Measurement and Data Strategies
In the history of hydrologic science, many of the significant advances have resulted from new measurements (NRC, 1991). Whereas there are often field experiments with specific research objectives, the hydrologic community has not placed a high priority on design and operation of observing systems that meet both operational and research needs. In addition, rarely has high priority been given to the establishment of long-term observing systems. Indeed, over the past decade, only modest progress has been made in hydrologic measurements, and data needs are much the same as delimited in Opportunities in the Hydrologic Sciences (NRC, 1991).
The science challenges identified in Chapter 2 have in common the need to characterize better the hydrologic cycle. The challenges fall into three categories. First, information is needed about the fluxes and storage of water as it moves through the hydrologic cycle and about associated fluxes of solutes, sediments, and energy. Second, hydrologic data are needed to monitor and understand changes in the quantity and quality of water. Third, in the traditional scientific sense, hydrologic data are needed to test hypotheses and models and to formulate new hypotheses. Data are needed at a variety of scales. However, the spatial and temporal scales of available data limit our ability to investigate hydrologic variability and to detect changes. For most fluxes, a fundamental impediment to progress is how to relate point measurements to fluctuations in hydrologic variables at large spatial scales, in heterogeneous terrain, and over long periods of time.
Existing data networks should be viewed as critical elements within the USGCRP that are needed to answer its identified research challenges. These networks should be viewed as opportunities upon which to build the data sets needed in the next century to address questions regarding global change. Resolution of the questions in Chapter 2 requires new data, with measurement location, frequency, duration, and accuracy characteristics that go beyond the capabilities of existing networks. Allocation of new resources for data collection and data archiving must seek a compromise between scientific and operational needs, between ground-based networks and remotely sensed measurements, and between information on water quantity and quality. In establishing new systems or revamping old ones, the needs of operational water management as well as those of research and long-term monitoring should be considered carefully. Such foresight can obviate costly retrofitting or attempts to analyze data trends from operational systems in which the data noise may be greater than the signal. The establishment of any data system must take into account cost-effectiveness, but should do so with a long-range view.
Maintaining and Upgrading Ground-Based Networks
Research in hydrologic science relies heavily on data from operational ground-based networks, which need to be upgraded to meet evolving needs. In some cases adding relatively low-cost sensors or expanding the coverage of existing networks could provide the measurements to help meet the scientific challenges identified in Chapter 2. For example, thermistors to measure soil and snow temperature, devices to estimate soil water content, and instruments to sense snow properties are readily available but are deployed only to a limited extent as part of operational networks. In other cases, even large financial investments for incremental expansion of instrument networks will leave important questions about hydrologic fluxes and reservoirs unanswered. Thus, technical and analytical innovations are necessary to overcome the paucity of useful hydrologic data being collected and collated.
The need for long-term measurements is becoming increasingly clear in investigations of environmental change, including changes to the hydrologic cycle. Despite the increasingly recognized importance of data records of long duration, only a few dedicated organizations have successfully maintained high-quality data-collection efforts over periods of 50–200 years. Research organizations have experienced difficulties in committing limited monies year after year to monitoring a task often not viewed as research. There are also problems associated with obtaining continuous and homogeneous observations from operational networks. These networks often designed for purposes other than long-term research tend to change in response to new technology, to changing operational needs, and to budget pressures. Detecting hydrologic change requires data sets of greater quality and reliability than are often needed for research on processes or for operational forecasting. Federal agencies that make up the USGCRP must demonstrate their commitment to the collection, archiving, and dissemination of data suitable for the hydrologic research challenges identified in the FY2000 Our Changing Planet. In particular, USGCRP and its member agencies should reevaluate existing ground-based networks and consider future networks in light of the critical need to detect hydrologic change in future decades.
Most hydrologic data have been collected with local-, regional-, or national-scale questions in mind. Some of the science issues in Chapter 2 are local or regional, but others are continental or global. Because it will not be possible with known technology and present-level resources to monitor hydrologic phenomena over the entire Earth, there is a need to actively promote international cooperation to implement networks that yield the maximum return.
Under the current budgets, there is a decreasing number of measurement points that federal, state, and local agencies can maintain, and there is a limit to the number of measurements that they can produce. For example, the national streamgaging network, maintained by the U.S. Geological Survey (USGS), is a unique and irreplaceable source of primary data supporting hydrologic research, planning, and management. It is of critical national importance that this source of long-term, consistent, and reliable data be maintained and upgraded to support the measurement of both streamflow and stream stage, especially in extreme conditions such as floods (USGS, 1998; NRC, 1999a). Federal agencies must implement effective measures to reverse the trend in river flow monitoring network losses. This can be achieved only through both inter- and intra-agency commitments to making long-term measurements in support of work on recognized national priorities in water resources planning, water quality management, aquatic ecosystem protection and restoration, hazard mitigation, and global change research. Implementation plans should address stable funding partnerships to maintain the networks.
