Search

The branching angles of stream network are the fingerprint of the processes that shape our landscape. However, the mechanisms that give rise to stream network patterns on Earth are not fully understood. Recent studies have shown controls of climate, tectonics, and lithology on channel incision and the planform geometry of stream networks. Our analysis of one million river junctions and over 4.2 million groundwater well observations across the contiguous United States shows for the first time that stream network branching angle vary systematically with the degree to which streams and groundwater interact. Streams that are losing their water to groundwater exhibit narrow branching angles while streams that are gaining water from groundwater exhibit wide branching angles on average. We show that the correlation between branching angle and fraction of losing streams is stronger than branching angle and other controls of stream network planform geometry. The systematic relationship between branching angle and losing fraction persist across a range of topographic gradient and across several stream orders. Our findings brings forward a mechanistic linkage between previously shown correlation between branching angles and climate.

Groundwater is one of the largest reservoirs of water on Earth but has relatively small fluxes compared to its volume. This behaviour is exaggerated at depths below 500 m, where the majority of groundwater exists and where residence times of millions to even a billion years have been documented. However, the extent of interactions between deep groundwater (>500 m) and the rest of the terrestrial water cycle at a global scale is unclear because of challenges in detecting their contributions to streamflow. Here, we use a chloride mass balance approach to quantify the contribution of deep groundwater to global streamflow. Deep groundwater likely contributes

Under pressure of population growth and increasing food demand, irrigated agriculture has become increasingly dependent on groundwater, which has resulted in high rates of groundwater depletion in many aquifers around the globe. Groundwater level declines increase pumping costs and may render irrigated agriculture uneconomic, however, the depths at which groundwater pumping ceases to be economically efficient remain unclear. Here, we estimate the maximum (deepest) depth to groundwater at which groundwater pumping becomes uneconomic when groundwater is used as a last resort for irrigating current crops. We show that the maximum depth to groundwater at which pumping for irrigation becomes uneconomic varies strongly between regions, with the maximum depths (>1000 m) found in large sedimentary basins where lucrative crops are grown. From the perspective of irrigated agriculture, the global volume of economically extractable groundwater currently under irrigation is estimated as 0.88 ± 0.08 million km3 which is estimated to be less than 8.4 ± 0.7% of the estimated volume of fresh groundwater in the Earth’s uppermost 2 km, and represents a current present value of 28.1 ± 7.4 trillion USD. Although globally the time until groundwater depletion renders pumping uneconomic is long in many areas, the remaining value of economically recoverable groundwater is likely to sharply decline in the coming century.

When your well runs dryGroundwater provides nearly half of the water used for agricultural irrigation and most of the drinking water for billions of people. It is essential, then, for this resource to remain secure. Jasechko and Perrone examined data from approximately 39 million wells in 40 countries worldwide to investigate their vulnerability to declining water levels (see the Perspective by Famiglietti and Ferguson). The authors found that construction of deeper wells is not occurring in some areas that are experiencing groundwater decline, a disconnect that poses risks for people who rely on well water.Science, this issue p.418; see also p.344

Most rivers exchange water with surrounding aquifers1,2. Where groundwater levels lie below nearby streams, streamwater can infiltrate through the streambed, reducing streamflow and recharging the aquifer3. These ‘losing’ streams have important implications for water availability, riparian ecosystems and environmental flows4–10, but the prevalence of losing streams remains poorly constrained by continent-wide in situ observations. Here we analyse water levels in 4.2 million wells across the contiguous USA and show that nearly two-thirds (64 per cent) of them lie below nearby stream surfaces, implying that these streamwaters will seep into the subsurface if it is sufficiently permeable. A lack of adequate permeability data prevents us from quantifying the magnitudes of these subsurface flows, but our analysis nonetheless demonstrates widespread potential for streamwater losses into underlying aquifers. These potentially losing rivers are more common in drier climates, flatter landscapes and regions with extensive groundwater pumping. Our results thus imply that climatic factors, geological conditions and historic groundwater pumping jointly contribute to the widespread risk of streams losing flow into surrounding aquifers instead of gaining flow from them. Recent modelling studies10 have suggested that losing streams could become common in future decades, but our direct observations show that many rivers across the USA are already potentially losing flow, highlighting the importance of coordinating groundwater and surface water policy. Direct observations of 4.2 million wells across the USA indicate that many streams are potentially losing water to underlying aquifers.

