BACKGROUND AND CONTEXT
Water security is essential for the sustainability of healthy human populations, economic growth, and the stability of communities and nations (e.g., Kreamer, 2012; World Bank, 2016). With water insecurity identified as a key factor contributing to geopolitical instability in different parts of the world, knowledge and understanding of water availability is thus an integral component of U.S. national security interests (e.g., Hartley et al., 2017).
Water of appropriate quantity and quality is essential for drinking, sanitation, and food, energy, and industrial production for any society and is derived for most needs from surface- or groundwater sources. The proportion of surface- to groundwater withdrawn and used in any region depends on a set of interactive factors including the dominant uses and needs for the water (human consumption, agriculture, etc.); the topographic and hydrogeological setting; seasonal temperatures and rainfall; economic and infrastructure capabilities (e.g., for storage, pumping, industrial and agricultural development); population size and change; and water management, policy, and other local or national water agreements.
Approximately 30 percent of all freshwater on Earth exists as groundwater.1 Although surface water withdrawals make up the majority of the water used for public supply and irrigation in the United States, groundwater provides drinking water to approximately 50 percent of the global population and more than 40 percent of the water used globally for irrigation (FAO, 2017; Maupin et al., 2014). Furthermore, agriculture represents approximately 70 percent of global freshwater withdrawals (WWAP, 2014). Studies also suggest that groundwater use in irrigation globally is increasing in total volume as well as a percentage of all water used for irrigation, with the demand for groundwater resources increasing as available primary surface water supplies are depleted (Siebert et al., 2010). Particularly in arid regions, groundwater may be the most accessible water supply for any purpose, leaving groundwater withdrawals concentrated in areas that are already experiencing water stress.
Perhaps the major factors in determining the importance of the use of groundwater in any region are the rate and quantity of withdrawals relative to water recharge, which in turn, relate to the tight coupling between surface- and groundwater quantities, flows, and change through time (NASEM, 2018). Groundwater is replenished by precipitation that seeps through
the upper level of soil (unsaturated zone) but assessing the rate at which this recharge occurs is dependent upon numerous variables, including measurements of precipitation and evapotranspiration (ET) and the hydraulic conductivity of near-surface soils and sediments, which themselves are often associated with large uncertainties (NASEM, 2018).
Even in the presence of direct ground observations and measurements of the water table, quantitative evaluation of groundwater storage, flow, or recharge at different scales also requires remotely sensed data and observations applied to groundwater models. Direct remote sensing of changes in groundwater storage is currently possible at a coarse scale using GRACE (NASA’s Gravity Recovery and Climate Experiment satellites) data, but resolving the interaction of groundwater storage, flow, and recharge at a scale at which basins are managed requires additional remotely sensed data (such as precipitation), and proxy data (such as land-use/land cover change over time, surface subsidence and rebound, or other information).
In June 2019, the Water Science and Technology Board of the National Academies of Sciences, Engineering, and Medicine convened a workshop2 to explore these topics and to identify scientific and technological research frontiers in monitoring and modeling groundwater recharge and flow in various regions of the world. The goals of the workshop were to assess regional freshwater budgets under major use scenarios (e.g., agriculture, industry, and municipal); examine state of the art research frontiers in characterizing groundwater aquifers, including residence time, quantity, flow, depletion, and recharge, using remotely sensed observations and proxy data; discuss groundwater model uncertainties and methods for mitigating them using sparse ground observations or data and other approaches; and consider our ability to detect which water management strategies (e.g., water reuse, irrigation efficiencies, and desalination) that affect groundwater flow and recharge are being used and any changes in their use over time.
