Tuesday, April 8, 2008

Is the Saraswati Still Flowing Underground

[ Update: Please refer to these posts for the latest results on the river Ghaggar

1) Yamuna and Sutlej stopped flowing into the Ghaggar by Early Holocene
2) Fluvial History and the Fortunes of the Harappan Civilization   ]
3) K.S. Valdiya on the glacial Saraswati in Current Science


I have been working on this on and off for the past few days. The Times of India (March 31 2008, couldn't find it online) carried a big new report on how the governments of Haryana and Rajasthan are trying to revive the ancient lost river Saraswati which Indologists have identified as the ephemeral river Ghaggar which flows through Haryana and Rajasthan. The news reports had lots of stuff about ancient channels under the Thar desert and as expected lots of misconceptions about geological terms and concepts. I thought I will clarify some of the geological understanding regarding the river Ghaggar /Saraswati.

1) Where is the river Saraswati?

There are many Saraswati's. The one of interest is the Ghaggar river system. The Rig-Ved describes the Saraswati as flowing between the Sutlej and the Yamuna. This has led Indologists to identify the Vedic Saraswati with the present day Ghaggar. The present day Ghaggar system originates in the Siwalik sub Himalayan region. There is also a small tributary in the Ghaggar river system named Saraswati. It originates in the Siwalik hills near Ambala and joins the Ghaggar river system in the plains of Haryana. I looked around for a good annotated map of this river system and could not find any. Most maps depict the entire north Indian plains and claim that as the Saraswati river basin. This is misleading. Below is an annotated map of the Ghaggar -Saraswati basin that I've prepared. Click on image for larger map.

2) In the past the Ghaggar-Saraswati system contained a lot more water than it does today.

This is true. Today the Ghaggar is an ephemeral river flowing only in the monsoons. Near Sirsa it dries up. Across the border in Pakistan it is called the Hakra. Remote Sensing studies have shown the Hakra to continue southward through the Thar desert terminating in the Rann of Kutch. Several channels have been identified below the Thar sands in western Rajasthan and are believed to represent the ancient channel system associated with the Ghaggar -Hakra drainage. There are hundreds of Harappan age sites along the Hakra and Ghaggar indicating that few thousand years ago, these rivers flowed year round. Paleochannels are also present in Haryana and likely represent abandoned channels of the Ghaggar river system. See image below. Arrows point to paleochannel.

3) In the past, glacially fed tributaries of the Sutlej and the Yamuna flowed into the Ghaggar river system. Tectonically induced stream piracy resulted in a diversion of these streams away from the Ghaggar system causing it to become seasonal.

This argument has been supported by geomorphological and petrological lines of evidence. Near Roopnagar (see map above) the Sutlej make an abrupt westward turn and eventually joins the Indus. Indologists have claimed that this westward turn is a relatively recent phenomenon induced by tectonic uplift in the Siwalik area. Before the Sutlej or major tributaries of it flowed into the Ghaggar. To support this they point to a widening of the Ghaggar channel where the paleo-Sutlej would have joined the Ghaggar. Literature on the web has maps showing the paleo-Sutlej joining the Ghaggar, but no such channel has been found in Haryana.