How much precipitation recharges groundwaters varies enormously across Earth's surface, but recharge rates are uncertain because field observations are sparse and modeled global estimates remain largely unvalidated. Here we show that annual recharge is predictable as a simple function of climatic aridity — the ratio of long-term potential evapotranspiration to precipitation — using a global synthesis of measured recharge of 5237 sites across six continents. We use this relationship to estimate long-term recharge globally outside of permafrost regions. Our estimates double previous global hydrological model estimates and are more consistent with empirical field observations. These revised higher estimates of global groundwater recharge imply that groundwater contributes more actively to evapotranspiration and streamflow than previously represented in global water cycle depictions or global hydrological and Earth system models. In addition, we quantify the sensitivity of groundwater recharge to changes in aridity using the empirical relationship between groundwater recharge rates and climatic aridity. This analysis indicates that recharge is most sensitive to climate aridity in mesic regions, where changes in the replenishment of aquifers will be amplified relative to projected changes in precipitation. Global hydrological models seem to underestimate changes in recharge with climate aridity. Thus, the impacts of climatic changes on the replenishment of Earth's largest liquid freshwater stores may be larger than previously anticipated.

Most stored groundwater is ‘fossil’ in its age, having been under the ground for more than ~12 thousand years. Mapping where wells tap fossil aquifers is relevant for water quality and quantity management. Nevertheless, the prevalence of wells that tap fossil aquifers is not known. Here we show that wells that are sufficiently deep to tap fossil aquifers are widespread, though they remain outnumbered by shallower wells in most areas. Moreover, the proportion of newly drilled wells that are deep enough to tap fossil aquifers has increased over recent decades. However, this widespread and increased drilling of wells into fossil aquifers is not necessarily associated with groundwater depletion, emphasizing that the presence of fossil groundwater does not necessarily indicate a non-renewable water supply. Our results highlight the importance of safeguarding fossil groundwater quality and quantity to meet present and future water demands.

Groundwater depletion is causing wells to run dry, affecting food production and domestic water access. Drilling deeper wells may stave off the drying up of wells—for those who can afford it and where hydrogeologic conditions permit it—yet the frequency of deeper drilling is unknown. Here, we compile 11.8 million groundwater-well locations, depths and purposes across the United States. We show that typical wells are being constructed deeper 1.4 to 9.2 times more often than they are being constructed shallower. Well deepening is not ubiquitous in all areas where groundwater levels are declining, implying that shallow wells are vulnerable to running dry should groundwater depletion continue. We conclude that widespread deeper well drilling represents an unsustainable stopgap to groundwater depletion that is limited by socioeconomic conditions, hydrogeology and groundwater quality. Groundwater wells in the United States are under more stress than ever before due to drought conditions and rising demand, but the extensive nature of deeper drilling has been unreported. This analysis compiles nearly 12 million groundwater wells across the United States to determine water vulnerability and sustainability.

Understanding the time that water takes to traverse the vadose zone is important to quantify groundwater recharge rates and to assess the vulnerability of aquifers to contamination. Tritium mass balance models provide measurement-driven estimates of soil water velocities, yet uncertainties embedded in these groundwater recharge estimates are rarely investigated. Here we analyze six >15-m tritium profiles collected from thick unsaturated zones located on China’s Loess Plateau and compare recharge rates obtained from various calculation approaches. Models used to estimate recharge based on unsaturated zone data include the tritium peak method and the tritium storage method; models used to estimate recharge using saturated zone data include the ‘modern groundwater’ method and a method that assumes groundwater is well-mixed in the aquifer. The tritium peak method presents reliable recharge; however, the recharge from the tritium storage method is two times of that from the tritium peak method. The well-mixed reservoir method overestimates groundwater renewal rates because the implicit assumption that groundwater is well-mixed in aquifers is not realistic. For potential recharge, considerable uncertainties arise from precipitation tritium reconstruction; while for actual recharge estimation, uncertainties come from the mismatched recharge mechanism between the model assumption and local characteristics. Selection of appropriate tritium mass balance models is of utmost importance for recharge estimation.