In addition to a series of guided panel discussions, attendees participated in small group conversations to examine how remotely sensed data can be utilized in regions where in situ measurements, observations, and instrumentation are particularly difficult. Participants considered potentially promising partnership and collaboration opportunities to help advance our understanding. This document summarizes workshop presentations and plenary discussions.3
OPENING REMARKS AND KEYNOTE PRESENTATIONS
Planning Committee Chair, Dr. Venkat Lakshmi (University of Virginia), opened the workshop and emphasized the importance of water availability and water stability, especially as it relates to U.S. national security. Many countries may already be experiencing water stress that could be exacerbated by groundwater withdrawals. In fact, many hydrological modelers now recognize the importance of including groundwater in their models as well as the connec-
2 This Proceedings of a Workshop was prepared by a workshop rapporteur as a factual summary of what occurred at the workshop. The planning committee’s role was limited to planning and convening the workshop. See the appendixes for the Statement of Task, planning committee biosketches, workshop agenda, participant list, and speaker abstracts.
3 Recordings of the workshop presentations and discussions can be viewed here: https://www.youtube.com/playlist?list=PLi6VVotVxseDqVsWoRLJn7xHRPCMQl2WY.
tions between groundwater and other parts of the hydrologic system. Dr. Lakshmi noted that remotely sensed and in situ observations can provide crucial information to ensure that these connections are made. Remotely sensed data, in particular, can be utilized in regions where it may be especially difficult to obtain in situ measurements.
Given the implications for U.S. national security, the National Geospatial-Intelligence Agency (NGA) is interested in learning more about groundwater and how changes in groundwater may affect key regions of the world. Dr. Tony Nguy-Robertson (NGA) shared insights from the perspective of the workshop sponsor, and noted that adequate support of the defense and intelligence communities requires geospatial intelligence to provide crucial information about the environment and to aid in decision making. NGA has a keen interest in international water security and water quality, especially to provide support and information to policy makers. Dr. Nguy-Robertson also noted that increased understanding of potential overuse of water in some individual nation-states will be important for future water resource stability and potential risks to aquifers. NGA is interested in ensuring that water is available for uses such as personal consumption, agricultural needs, and industrial uses. In addition, there is a great deal of interest in increased understanding of groundwater resources to support U.S. Department of Defense operations and to ensure adequate water supplies for military operations and diplomatic missions, including humanitarian and disaster relief support teams. Dr. Nguy-Robertson noted that NGA support for warfighters, humanitarian assistance workers, and policy makers relies on improvements in estimating groundwater recharge and flow to anticipate future crises and to provide operational support.
As noted in the opening remarks, people throughout the world rely on groundwater as their primary source of freshwater, however groundwater is being depleted at a significant rate. In fact, trends show that global water demand is increasing at the same time that groundwater is being depleted (see Figure 1.1; Wada et al., 2010). Irrigated agriculture is one of the major causes of this depletion (Dalin et al., 2017), and drier areas of the world tend to rely more on groundwater and thus have larger fractions of depletion. To address how remote sensing can play a role in increasing our understanding of these issues, keynote speaker Dr. Matthew Rodell (NASA Goddard Space Flight Center) provided a historical perspective on data availability. He noted that in the 20th century, almost all data originated from in situ observations such as well monitoring, well logs, ground penetrating radar, and other geophysical methods and geological mapping techniques. In addition, older groundwater models were fairly crude and limited to local scales. However, useful information can still be obtained from these in situ observations (see, for example, the discussion in Li et al., 2015, on groundwater variability). In the United States there are a number of wells that are useful for climate monitoring, as indicated by the U.S. Geological Survey (USGS) Groundwater Climate Response Network, however long-term and reliable in situ data records are sparse. Although most countries have at least some groundwater data, often they are either not digitized, not centralized, or not made publicly available. In fact, there are only 17 countries that contribute to the Global Groundwater Monitoring Network. Therefore, scientists face coverage gaps both spatially and temporally as well as delays in data availability and issues with measurement consistency. In many cases, political restrictions exacerbate this problem; wells may be monitored, but the data are not made available.