The hypothesis that the Yamuna or channels of the Yamuna systems flowed into the Ghaggar system is based on petrology. The Marakand river is one of the tributaries of the Ghaggar. It originates in the Siwalik hills (Sub Himalaya) in the western part of the Paonta Valley. This river has a series of abandoned terraces at different elevations which indicate continued down-cutting of the river. Think of these terraces as where the river channel used to be. The topmost terraces are the oldest dated to around 5000 year before present. These older terraces have been found to contain pebbles of quartzite and other metamorphic rocks (V.K.M Puri and B.C Verma, 1998; Glaciological and Geological Source of the Vedic Saraswati in the Himalayas, Itihas Darpan Vol IV, No 2: 7-36). The present drainage of the Marakand flows through only the lesser and sub Himalayas which is mostly a sedimentary terrain along with very low grade metamorphic rocks and the present channel does not contain metamorphic clasts of the type found in the older terraces. This has been taken to indicate that around 5000 years ago the Marakand was receiving sediments from streams draining metamorphic terrain and this source was cut off sometime later. Map below shows this scenario. The only way for the Marakand to have received sediments from metamorphic terrains was for the tributaries of the Yamuna - which have their source in the high Himalayas composed of metamorphic rocks - to have flowed west-north west through the Paonta Valley and joined the Marakand. Later the Yamuna captured these tributaries and this sediment supply stopped. I have not read the original paper and cannot assess the details of the sediment petrology, but taken on face value this finding supports the hypothesis that the Ghaggar was once supplied by glacial streams from the high Himalayas. Map below shows the Paonta Valley. T refers to river terraces.

Another study using geochemistry of the Ghaggar sediments has refuted this interpretation. This study uses strontium (Sr) and neodymium (Nd) isotope composition of sediments to track down the sources of sediment. This is how it works. Radiogenic Sr is enriched in crustal components which are highly differentiated from their original mantle sources. So rocks like granites, sediments derived from their weathering and their metamorphosed equivalents tend to be enriched in radiogenic Sr. The high Himalayas where all the glacial streams originate are composed of such rocks. Crustal components such as oceanic basalts, deep sea sediments and their weathered and metamorphosed equivalents are enriched in Nd isotopes reflecting less differentiation from the original mantle sources. The Indus Tsangpo suture zone, the Tibetian sedimentary series and sub-Himalayan terrains such as the Subathu formation have such high Nd signatures. So a plot of Sr vs Nd isotope composition of sediment works well to discriminate sources of sediment. Such an analysis has shown that Ghaggar sediments for the past 20 thousand years are much less Sr and more Nd than sediments of the Yamuna, Ganga and Sutlej. The inference is that the Ghaggar did not receive sediments from high Himalayan Sr rich terrain. It was never glacially fed and the drying of the Ghaggar was a result of aridification in Rajasthan beginning around 3500 B.C. and not due to tectonic controls on drainage patterns. This paper does mention the clast hypothesis favoring a glacial origin but besides presenting geochemical data does not give an alternative explanation for the metamorphic clasts.

Isotope composition of fresh water in buried channels of Jaisalmer district, Rajasthan interpreted as part of the Ghaggar system, using isotopes 2H, 18O, 3H and 14C appears to indicate that this water is ancient on the order of several thousand years old, was replenished when the river was flowing, but the d18O (ratio of 18O to 16O) composition does not show any signatures of glacial origin. The oxygen isotope composition of rainfall in affected by altitude. More than 99% of oxygen is 16O and 1% is 18O. As water vapor condenses at a particular altitude 18O is preferentially removed in rainwater leaving rising clouds depleted in 18O. This is really a temperature effect on oxygen isotope fractionation. At lower altitudes and therefore higher temperatures the lighter isotope 16O will fractionate preferentially in the vapor phase. This effect is reduced at low temperatures. So rainfall falling on the Siwaliks at lower altitudes will have a higher d18O value compared with rainfall/snow falling in the low temperature glacial region of the high Himalaya. The composition of ancient fresh water in the paleo-channels of Rajasthan has higher d18O value compared with Himalayan glacial standard indicating that glacial water did not contribute to this ancient river. This further supports the hypothesis that the Ghaggar system never had any glacial connections and always drained only the Siwalik- Sub Himalayan region.

So the petrology of the clasts in Markand terraces support a past glacial contribution to the Ghaggar while the geochemistry of sediments and ancient water does not. I would really like to take a look at those metamorphic clasts.

There is another line of evidence which seems not to have been explored in this context and that is paleo-current analysis of the older terraces along the Paonta valley. Paleocurrents and therefore the direction of stream flow can be found out by analyzing the orientation of sedimentary structures such as channel bedding surfaces, imbricate patterns of pebbles in conglomerates, ripple marks in finer sands etc. This type of analysis can tell us if at any point in the past streams flowed west-north west along the Paonta valley which would support the glacial origin hypothesis.