International audience; Groundwater pollution threatens human and ecosystem health in many regions around the globe. Fast flow to the groundwater through focused recharge is known to transmit short-lived pollutants into carbonate aquifers, endangering the quality of groundwaters where one quarter of the world’s population lives. However, the large-scale impact of such focused recharge on groundwater quality remains poorly understood. Here, we apply a continental-scale model to quantify the risk of groundwater contamination by degradable pollutants through focused recharge in the carbonate rock regions of Europe, North Africa, and the Middle East. We show that focused recharge is the primary reason for widespread rapid transport of contaminants to the groundwater. Where it occurs, the concentration of pollutants in groundwater recharge that have not yet degraded increases from

The depth at which groundwaters transition from fresh to more saline-the "base of fresh water"-is frequently used to determine the stringency and types of measures put in place to manage groundwater and protect it from contamination. Therefore, it is important to understand salinity distributions and compare defined bases of fresh water with salinity distributions and groundwater well depths. Here we analyze two distinct datasets: 1) a large set of total dissolved solids concentration (TDS) measurements (n = 216,754) and 2) groundwater well locations and depths (n = 399,454) across California. We find that 19 to 56% of the groundwater TDS measurements made at depths deeper than defined bases of fresh water pump fresh groundwater (TDS < 2,000 mg/L). Because fresh groundwater is found at depths deeper than the base of fresh water, current policies informed by base of fresh water assessments may not be managing and protecting large volumes of deep fresh groundwater. Furthermore, we find that nearly 4% of existing groundwater wells penetrate defined bases of fresh water, and nearly 16% of wells overlie it by no more than 100 m, evidencing widespread encroachment on the base of fresh water by groundwater users. Consequently, our analysis suggests that groundwater sustainability in California may be poorly safeguarded in some places and that the base-of-fresh-water concept needs to be reconsidered as a means to define and manage groundwater.

Most rivers exchange water with surrounding aquifers

Seawater intrusion into coastal aquifers can increase groundwater salinity beyond potable levels, endangering access to freshwater for millions of people. Seawater intrusion is particularly likely where water tables lie below sea level, but can also arise from groundwater pumping in some coastal aquifers with water tables above sea level. Nevertheless, no nation-wide, observation-based assessment of the scope of potential seawater intrusion exists. Here we compile and analyze ~250,000 coastal groundwater-level observations made since the year 2000 in the contiguous United States. We show that the majority of observed groundwater levels lie below sea level along more than 15% of the contiguous coastline. We conclude that landward hydraulic gradients characterize a substantial fraction of the East Coast (>18%) and Gulf Coast (>17%), and also parts of the West Coast where groundwater pumping is high. Sea level rise, coastal land subsidence, and increasing water demands will exacerbate the threat of seawater intrusion., Nature Communications, 11 (1), ISSN:2041-1723

Hydraulic fracturing operations are generating considerable discussion about their potential to contaminate aquifers tapped by domestic groundwater wells. Groundwater wells located closer to hydraulically fractured wells are more likely to be exposed to contaminants derived from on-site spills and well-bore failures, should they occur. Nevertheless, the proximity of hydraulic fracturing operations to domestic groundwater wells is unknown. Here, we analyze the distance between domestic groundwater wells (public and self-supply) constructed between 2000 and 2014 and hydraulically fractured wells stimulated in 2014 in 14 states. We show that 37% of all recorded hydraulically fractured wells stimulated during 2014 exist within 2 km of at least one recently constructed (2000-2014) domestic groundwater well. Furthermore, we identify 11 counties where most ([Formula: see text]50%) recorded domestic groundwater wells exist within 2 km of one or more hydraulically fractured wells stimulated during 2014. Our findings suggest that understanding how frequently hydraulic fracturing operations impact groundwater quality is of widespread importance to drinking water safety in many areas where hydraulic fracturing is common. We also identify 236 counties where most recorded domestic groundwater wells exist within 2 km of one or more recorded oil and gas wells producing during 2014. Our analysis identifies hotspots where both conventional and unconventional oil and gas wells frequently exist near recorded domestic groundwater wells that may be targeted for further water-quality monitoring.