Similarly, a national network of groundwater monitoring wells is essential for groundwater recharge characterization, an important scientific challenge presented in Chapter 2. These data are also central to the identification of long-term trends related to pumping, drought, and climate and land-use change. Although the cost of making water-level measurements is often relatively low (measurements are sometimes made by volunteers), the costs of coordination, data management, and well maintenance are relatively high. Current funding is insufficient to maintain the present network, with the result that some states have been forced to drop wells from their networks (e.g., 43 wells25 percent of the total were droppedfrom the network in Wisconsin owing to decreased funding). New approaches should be explored to ensure that a long-term network of monitoring wells for both water level and water chemistry is maintained, such as the establishment of new partnerships between the federal and state agencies that manage the network. Furthermore, funding sources should be stabilized to ensure that the network is maintained.
Integration of Remote Sensing with Ground-Based Measurements
Over the past decade the ability of remote sensing to provide spatial data on rainfall, soil moisture, snow cover, snow water equivalence, and other key hydrologic variables has been demonstrated. As recognized in FY 2000 Our Changing Planet, the capabilities of remote sensing are powerful, in part because remote sensing has the unique capability of providing consistent, global coverage. Remote sensing has the potential to improve our understanding of terrestrial hydrology. However, much of this capability remains underused in terrestrial hydrologic research and applications because of a lack of investment in research and technology by the USGCRP agencies. Important technological questions remain to be answered before data with sufficient accuracy and reliability can be collected and systematically applied for measuring and detecting changes in land surface hydrologic properties. Moreover, hydrologic science has yet to articulate, or in some cases even recognize, the necessary synergy between remotely sensed and ground-based measurements. Remote sensing will. not preclude the need for ground-based measurements. However, it is not likely that existing ground-based networks are adequately configured to take maximum advantage of remote sensing measurements. Further research is required to determine the mix of remote sensing and ground-based networks to address several of the questions posed in Chapter 2.
The four areas that offer the greatest promise for satellite remote sensing to further hydrologic research and science are rainfall, soil moisture, snowpack properties, and surface-water level monitoring. Significant investment in remote sensing technology and hydrologic research is needed to prepare adequately for future missions in these areas. Two additional areas that have been initiated by other disciplines, but in which hydrologic science has an obvious interest, are vegetation properties and ice sheet mass balance.
Following the success of the Tropical Rainfall Monitoring Mission (TRMM) in measuring rainfall using a combination of active and passive microwave observations, the National Aeronautics and Space Administration (NASA) initiated planning for a global precipitation mission. Rainfall is associated with weather systems that display considerable spatial and temporal variability, requiring a sampling frequency of multiple times per day. The proposed global precipitation mission provides measurements that are central to the understanding of the predictability and variability of regional and global water cycles (Chapter 2). It also directly addresses USGCRP priorities in the global water cycle, particularly weather prediction and weather-climate relationships.
Space-based remote sensing offers the potential of measuring soil moisture, one of the main hydrologic variables that is not measured in any large spatial array of ground-based instruments. Soil moisture is the most significant indicator of the state of the terrestrial hydrologic system, and it is the governing parameter for partitioning rainwater among evaporation, infiltration, and runoff. It also provides a critical link between the physical climate and biogeochemical cycles. No existing or planned instrument meets these needs; thus, NASA has initiated studies for an exploratory mission. Soil moisture measurement is clearly a high priority within the global water cycle, and planning should be initiated for a measurement system that can effectively blend ground-based and satellite measurements.
With the launch of Terra, the hydrologic community will have the possibility of developing more accurate snow-covered-area products using the moderate resolution spectrometer (MODIS) instrument. The gain in accuracy comes from the improved spatial and spectral resolution of MODIS over the NOAA's Advanced Very High Resolution Radiometers (AVHRR) instruments. However, as with AVHRR, MODIS's surface observations will be hampered by clouds. An instrument with at least the resolution of MODIS should be made available to provide snow cover data for the foreseeable future. Research has shown that further advances in measuring snowpack properties from space will come only with a multispectral, multifrequency microwave sensor, either through active synthetic aperture radar (SAR) or through a high-resolution radiometer. Both systems offer the promise of measuring snow water equivalent, with a lower-frequency, higher-resolution SAR being preferred for measuring deep snow properties in mountainous regions, and with higher-frequency passive sensors being preferred for thinner snow packs on the Great Plains. A SAR sensor for snow water equivalence thus helps address perhaps the most significant water resources management challenge in the western United Statesthat of seasonal water supply forecasting. Over 85 percent of the Colorado River streamflow derives from snowmelt from mountainous regions, and late-spring water-supply forecasts based on estimates of how much snow is on the ground still differ from measured seasonal runoff by 20 percent to 50 percent in the various subbasins. SAR data have also been shown to be useful in estimating ice sheet dynamics. Better snow observations in the plains can help forecast floods like the record 1997 flood on the Red River of the North, and they can help in predicting moisture availability in agricultural areas on the Great Plains.
