the richest litetature.

    Genitic studies have also pointed towards an Indian origin of Vedic people. Also Indian learning system was mainly vocal and not written..means knowledge sharing was through vocal medium in ancient times. So in terms of history, i think we need a bit of re-writing as present history is heavily leftist/macaulay inspired. As for geological proof, I sincerely pray that you do the best possible study and further elighten us about true scenario


  6. Bhargav- thanks for reading.. regarding your statement that Genetic studies have pointed towards an Indian origin of Vedic people.. I am not so sure.. please see this post which points to some recent papers and has a commentary on the state of the science.. there is genetic evidence of waves of admixture of Eurasian and Indian populations dating to around 2000 B.C. or so ..