The turnover of groundwater through recharge drives many processes throughout Earth's surface and subsurface. Yet groundwater turnover rates and their relationship to regional climate and geology remain largely unknown. We estimated that over 200 million km3 of groundwater has recharged since the Last Glacial Maximum (LGM), which is ten times the volume of global groundwater storage. However, flushing is very unevenly distributed throughout Earth's one million watersheds, with some aquifers turned over thousands of times to others with

Transformations of precipitation into groundwater and streamflow are fundamental hydrological processes, critical to irrigated agriculture, hydroelectric power generation and ecosystem health. Our understanding of the timing of groundwater recharge and streamflow generation remains incomplete, limiting our ability to predict fresh water, nutrient, and contaminant fluxes, especially in large basins. Here we analyze thousands of rain, snow, groundwater and streamflow δ18O and δ2H values in the Nelson River basin, which covers 1.2 million km2 of central Canada. We show that the fraction of precipitation that recharges aquifers is ~1.3-5 times higher for precipitation falling during cold months with subzero mean monthly temperatures than for precipitation falling during warmer months. The near-ubiquity of cold-season-biased groundwater recharge implies that changes to winter water balances may have disproportionate impacts on annual groundwater recharge rates. We also show that young streamflow—defined as precipitation that enters a river in less than ~2.3 months—comprises ~27 % of annual streamflow, but varies widely among tributaries in the Nelson River basin (1-59 %). Young streamflow fractions are lower in steep catchments, and higher in flatter catchments such as the transboundary Red River basin. Our findings imply that flat, lower-permeability, heavily-tiled landscapes favor more rapid transmission of precipitation into rivers, possibly mobilizing excess soluble fertilizers and exacerbating eutrophication events in Lake Winnipeg.

Even though knowing the contributions of transpiration (T), soil and open water evaporation (E), and interception (I) to terrestrial evapotranspiration (ET = T + E + I) is crucial for understanding the hydrological cycle and its connection to ecological processes, the fraction of T is unattainable by traditional measurement techniques over large scales. Previously reported global mean T/(E + T + I) from multiple independent sources, including satellite-based estimations, reanalysis, land surface models, and isotopic measurements, varies substantially from 24% to 90%. Here we develop a new ET partitioning algorithm, which combines global evapotranspiration estimates and relationships between leaf area index (LAI) and T/(E + T) for different vegetation types, to upscale a wide range of published site-scale measurements. We show that transpiration accounts for about 57.2% (with standard deviation ± 6.8%) of global terrestrial ET. Our approach bridges the scale gap between site measurements and global model simulations,and can be simply implemented into current global climate models to improve biological CO2 flux simulations.

Quantifying seasonal variations in precipitation δ2H and δ18O is important for many stable isotope applications, including inferring plant water sources and streamflow ages. Our objective is to develop a data product that concisely quantifies the seasonality of stable isotope ratios in precipitation. We fit sine curves defined by amplitude, phase, and offset parameters to quantify annual precipitation isotope cycles at 653 meteorological stations on all seven continents. At most of these stations, including in tropical and subtropical regions, sine curves can represent the seasonal cycles in precipitation isotopes. Additionally, the amplitude, phase, and offset parameters of these sine curves correlate with site climatic and geographic characteristics. Multiple linear regression models based on these site characteristics capture most of the global variation in precipitation isotope amplitudes and offsets; while phase values were not well predicted by regression models globally, they were captured by zonal (0–30∘ and 30–90∘) regressions, which were then used to produce global maps. These global maps of sinusoidal seasonality in precipitation isotopes based on regression models were adjusted for the residual spatial variations that were not captured by the regression models. The resulting mean prediction errors were 0.49 ‰ for δ18O amplitude, 0.73 ‰ for δ18O offset (and 4.0 ‰ and 7.4 ‰ for δ2H amplitude and offset), 8 d for phase values at latitudes outside of 30∘, and 20 d for phase values at latitudes inside of 30∘. We make the gridded global maps of precipitation δ2H and δ18O seasonality publicly available. We also make tabulated site data and fitted sine curve parameters available to support the development of regionally calibrated models, which will often be more accurate than our global model for regionally specific studies., Hydrology and Earth System Sciences, 23 (8), ISSN:1027-5606, ISSN:1607-7938

