20 30J u ly 2 0 0 2 J a n . 2 0 0 3 J u ly 2 0 0 3 J a n . 2 0 0 4 J u ly 2 0 0 4 J a n . 2 0 0 5 J u ly 2 0 0 5 J a n . 2 0 0 6 J u ly 2 0 0 6 J a n . 2 0 0 7 J u ly 2 0 0 7 J a n . is 5.9 cm, compared with 10.3 cm for soil water and 13.8 cm for TWS. These relationships are consistent with those reported in previous studies of soil-water/groundwater covariability 5,14,22 . The soil-water time series reflects rainfall anomalies during the period (discussed below) and exhibits no significant trend. On the other hand, TWS and groundwater decline steadily from 2003 onwards. We calculate the rate of depletion of groundwater to be 4.0 6 1.0 cm yr 21 . Assuming 2 a specific yield of 0.12, the regional mean rate of water As best can be determined from the coarse GRACE observations, maximum rates of groundwater depletion are centred on Haryana. Groundwater levels also appear to be declining quickly in western Uttar Pradesh, to the east of Haryana. If there is groundwater depletion in Pakistan, to the northwest, it seems to be much less severe.Although six years is a short period from which to assess a longterm trend with confidence, two pieces of evidence support our conclusion that severe groundwater depletion is occurring as a result of human consumption rather than natural variability. First, the Indian Ministry of Water Resources reports that groundwater withdrawals exceed recharge in the three states we studied 2 . Irrigation accounts for about 95% of the consumption 2 ; about 28% of the area is irrigated 23 . Second, there was no shortage of rainfall in the region to cause a natural decline in water storage. GLDAS modelled soil-water fields integrate the effects of precipitation, solar radiation, air temperature and other meteorological factors that directly or indirectly influence groundwater storage 4 . The trend in simulated soil-water storage during the period of study was 0.4 cm yr 21 . This supports the notion that groundwater declines were not caused by natural climate variability. It also confirms that the computed groundwater trend is not a mathematical artefact caused by the subtraction of a large positive soil-water trend from the GRACE-derived TWS trend.We conclude that withdrawals for irrigation and other uses are depleting the groundwater reserves of Rajasthan, Punjab and Haryana at a rate of 4.0 6 1.0 cm yr 21 equivalent height of water, or 17.7 6 4.5 km 3 yr 21 . The Indian Ministry of Water Resources reports that the difference between annual available recharge and annual withdrawals in the region is a 13.2 km 3 yr 21 deficit 2 . Our result implies that the portion of irrigated water that replenishes the aquifers is less and/or the rate of withdrawal is more than the Indian government has estimated. Apparently, most of the groundwater withdrawn subsequently is lost from the region as a result of increases in run-off and/or evapotranspiration. Between August 2002 and October 2008, the region lost 109 km 3 of groundwater, which is double the capacity of India's largest reservoir, the Upper Waingang...

10As the world's largest distributed store of freshwater, groundwater plays a central role in 11 sustaining ecosystems and enabling human adaptation to climate variability and change. 12The strategic importance of groundwater to global water and food security will intensify 13 under climate change as more frequent and intense climate extremes (droughts, floods) 14 increase variability in soil moisture and surface water. Here we critically review recent 15 research assessing climate impacts on groundwater through natural and human-induced 16 processes as well as groundwater-driven feedbacks on the climate system.

Freshwater availability is changing worldwide. Here we quantify 34 trends in terrestrial water storage observed by the Gravity Recovery and Climate Experiment (GRACE) satellites during 2002-2016 and categorize their drivers as natural interannual variability, unsustainable groundwater consumption, climate change or combinations thereof. Several of these trends had been lacking thorough investigation and attribution, including massive changes in northwestern China and the Okavango Delta. Others are consistent with climate model predictions. This observation-based assessment of how the world's water landscape is responding to human impacts and climate variations provides a blueprint for evaluating and predicting emerging threats to water and food security.