Update: I just had to write this update seeing this post has generated a lot of traffic. Saying that metamorphic clasts are present in the river terrace sediments of Marakand river is not enough evidence that there was a source from the high Himalayas. The lesser Himalayas in which the Marakand originates contain some low grade metamorphic rocks. The high Himalayas contain medium to high grade metamorphic rocks. So specifying the grade of metamorphism is important and at least in the literature available I did not see any specific description. Even if high grade metamorphic fragments were present in the Marakand terrace sediments it does not necessarily mean that the present day Marakand had a source in the high Himalayas. It is important to understand what type of component do the high grade metamorphic fragments form within the terrace pebbles. Are the pebbles entirely composed of the high grade metamorphic minerals? That would indicate that the pebbles are broken off pieces of high grade metamorphic rocks indicating a primary source. Or are the pebbles composed of older sedimentary rocks (lithoclasts) which contain high grade metamorphic minerals as detrital grains. That would indicate that the high grade metamorphic minerals have been recycled into the Marakand terraces by the weathering of an earlier sedimentary sequence, which in turn were made up of detritus eroded from older high grade metamorphic terrains. Simply put rocks of higher Himalayas and lesser Himalayan sequences eroding to form the Siwaliks, which in turn eroding to shed detritus in the present day Marakand. Metamorphic minerals recycling as detrital grains through earlier sedimentary sequences into the present day river terraces without the Marakand river being connected to streams draining high grade metamorphic terrains. Sediment provenance studies can be complicated in terrains like the Himalayas which has seen successive orogenic episodes and recycling of the sediments, and I want to find the original paper on these clasts to write a more detailed post.

4) The Saraswati is still flowing underground

A river system has several flow regimes associated with it. The most common is the surface flow. At present the Ghaggar has surface flow only during the monsoons. River channels have sometimes hundreds of feet of sand and sediment deposits. There is groundwater in these sands and this groundwater flows through this sand or even through subsurface bedrocks. Such a flow must exist below the present channels of the Ghaggar river system at least in Haryana where the river still flow during monsoons. Another type of underground flow is when there is a subterranean cave system. Surface waters can disappear underground and flow through such caves. This type of terrain is not present along the Ghaggar-Hakra. So the underground flow of the Ghaggar-Saraswati river is really groundwater flowing along the present and abandoned channels replenished by monsoon water from the channels and the surrounding plains. At places this groundwater appears at the surface as springs. There are temples with holding ponds to tap these spring water at several places in Haryana such as Kapil Muni. These springs are taken as evidence that the river is flowing underground. Rates of groundwater flow vary depending on permeability of the aquifer and hydraulic gradient, usually averaging to 10's of meters per day. Such rates are possible for groundwater flowing under the Ghaggar channels in the Siwalik region and the upper parts of the stream in Haryana and may explain reports of "waves" coming out of springs, the "waves" being groundwater flowing out as a spring at a fair rate. Downstream towards Rajasthan is a different story. Radioactive tracer studies have shown that at least in Rajasthan the flow in buried channels is measured in few tens of cm per year, suggesting near stagnation of groundwater and no replenishment from the Himalayan headwaters of the Ghaggar system. This water is likely to be old water which recharged the aquifer when Rajasthan was wetter. This has been borne out by isotope studies carried out by India's Bhabha Atomic Research Centre. Isotope data also indicates no glacial contribution to this old water. So the groundwater system associated with the Ghaggar drainage is likely to be a series of disconnected aquifers being recharged primarily through monsoon discharge of the river and through surrounding plains in Haryana and possibly not being replenished much in Rajasthan.

5) Targeted drilling in these paleochannels will yield a lot of water.