Global water cycles, ecosystem assemblages, and weathering rates were impacted by the ∼4°C of global warming that took place over the course of the last glacial termination. Fossil groundwaters can be useful indicators of late-Pleistocene precipitation isotope compositions, which, in turn, can help to test hypotheses about the drivers and impacts of glacial-interglacial climate changes. Here, a global catalog of 126 fossil groundwater records is used to interpolate late-Pleistocene precipitation δ18O across the global landmass. The interpolated data show that extratropical late-Pleistocene terrestrial precipitation was near uniformly depleted in 18O relative to the late Holocene. By contrast, tropical δ18O responses to deglacial warming diverged; late-Pleistocene δ18O was higher-than-modern across India and South China but lower-than-modern throughout much of northern and southern Africa. Groundwaters that recharged beneath large northern hemisphere ice sheets have different Holocene-Pleistocene δ18O relationships than paleowaters that recharged subaerially, potentially aiding reconstructions of englacial transport in paleo ice sheets. Global terrestrial late-Pleistocene precipitation δ18O maps may help to determine 3-D groundwater age distributions, constrain Pleistocene mammal movements, and better understand glacial climate dynamics.

Groundwater age—defined as the time elapsed since the groundwater entered the subsurface—is a helpful indicator of groundwater quality and renewal. Some tracer-based groundwater age calculations require assumptions about the distribution of ages in the groundwater sample that, unfortunately, cannot be validated. Here, straightforward mixing models are developed that do not require a priori knowledge of age distributions and are able to quantify the fraction of a groundwater sample that recharged more recently than calendar year 1953 (using tritium) or within the past ~ 12 thousand years (using carbon-14). Calculations of young and old groundwater can be used to map groundwater velocities in flow systems driven by diverse climate conditions, groundwater extraction rates, rock types, topographic gradients and plant populations.

Streamflow is a mixture of precipitation of various ages. Oxygen isotope data suggests that a third of global river discharge is sourced from rainfall within the past few months, which accounts for less than 0.1% of global groundwater.

A U.S. court decision unlocks vast potential to improve sustainable freshwater management

Groundwater recharged less than 50 years ago is vulnerable to contamination and land-use changes. Data and simulations suggest that up to 6% of continental groundwater is modern—forming the largest component of the active hydrologic cycle.

Reconstructions of Quaternary climate are often based on the isotopic content of paleo-precipitation preserved in proxy records. While many paleo-precipitation isotope records are available, few studies have synthesized these dispersed records to explore spatial patterns of late-glacial precipitation δ18O. Here we present a synthesis of 86 globally distributed groundwater (n = 59), cave calcite (n = 15) and ice core (n = 12) isotope records spanning the late-glacial (defined as ~ 50 000 to ~ 20 000 years ago) to the late-Holocene (within the past ~ 5000 years). We show that precipitation δ18O changes from the late-glacial to the late-Holocene range from −7.1 ‰ (δ18Olate-Holocene > δ18Olate-glacial) to +1.7 ‰ (δ18Olate-glacial > δ18Olate-Holocene), with the majority (77 %) of records having lower late-glacial δ18O than late-Holocene δ18O values. High-magnitude, negative precipitation δ18O shifts are common at high latitudes, high altitudes and continental interiors (δ18Olate-Holocene > δ18Olate-glacial by more than 3 ‰). Conversely, low-magnitude, positive precipitation δ18O shifts are concentrated along tropical and subtropical coasts (δ18Olate-glacial > δ18Olate-Holocene by less than 2 ‰). Broad, global patterns of late-glacial to late-Holocene precipitation δ18O shifts suggest that stronger-than-modern isotopic distillation of air masses prevailed during the late-glacial, likely impacted by larger global temperature differences between the tropics and the poles. Further, to test how well general circulation models reproduce global precipitation δ18O shifts, we compiled simulated precipitation δ18O shifts from five isotope-enabled general circulation models simulated under recent and last glacial maximum climate states. Climate simulations generally show better inter-model and model-measurement agreement in temperate regions than in the tropics, highlighting a need for further research to better understand how inter-model spread in convective rainout, seawater δ18O and glacial topography parameterizations impact simulated precipitation δ18O. Future research on paleo-precipitation δ18O records can use the global maps of measured and simulated late-glacial precipitation isotope compositions to target and prioritize field sites.