Dr. Rodell commented on the expanded use of remote sensing in the 21st century and the exciting new insights enabled by these technologies, particularly satellite gravimetry and airborne and satellite interferometric synthetic aperture radar (InSAR). He also noted the importance of the development of regional to global land surface models with groundwater budget and 3D flows. Scientists have been able to explore atmospheric coupling and data assimilation, where models are constrained using available observations. Several NASA Earth observing satellites are highly relevant to understanding the water cycle and to conduct groundwater studies. GRACE and GRACE Follow On (GRACE FO) missions are particularly valuable for groundwater research. While most remote sensing missions rely on radiation-based approaches (i.e., measuring light that is either emitted or reflected from the surface to estimate quantities like snow cover, vegetation type, ice, rainfall, or soil moisture) and are only able to penetrate the first few centimeters below the surface, GRACE and GRACE FO are unique in their ability to monitor water at all levels, down to the deepest aquifer. GRACE (2002-2017) and GRACE FO (2018-present) are twin satellite missions that rely on an understanding of how heterogeneities in Earth’s gravity field perturb the orbits of the satellites; precise measurements of the distance between the twin satellites (one following the other about 200 km apart) allows researchers to infer temporal changes in the gravity field, which can be directly related to mass redistribution. On monthly to multi-annual time scales, the largest sources of mass redistribution are ocean circulation and tides, atmospheric circulation, and changes in terrestrial water
storage. By modeling and removing the first two, changes in terrestrial water storage (i.e., the sum of groundwater, soil moisture, surface waters, snow, and ice) can be estimated. Using this type of data, Dr. Rodell illustrated how anomalies of terrestrial water storage (the departure from the long-term mean at a given time and location) and their changes over time can be calculated globally (see Figure 1.2). Using auxiliary data and/or models, groundwater storage changes can then be isolated from the terrestrial water storage data. He noted that GRACE cannot, however, provide information about the total amount of water in an aquifer, it can only give information about how it is changing over time. Although long-term trends in terrestrial water storage can be quantified using these methods, it can be challenging to determine if they are caused by natural interannual variability, water management (e.g., groundwater pumping), reservoir filling, climate change, or some combination of those.
Some of the challenges associated with satellite gravimetry highlighted by Dr. Rodell include low spatial resolution (roughly 150,000 km2 at mid-latitudes), monthly temporal resolution, data latency of up to a few months, and lack of vertical information (because this method reveals the sum of all water components, not measurements of individual components such as groundwater, soil moisture, snow, or surface water). One way to address the issues of low spatial and temporal resolution could be to have multiple pairs of GRACE-like satellites, however, advanced technologies would still be needed to obtain orders of magnitude higher spatial resolution, he stated. Some new technologies and tools that are being studied include laser interferometry paired with lower altitude and drag-free spacecraft, cold atom gradiometry, and data assimilation methods to push these boundaries and help obtain higher resolution data. InSAR technology relies on satellite measurements of changes in the elevation of the Earth’s surface. When the groundwater level changes in an aquifer, the response of the land
surface can be tracked over time. Dr. Rodell noted that this method can be used to estimate dynamics of aquifer systems, including how much of the groundwater is removed, though this can be a challenging process. The advantage of using this technology is high spatial resolution, but the non-elastic aquifer response (i.e., an aquifer may not completely recover from the compaction and subsidence caused by de-watering) can create challenges.
Dr. Rodell pointed out that the recent Decadal Survey for Earth Observation from Space (NASEM, 2018) highlighted several NASA missions including the Surface Water Ocean Topography (SWOT) mission, which will measure surface water elevations and estimate rates of river discharge. Among the many references to hydrology in the 2018 Decadal Survey (as well as the 2007 Decadal Survey; NRC, 2007), the report also recommended a mission to increase our understanding of aerosols, clouds, convection, and precipitation (crucial for hydrology research); a mission to understand surface deformation and change; and a mass change mission (a follow on to GRACE FO). He emphasized that there are many other observing methods such as sensors on commercial aircraft, sensors on the space station, and ground-based citizen science technologies, for example, that can be used to monitor the water cycle.