A fourth exploratory area that shows great potential is river discharge and changes in inland lake levels. Although river stage can be conveniently observed in situ for small to moderate rivers, discharge information for many of the world's major rivers is not available. Monitoring river discharge and lake storage would allow for the closure of the terrestrial water balance at continental scales and would provide important data on the linkage between the terrestrial and oceanic hydrospheres. Such monitoring would potentially yield an important indicator of global change. Precision altimetric sensors developed for oceanographic and ice-sheet measurements are potentially capable of detecting changes of a few centimeters in inland water surfaces.
In addition to these areas, continued work in quantifying and characterizing land cover changes is of critical interest to the hydrologic science community, as well as to the fields of ecological and biogeochemical research (NRC, 1999a). Terrestrial vegetation is an important determinant of hydrologic conditions through its effects on the surface radiation balance, on precipitation-runoff partitioning, and on evaporation. Remotely sensed data provide the only means by which variables such as growth form (e.g., woody vs. herbaceous vegetation), leaf type, and growing season onset and termination can be estimated at high resolution. Research and applicationscurrently using AVHRR, Landsat, and SPOT (Satellite Pour l'Observation de la Terre), and by extention MODISneed to be extended to include data from the Vegetation Canopy
Lidar (VCL) mission. The use of VCL data for biological and hydrologic monitoring, process studies, and model parameter estimation needs to be expanded, since this mission will provide unique data on vegetation structure that may identify vegetation stress related to climatic factors and landscape (or land use) change. As discussed subsequently, multivariate model assimilation schemes might be used to incorporate these and other in situ ecosystem observations with other in situ and remotely sensed hydrometeorological data sources into a consistent framework. Such assimilations would facilitate subsequent hydrologic (and ecological) process studies, as well as provide a check on the underlying physics and parameterizations of the coupled ecological-hydrologic models used in the assimilation process.
Earth satellite missions, data management, and science to support development and use of data from these missions form the centerpiece of the USGCRP (e.g., NASA's Earth Science Enterprise). Potentially, this arrangement can serve hydrology well. However, for hydrologic science and applications to get the maximum benefit from the missions, a much closer link with ground-based network design, data management, and science support will be needed. Further, the ready availability of remote sensing information poses several specific challenges to the hydrologic community: (1) classical data-poor hydrologic models designed to function (and calibrate) with sparse in situ observations of precipitation and stream discharge need to be augmented and possibly replaced with new, process-resolving distributed hydrologic models that are forced with multispectral remotely sensed (and ground-based) observations, (2) basin-scale and regional validation databases consisting of coordinated in situ and remote-sensing data collection programs need to be established, and (3) hydrologists must adjust from being passive recipients of limited remote-sensing observations to acting as a unified scientific community engaged in supporting the definition, design, and implementation of satellite remote sensing missions. The USGCRP agencies with research and applications program in hydrologic science must be proactively supportive of these goals.
Data Interpretation: Synergy in Modeling and Measurement
Advances in data interpretation capabilities are required to use the measurements effectively in the context of science questions and societal needs. Data assimilation (merging of models and data), solution techniques for inverse problems (inferring system properties from sparse data with the aid of models of the system), nested regional atmospheric models (and hydrologic models), and expanded analysis capabilities of geographical information systems (GIS) are opening new possibilities for the interpretation and use of data.
Data assimilation, an approach routinely used in meteorology and oceanography, is a particularly promising aid in the integration of large volumes of multisource and multiscale data as well as in the interpretation of measurements. It is an extension of standard model calibration. However, data assimilation is data-driven whereas calibration is model- and data-driven. Thus, one of the primary purposes of data assimilation is to produce data products that are directly useful for hydrologic analyses. For example, sequential remotely sensed data of ground temperature and microwave emissivity for the top few centimeters of soil may be used in conjunction with a soil column model to infer profiles of soil temperature and moisture down to tens of centimeters. In the context of groundwater, tracer and water level measurements in a sparse network of monitoring wells may be assimilated into a numerical groundwater model to infer soil hydraulic properties throughout the aquifer. Another use of data assimilation is to demonstrate the impact of under used and nontraditional data types (e.g., remote sensing, groundwater levels, and new in situ networks) on
hydrologic predictions. The framework may be used to analyze the value of various observing systems through so-called ''data-denial" experiments. In these experiments the value of a particular data source can be quantified in terms of the effect it has on estimates of a particular hydrologic variable. The design of new, cost-effective observing networks may be similarly done in the framework of data assimilation.