Soil water is usually assumed to be equally available for all purposes, supplying plant transpiration as well as groundwater and streamflow; however, a study of hydrogen and oxygen isotopes from 47 globally distributed sites shows that in fact the water used by plants tends to be isotopically distinct from the water that feeds streamflow. Soil water is usually assumed to be available for all purposes in equal measure, supplying plant transpiration as well as groundwater and streamflow. Building on prior but limited studies, Jaivime Evaristo et al. have assembled a dataset of hydrogen and oxygen isotopes — drawn from widely distributed sites — and show that ecohydrological separation is the rule. Water used by plants tends to be isotopically distinct from that used for streamflow, suggesting that hydrological separation of precipitation inputs creates distinct pools of water resources. This finding implies that that existing land surface model parameterizations of plant physiological processes and streamflow can be made more realistic through the incorporation of ecohydrological separation. Current land surface models assume that groundwater, streamflow and plant transpiration are all sourced and mediated by the same well mixed water reservoir—the soil. However, recent work in Oregon1 and Mexico2 has shown evidence of ecohydrological separation, whereby different subsurface compartmentalized pools of water supply either plant transpiration fluxes or the combined fluxes of groundwater and streamflow. These findings have not yet been widely tested. Here we use hydrogen and oxygen isotopic data (2H/1H (δ2H) and 18O/16O (δ18O)) from 47 globally distributed sites to show that ecohydrological separation is widespread across different biomes. Precipitation, stream water and groundwater from each site plot approximately along the δ2H/δ18O slope of local precipitation inputs. But soil and plant xylem waters extracted from the 47 sites all plot below the local stream water and groundwater on the meteoric water line, suggesting that plants use soil water that does not itself contribute to groundwater recharge or streamflow. Our results further show that, at 80% of the sites, the precipitation that supplies groundwater recharge and streamflow is different from the water that supplies parts of soil water recharge and plant transpiration. The ubiquity of subsurface water compartmentalization found here, and the segregation of storm types relative to hydrological and ecological fluxes, may be used to improve numerical simulations of runoff generation, stream water transit time and evaporation–transpiration partitioning. Future land surface model parameterizations should be closely examined for how vegetation, groundwater recharge and streamflow are assumed to be coupled.

Stable isotopes have been used to study large-scale changes in hydrology during the Pleistocene epoch. Many of these efforts have required an estimate of the δ18O value of runoff generated by melting ice sheets. There is no consensus on representative values. Here we examine δ18O values from fossil groundwater samples and isotope-enabled general circulation models (GCMs) to better understand the isotopic composition of the Laurentide Ice Sheet (LIS). Groundwater δ18O values ranged from −12.5 to −25.3‰ and tended to increase southward. The δ18O precipitation values predicted by GCMs follow a similar trend but increase more steeply southward. The difference in groundwater and GCM output can be explained by invoking movement of glacial ice and meltwater, along with mixing within groundwater systems. Most groundwater δ18O values are higher than an average LIS δ18O value of −25.4 ± 2.5‰ calculated based on estimated ice sheet volumes and sea level data.

The vulnerability of groundwater to contamination is closely related to its age. Groundwaters that infiltrated prior to the Holocene have been documented in many aquifers and are widely assumed to be unaffected by modern contamination. However, the global prevalence of these ‘fossil’ groundwaters and their vulnerability to modern-era pollutants remain unclear. Here we analyse groundwater carbon isotope data (12C, 13C, 14C) from 6,455 wells around the globe. We show that fossil groundwaters comprise a large share (42–85%) of total aquifer storage in the upper 1 km of the crust, and the majority of waters pumped from wells deeper than 250 m. However, half of the wells in our study that are dominated by fossil groundwater also contain detectable levels of tritium, indicating the presence of much younger, decadal-age waters and suggesting that contemporary contaminants may be able to reach deep wells that tap fossil aquifers. We conclude that water quality risk should be considered along with sustainable use when managing fossil groundwater resources. Groundwater that predates the Holocene is commonly assumed to be unaffected by modern contamination. A global analysis of fossil groundwater suggests that modern contaminants are present in deep wells that tap fossil aquifers.