Land surface models use a system of physical equations to understand the interaction of energy, momentum, and mass between the surface and the atmosphere. Dr. Rodell illustrated the high spatial and temporal resolution provided by model parameters and other inputs such as precipitation and solar radiation. GRACE data can then be used to constrain the model via data assimilation, resulting in a product with higher resolutions than GRACE and better accuracy than the model alone. More advanced groundwater flow models such as ParFlow and PCR-GLOBWB can simulate 3D groundwater flow. Groundwater modeling requires observations such as precipitation, soil moisture, surface water levels, streamflow, terrestrial water storage, snow cover, evapotranspiration, land surface temperature, and vegetation cover, but additional observations could help make further improvements in modeling: flow velocities, root-zone soil moisture, terrestrial water storage at 100 km2 resolution, snow water equivalent, and hydrogeological parameters (e.g., aquifer extent and depth, permeability, and specific yield). Integrating this data into land surface models can help with downscaling, disaggregating, and interpreting groundwater-relevant observations from disparate sources.
Although remote sensing is clearly a powerful tool in improving groundwater research, in situ data can also be used to understand changes in groundwater management. Keynote speaker Dr. Holly Michael (University of Delaware) noted that recent advances in spatial data sets such as bulk aquifer properties, recharge, discharge, and projected changes are enabling new insights in process understanding, vulnerability identification, and targets for mitigation on regional and global scales. In terms of practical management on a basin scale, in situ data are needed to supplement and ground-truth coarser data sets and estimates. Three examples of in situ research were used by Dr. Michael to highlight some of the associated challenges and transboundary problems: (1) conjunctive use management4 in the Upper Ganges Basin, (2) arsenic contamination in the Bengal Basin, and (3) mega-city pumping in Dhaka, Bangladesh.
The first example highlighted by Dr. Michael illustrates the conjunctive use of groundwater and surface water in the Ganges Basin of India and Bangladesh. This is an example of
4 Conjunctive use of surface and groundwater consists of harmoniously combining the use of both sources of water in order to minimize the undesirable physical, environmental and economical effects of each solution and to optimize the water demand/supply balance.
a transboundary water quantity problem that also suffered from a lack of data availability. The enormous size and complexity of the Ganges Basin means that there are many surface water diversions that have international policy implications. An international treaty plays an important role in governance of water resources in this region, especially given the seasonality of the hydrologic changes (i.e., flooding in the monsoon season and water diversion in the dry season, see Figure 1.3) as well as problems with both waterlogging and groundwater depletion. This has serious implications for water availability for both people and ecosystems. This study included numerical modeling to assess hydrologic effects and economic analysis, which required data for model boundary conditions and forcing (recharge rate, river width, distance between rivers, and irrigation return flow); data for model parameterization (hydraulic conductivity, specific yield, and connectivity between river beds and canal beds); data for model calibration (in situ hydraulic head, river discharge, and river levels); and data to understand the economics (energy source and cost for pumping, pump efficiency, infrastructure cost, and dry season river discharge). Some of this was available directly (e.g., river discharge), but most of the information was aggregated from indirect sources (such as literature and government reporting) or informed guesses and estimates. In some cases, no data were available at all (e.g., in situ hydraulic head and river levels).
As noted by Dr. Michael, four scenarios were considered in this study: (1) the current state of surface irrigation, (2) the Ganges Water Machine as described above, (3) pumping
along canals, and (4) distributed pumping and recharge. Simple models and large-scale sensitivity analyses were used to incorporate variability across the basin and uncertainties in many of the parameters. The study looked at the potential for reduction in monsoon season river flow, water that would be made available for irrigation, and pumping costs for the different scenarios. It became clear that the Ganges Water Machine was not the most viable scenario, even when the uncertainties due to lack of data were taken into consideration (Khan et al., 2014). Therefore, this example illustrated that informative results can sometimes be obtained even in the absence of data.
