(Carte de la sarasvati védique from Wikipedia)
(Science, Vol 332 had an article titled In Indus Times, the River Didn’t Run Through It by Andrew Lawler. In the article, three independent studies were cited to argue that Ghaggar-Hakra was a seasonal stream during the Mature Harappan phase (2500 – 1900 B.C.E) and not the mighty river mentioned in the Vedas. The studies also show that the river may have dried up around 10,000 years back. I asked Michel Danino, the author of The Lost River:On the Trail of the Saraswati, for his comments on these studies – Ed)
A few thoughts on the recent challenge posed by a few geological studies mentioned on this group (mainly Sanjeev Gupta et al.; H. Maemoku et al.). Some links to them (1,2, 3,4)
Those studies conclude that “the data suggests there was no big river here” (near Kalibangan in northern Rajasthan, on the bank of the Ghaggar) or that “the Ghaggar was not the mighty Saraswati during mature Harappan period because sand dunes on either side of the Ghaggar had been formed before that”. This is contrary to the view adopted by most archaeologists that the Ghaggar and its tributaries flowed during Harappan times, watering the hundreds of sites that have been found in this region. (Since the mid-nineteenth century, the Ghaggar was additionally identified with the Sarasvati river of the Rig-Veda.)
In reality, the issues are not so simple. First there is nothing really new in the claim that we don’t have a “mighty Sarasvati” during the Mature phase. In my recent book Lost River: On the Trail of the Sarasvati, I listed various evidences showing that the Ghaggar was dwindling during the Mature phase. In summary:

1. The Pakistani archaeologist M Rafique Mughal’s observation of a break in the settlement pattern between the Early and Mature phases (around 2600 BCE) just west of the international border for some 100 kilometres. His conclusion was that the Ghaggar had stopped flowing into Cholistan before the Mature phase; this means it was much weakened by that time.
2. In the same region, the German scientists M.A. Geyh and D. Ploethner detected a huge and shallow body of fresh groundwater. A tritium-based isotope study pointed to “a range of the actual water age from 12900 to 4700 years BP”, i.e. till about 2700 BCE, which matches Mughal’s conclusion.
3. A 2008 U.S.-Pakistan study directed by Peter Clift tested the Ghaggar-Hakra’s floodplain in Pakistan’s Punjab and concluded, “Provisional age data now show that between 2000 and 3000 BCE, flow along a presently driedup course known as the Ghaggur-Hakkra River ceased, probably driven by the weakening monsoon and possibly also because of headwater capture into the adjacent Yamuna and Sutlej Rivers.” This is again consistent with the above.

What is new and challenging in the recent geological studies is a suggestion that the Ghaggar went dry many millennia earlier. Actually that statement is found in A. Lawler’s article in Science, not in S. Gupta et al.’s abstract — and we don’t have the full papers as far as I know. Maemoku et al.’s conclusion that “the Ghaggar was not the mighty Saraswati during mature Harappan period because sand dunes on either side of the Ghaggar had been formed before that” can be readily dismissed because the age of the sand dunes is irrelevant to the question of water flow at various later dates.
Apart from the above views of Mughal, Geyh and Ploethner, and Clift, there are several major objections to a completely dry Ghaggar during Mature times. For instance:

1. There were undoubtedly numerous streams flowing down from the Shivaliks, and most recent climatic studies agree that the climate was wetter during the Mature phase, though on the way to aridity: all these streams (Sarsuti, Markanda, Dangri, Ghaggar, Patialewali, Wah and the three Naiwals being the chief ones) carried more water than they do at present. Moreover, nowadays, whatever water flows seasonally in those streams is largely diverted to irrigation through canals; in Harappan times, the diversion would have been much less. As a result, all this surplus water must have accumulated somewhere — where, if not in the Ghaggar? In fact, during last year’s abundant summer monsoon, the Ghaggar was full to the brim well into India’s Punjab, and we have records to show that it flowed all the way to Anupgarh decades earlier. Because of higher precipitation and lesser diversion, its flow in Harappan times could have been larger as well as for longer periods of the year.
2. Why are sites such as Banawali or Kalibangan built on the edge of well-defined paleochannels if those channels had no flowing water?
3. In particular, there are two crucial messages from Kalibangan’s urban layout: 1) the absence of large reservoirs (such as those at Dholavira) and a relatively small number of wells, both of which together point to a perennial source of water nearby; 2) the recessed entrance to the upper town precisely facing the Ghaggar below. The architectural message is unmistakable: the Kalibangan citizens had access to a flowing Ghaggar, both for water supply and for communication.
4. An earlier study by JK Tripathi et al.concluded that “The Palaeo-Ghaggar must have been a mighty river”. Though I regard that study as substandard (I explained why in my book), I mention is to show that experts can and do disagree.
5. Another case of disagreement can be found among the recently mentioned abstracts, see p. 23 of the pdf file: P. Clift et al., “Evolving Holocene Drainage Geometries and Environmental Conditions in the Indus River Basin”. I quote from the abstract:

“This trend became more intense after 4.5 ka [i.e. 2500 BCE ] when the last evidence for an active river was found in the region close to the archeological sites. … We suggest that in the Early-Mid Holocene the area of heaviest Harappan Settlement was one of significant fluvial confluence. The Sutlej and an independent Beas River flowed much closer to the Thar Desert than they do now. Moreover, we propose that the Yamuna, which now flows east into the Ganges, must have contributed to the sediment flux in the recent geological past, although the precise age of capture is, so far, not yet well determined. The end of the Mature Harappan Phase of settlement around 1900 BCE appears to shortly postdate the end of major river flow in the region, as the Sutlej migrated north, capturing the Beas. This change in the course of the Sutlej, together with the probable loss of the Yamuna resulted in the much smaller Ghaggar-Hakra being unable by itself to maintain significant flow into the desert, especially in the context of a weakening summer monsoon. The effect of this reorganization may have been as catastrophic to agriculture as the proposed abrupt weakening of the summer monsoon rains.”

This scenario is the same as that proposed by numerous experts earlier. It implies a flowing Ghaggar, partly fed by waters from the Sutlej and the Yamuna. Why hasn’t Lawler mentioned this alternative view in his Science article? And why do Sanjeev Gupta et al. fail to notice recent contributions from the Sutlej into the Ghaggar system, which are well attested by an Islamic chronicle of the 15th century and by the 1908 Imperial Gazetteer (which mentions the Sutlej finally leaving the Ghaggar in 1796)? I think the answer lies in the treacherous nature of the sediments in the Ghaggar region, which are notoriously difficult to interpret (even earlier, there were major disagreements among experts, e.g. Raikes and Courty).

These new geological efforts are welcome but I think we need to give them a few years to expand their scope and stabilize their findings. We also need to hear detailed discussions from geologists who have worked on the Sarasvati problem for a long time, such as KS Valdiya, VMK Puri, BC Verma etc.
Finally, nothing less than a multidisciplinary approach will provide a convincing answer to the question of water sources for the hundreds of Harappan sites in Haryana, Indian Punjab and northern Rajasthan: not just geology, but climatology, isotope studies of palaeo-waters, and of course archaeology (especially more refined studies of the evolution of settlement patterns). We are still far from such a multidisciplinary convergence.