Groundwater recharged by meteoric water supports human life by providing two billion people with drinking water and by supplying 40% of cropland irrigation. While annual groundwater recharge rates are reported in many studies, fewer studies have explicitly quantified intra-annual (i.e., seasonal) differences in groundwater recharge. Understanding seasonal differences in the fraction of precipitation that recharges aquifers is important for predicting annual recharge groundwater rates under changing seasonal precipitation and evapotranspiration regimes in a warming climate, for accurately interpreting isotopic proxies in paleoclimate records, and for understanding linkages between ecosystem productivity and groundwater recharge. Here we determine seasonal differences in the groundwater recharge ratio, defined here as the ratio of groundwater recharge to precipitation, at 54 globally distributed locations on the basis of 18O/16O and 2H/1H ratios in precipitation and groundwater. Our analysis shows that arid and temperate climates have wintertime groundwater recharge ratios that are consistently higher than summertime groundwater recharge ratios, while tropical groundwater recharge ratios are at a maximum during the wet season. The isotope-based recharge ratio seasonality is consistent with monthly outputs from a global hydrological model (PCR-GLOBWB) for most, but not all locations. The pronounced seasonality in groundwater recharge ratios shown in this study signifies that, from the point of view of predicting future groundwater recharge rates, a unit change in winter (temperate and arid regions) or wet season (tropics) precipitation will result in a greater change to the annual groundwater recharge rate than the same unit change to summer or dry season precipitation.

Plant transpiration is the largest continental water flux. Research now shows that climate and water availability projections are highly sensitive to the ways that plant responses to changing atmospheric conditions are represented.

An analysis of the relative effects of transpiration and evaporation, which can be distinguished by how they affect isotope ratios in water, shows that transpiration is by far the largest water flux from Earth’s continents, representing 80 to 90 per cent of terrestrial evapotranspiration and using half of all solar energy absorbed by land surfaces. Water fluxes from the land surface to the atmosphere are divided between evaporation, and transpiration from leaf stomata. Although a seemingly basic division between the physical and biological, there is still no consensus on the global partitioning between the two fluxes, resulting in uncertainties as to responses to future climate variations. Now, Scott Jasechko and colleagues use the isotopic signatures of transpiration and evaporation from a global data set of large lakes and reveal that enormous quantities of water — as much as 90% of total terrestrial evapotranspiration — are cycled through vegetation via transpiration. One conclusion to be drawn from this study is that the accuracy of biological — rather than physical — fluxes should be prioritized in work to improve climate models. Renewable fresh water over continents has input from precipitation and losses to the atmosphere through evaporation and transpiration. Global-scale estimates of transpiration from climate models are poorly constrained owing to large uncertainties in stomatal conductance and the lack of catchment-scale measurements required for model calibration, resulting in a range of predictions spanning 20 to 65 per cent of total terrestrial evapotranspiration (14,000 to 41,000 km3 per year) (refs 1, 2, 3, 4, 5). Here we use the distinct isotope effects of transpiration and evaporation to show that transpiration is by far the largest water flux from Earth’s continents, representing 80 to 90 per cent of terrestrial evapotranspiration. On the basis of our analysis of a global data set of large lakes and rivers, we conclude that transpiration recycles 62,000 ± 8,000 km3 of water per year to the atmosphere, using half of all solar energy absorbed by land surfaces in the process. We also calculate CO2 uptake by terrestrial vegetation by connecting transpiration losses to carbon assimilation using water-use efficiency ratios of plants, and show the global gross primary productivity to be 129 ± 32 gigatonnes of carbon per year, which agrees, within the uncertainty, with previous estimates6. The dominance of transpiration water fluxes in continental evapotranspiration suggests that, from the point of view of water resource forecasting, climate model development should prioritize improvements in simulations of biological fluxes rather than physical (evaporation) fluxes.