Nested regional climate models provide a bridge between the spatial scales of atmospheric and land surface processes. Understanding predictability in hydrologic systems and tapping that predictability for planning, assessments, and forecasts clearly require bridging tools that can make use of detailed spatial data to estimate spatial responses. In the western United States, capturing precipitation and runoff patterns requires topographic information of detail considerably greater than that found in even the most detailed general circulation models.
Advanced GIS tools make spatial modeling and interpretation possible. For example, assessing the impact of land-use changes on runoff or recharge requires information on the time evolution of land use as inputs to hydrologic models. GIS tools are likely to become a central part of hydrologic modeling.
Supporting Long-Term Experimental Sites
Long-term experimental sites for characterizing the water balance, flow pathways, and reactive transport provide the data essential for developing models and methods to scale hydrological variables, characterize basin-scale variability, and understand the limits of predictability. In the future they should also provide the data essential to understanding the coupling between hydrologic systems and ecosystems, and they should provide measurements for the calibration and validation of remote sensing data. The goal is to operate dozens of such basins worldwide over decades and across a range of bioclimatic zones, geological settings, and land-use characteristics. Existing program in the United States (e.g., the National Science Foundation's Long-Term Ecological Research sites; the U.S. Department of Agriculture's experimental watersheds; the Water, Energy, and Biogeochemical Budget research watersheds of the USGS; and the AmeriFlux CO2 sites partially supported by the U.S. Department of Energy and NASA) provide a useful starting point; some were designed as hydrological study sites, bit others do not have a significant hydrological component. Agencies that support these sites should encourage multidisciplinary observations and studies. International and interagency coordination activities need to be launched to augment these sites for hydrologic research in a cost-effective manner. At larger scales, activities such as the USGS's National Water Quality Assessment (NAWQA) program (Box D) are needed to monitor and to understand the trends in the quality (and quantity) of the nation's surface and subsurface water resources for both science and policy purposes. Furthermore, more sites are needed in the United States and in selected climatic zones around the world in order to characterize hydrologic processes across the Rill range of environmental conditions.
Data Accessibility and Quality Control
Advances in hydrologic science will depend largely on how well investigators can integrate reliable, large-scale, long-term data sets. Inevitably, many scientists from a variety of disciplines and backgrounds will be involved in data collection and analysis over a significant period of time. Creating effective data
systems for assembling and distributing scientific data sets is not trivial. For example, better access to data from the Next Generation Weather Radar (NEXRAD) WSR-88D Doppler weather radar network with national coverage can significantly enhance the capability of monitoring precipitation fields for operational and research applications in the United States (see Box E). Many successful data systems have been constructed within the scientific and operational communities. Despite technical and political challenges, these data systems have high scientific value.
Many hydrologic data are currently not accessible and are in danger of being lost because of poor archiving by agencies or by responsible individuals as they retire or move to new assignments. Underuse of other hydrologic data stems from problems with quality control. For example, the limited use of ground-based radar rainfall data outside of the operational environment is partially attributed to the lack of research-quality data products and partially to poor archiving practices. Hence, resolving these issues would enhance the scientific basis and the routine use of these data in hydrology. In another example, downhole geophysical logs collected during the drilling of groundwater wells provide accurate continuous records of subsurface properties and are a valuable source of information. These records are not maintained in an electronic archive dig is readily accessible and usable and they represent a tremendous missed opportunity to capture data that could have many and varied future uses. The logs, plus information regarding the location and accuracy of well logs from various sources, including commercial well drillers, need to be archived.
There is clearly a need for spatially distributed, regular information on water quality to address the priority science questions in the coupling of hydrologic systems and ecosystems through chemical cycles noted above. Although the long-term water quality monitoring efforts in the United States have been well designed to meet their own, often focused, goals, they are generally not connected, do not share data, and do
not provide a broad, comprehensive assessment of water quality at continental, regional, or watershed scales. At present there is not an adequate, readily accessible archive for water quality data for the nation. The largest water quality archive in the United States is the Storage and Retrieval system (STORET), maintained by the U.S. Environmental Protection Agency. STORET incorporates data from other governmental agencies, from volunteers, and from any other programs or agencies that choose to submit data (see Box F). In 1997, STORET received data from 67,000 sites, 1,200 of which were from the USGS networks. As a long-term archive, STORET needs to be unproved with respect to the documentation of its quality assurance and quality control procedures, and the more consistency of its reporting in order to reduce large spatial and temporal data gaps. Systematic criteria that are cost-effective for collecting and archiving valuable water quality data need to be defined and implemented as part of operational networks.