Recently published on-line in
Geology is a paper which might not appear on the surface to be very archaeobotanical, but which is important for thinking about the past agriculture of the Indus valley. This is by Clift et al (2012) on "
U-Pb zircon dating evidnece for a Pleistocene Sarasvati River and capture of the Yamuna River".
This paper provides sources for the headwater sediments in the various rivers of the Indus system based on zircon finger-printed (geological source dating in the 1000s of millions of years). These dated source profiles in turn are stratified in the Pleistocene and Holocene river sequences which have been dated by OSL.
These river systems include the now extinct Ghaggar-Hakra river, often equated with the 'lost Saraswati" of Indian epic.
The paper shows that while the Ghaggar-Hakra used to be much larger in the Pleistocene, drawing on the headwaters that now feed the Yamuna, tha Yamuna had begun to flow east into the Ganges before the End of the Pleistocene, and therefore well before the start of Harappan urban societies. 
(but Indian civilization is millions of years old so who knows right?)
Throughout the Holocene, including the Harappan period this river was fed only by seasonal monsoon rain in the east. This rain-fed Ghaggar-Hakra was active until after 4.5 ka and was then covered by dunes before 1.4 ka. What this means is that the Ghaggar-Hakra, unlike any of the major Indus tributaries, was not fed by snow melt, which begins in Spring and may be unpredictable, but was entirely reliant on swelling its banks from the summer monsoon.
This means it would have been an ideal river for winter crop agriculture, along the lines of the Nile flood regime which is keyed to the Blue Nile's monsoon source, with sowing of wheat and barley in Oct.-Nov. as the monsoon flood began to recede to leave behind a rich floodplain. These could then be left to mature until harvests in March or April, without fear of early snowmelt floods ruining crops. It really should come as no surprise then that so many Harappan Bronze Age sites concentrated in this valley.
Nevertheless as monsoons gradually weakened (already underway during the Harappan period) with the flood water source retreating eastwards, and the Thar desert expanding, the valley became gradually drier and eventually choked with desert sands. This, however happened in Iron Age or post-Iorn Age times, so thus there is no basis for correlating any catastrophic shift in the Ghaggar-Hakra with the end of the Harappan civilization-- a notion which has often appealed to archaeologists.
[edited for typos 9.2.2102 DF]