The second example of a transboundary and water quality problem discussed by Dr. Michael concerns arsenic contamination in the Bengal Basin. Some data were available in this case, but they were sparse and low-quality. There is naturally occurring arsenic in shallow groundwater in this region and more than 150 million people rely on the groundwater for drinking and irrigation; many are at risk for serious health effects due to the arsenic. However, the level of arsenic concentration decreases dramatically at about 150 m well depth. Therefore, deep wells have an important role to play when considering mitigation options.
Dr. Michael’s research examined the sustainability of this deep groundwater resource, if it is pumped, to determine whether the arsenic would migrate to deeper levels. A model was developed to test scenarios, and the associated data needs were: information for model boundary conditions and forcing (topography, domestic pumping, and irrigation pumping); model parameterization (hydraulic conductivity); and parameter estimation (in situ hydraulic head, groundwater age, driller logs, and river discharge). There were difficulties collecting these data through both governmental and non-governmental sources in Bangladesh and India. Data were poor quality, missing key pieces, and difficult to get. Navigating these difficulties and acquiring needed data required collaboration and plenty of time to meet with relevant agencies.
Again, in this case, Dr. Michael noted that direct data were supplemented by information aggregated from indirect sources and informed guesses and estimates. No data were available for river discharge, and problems with the data included information only from Bangladesh, not India; elevations very different from the topography; only shallow wells available; only four data points available for deep groundwater ages; and quality control issues related to different drilling methods and interpretations. Two management alternatives were considered: (1) deep irrigation and deep domestic pumping and (2) shallow irrigation and deep domestic pumping. With the split pumping scheme, 90 percent of the area was sustainable compared to 14 percent sustainability with all of the pumping occurring deep in the system (Michael and Voss, 2008). Therefore, the researchers in this study suggest regulating the indiscriminate abstraction of deep groundwater and using it for domestic pumping rather than irrigation. This illustrates how sparse, low-quality data can still be used to provide high-level management guidance.
The third example illustrates large scale pumping of groundwater in Dhaka, Bangladesh, and Dr. Michael noted that this is a combined quantity and quality problem, with high-quality data available for the research. Huge populations in mega-cities like Dhaka require increasing amounts of water and this results in massive groundwater extractions. These types of “mega-depletion” cases are occurring around the world, not just in Dhaka (see Figure 1.4; Werner et al.,
2013). Key considerations include the effects of extreme pumping on water supply and water quality as well as investigating if this water use is worsening the arsenic problem. For example, are there implications associated with large amounts of pumping and impacts on contaminant transport in the aquifer outside of the city, potentially contaminating deep wells? This problem is difficult because of the complexity of the transport and geochemistry associated with arsenic movement.
In this final case study, Dr. Michael outlined an approach using a detailed model within the basin-scale framework to understand vulnerability of the deep groundwater surrounding Dhaka to arsenic migration. Many of the data needs are similar to the cases noted above, but in this case much more direct and better quality data were available to the researchers from their own wells and loggers, partly due to the smaller area of study. Though this is a serious issue for the region over the next decade, historically data collection by local governments is sporadic. This is a concern for modeling and managing these issues in the near term.
The resulting modeling and observations revealed a very complex hydrological system with a corresponding heterogeneous sub-surface system, as explained by Dr. Michael. When looking at the probability of contamination in 200 years, the models revealed some areas that would likely be contaminated, but the uncertainty is large throughout the study area (Khan et al., 2016). Based on the results of this study, Dr. Michael’s team found that it is important to incorporate geologic data, and that monitoring is critical to protect the health of the population that relies on this water for drinking. The existence of better quality data in this case allowed for increased understanding, even with the large uncertainties. However, increased availability of in
situ sub-surface data is needed to constrain the model simulations, and could lead to improved large-scale analysis including multiple processes.
In an ideal situation, Dr. Michael noted that a global database of fully quality-controlled and densely spaced in situ hydrologic, geologic, and geochemical data would be made publicly available in real time; but challenges associated with cost, international data sharing, technology transfer, infrastructure, quality control, and several other factors should first be addressed.