It is likely that these paleochannels will yield a lot of water. In Punjab and Haryana there are already thousands of bore wells extracting groundwater. Many of them no doubt have been unknowingly sunk in buried channels. Groundwater levels are falling in both these states through over-exploitation. Exploitation of these buried aquifers by targeted drilling has the potential to benefit Rajasthan more, though I feel that claims that these buried channels may contain as much water as the Sutlej canals are exaggerated. Preliminary drilling in Rajasthan along interpreted buried channel has yielded fresh water. These fresh water zones are linear and roughly trend N-S, coinciding with the direction of the ancient drainage. But more work needs to be done to understand the quantity and quality of these water resources in Rajasthan.

6) In Libya an ancient river channel has been discovered and rejuvenated into a man made river. The Saraswati can similarly be rejuvenated.

In Libya groundwater is being extracted from the Nubian sandstone aquifer which is made up of marine sedimentary sequences ranging in age from lower Paleozoic to the Cretaceous. The aquifer was last replenished several thousand years ago when north Africa was wetter. Water extracted from this aquifer is being distributed through a connected system of underground pipelines which is called the man made river.

Plans to rejuvenate the Saraswati are really linked to projects like the National Water Grid. The idea is to use excess water from the Indira Gandhi Nahar Project (IGNP) and the floodwaters from the Ghaggar river to recharge the aquifers in the Thar desert, some of them coinciding with buried channels of the Ghaggar and Hakra system. Bombastic claims have been made that a sustainable system of a million tube wells can be developed. Sustainability depends not on the original water stored in these aquifer but upon recharge of groundwater from surface water brought by the Rajasthan canal which is being extended into Gujarat. I have to say that geo-engineering projects may become necessary as climate change will affect the water resources in the Gangetic river system. Say Himalayan glaciers decline and monsoon becomes concentrated in short bursts, how do we deal with an overall decline in water but also an excess during very short periods. Is storing water in underground aquifers like to one's identified in Rajasthan a viable option? What are the engineering hurdles and environmental problems likely to crop up?


  1. Excellent post Suvrat, you have explained things very well. Thanks for enlightening me.

  2. "In the past, glacially fed tributaries of the Sutlej and the Yamuna flowed into the Ghaggar river system. "

    Please see the article: "Is River Ghaggar, Saraswati? Geochemical constraints" http://www.ias.ac.in/currsci/oct252004/1141.pdf

    How do these findings change our views on what vedic Saraswati is?

  3. Surya-

    I think you have taken what i wrote out of context. the post is written as a FAQ. one FAQ is the sentence you quoted.

    I have given evidence that has been interpreted by some Indologists to support a glacial fed Ghaggar, and then given evidence (much more convincing in my opinion) that does not support a glacial origin. I have in fact linked to the very paper that you have given in your comment.

    Looking at the geological data I don't find much convincing evidence that the Ghaggar was ever glacially fed.

    1. Bhabha Atomic Research Centre had done some investigation by isotopic analysis of the alluvial sands of Saraswathi River and had confirmed that the Saraswathi river originated from the himalayas.

    2. Isotope work done suggests a source in the Lower Himalaya (Shivaliks) and not to glaciers. see this paper-


      If you know of any other isotope work, do post a link.

  4. Me and my wife were discussing how river Saraswati got disappeared? And Google God landed us on your page. I would say you have deep understanding of earth and more. Please accept our best wishes.

    PS: If you have any latest update please let us know here- abhayanand.bhagwat@gmail.com


  5. very good job done suvrat sir,

    i am a pursuing geoscience engineering. The post really helped me get a view of the paleochannel for the river SARASWATI and GHAGGAR.

    i`d like to be in touch..!!


  6. Mystery of River Saraswati is no loner a mystery after reading your blog and with the help of the Google Maps.Wonderful explanation that a person with basic Geo knowledge could understand.Thanks.

  7. Hum hanumangarh me ghaghar se 13 km door h bina kisi bdi canal k 500 feet tk groundwater fresh h paas me koi bhi recharge source nhi h sand dune ka area h fir bhi groundwater bilkul fresh h