The National Acid Precipitation Assessment Program
The Acid Precipitation Act of 1980 (Public Law 96-294) established a comprehensive 10-year research program to achieve the following purposes:
to identify the causes and sources of acid precipitation [defined as "wet or dry deposition from the atmosphere of acid chemical compounds"];
to evaluate the environmental, social, and economic effects of acid precipitation; and
based on the results of the research program established by this subtitle and to the extent consistent with existing law, to take action to the extent necessary and practicable (A) to limit or eliminate the identified emissions which are sources of acid precipitation, and (B) to remedy or otherwise ameliorate the harmful effects which may result from acid precipitation.
The terms of the statute established an Acid Precipitation Task Force, of which the Secretary of Agriculture, the Administrator of the Environmental Protection Agency (EPA), and the Administrator of the National Oceanic and Atmospheric Administration (NOAA) were designated as joint chairpersons. Membership of the task force also included one representative each from the Department of Interior, the Department of Health and Human Services, the Department of Commerce, the Department of Energy, the Department of State, the National Aeronautics and Space Administration, the Council on Environmental Quality, the National Science Foundation, and the Tennessee Valley Authority, in addition to the
directors of the Argonne, the Brookhaven, the Oak Ridge, and the Pacific Northwest National Laboratories, as well as four additional members appointed by the President. The four National Laboratories were designated as a research management consortium to carry out a comprehensive and coordinated research plan, and the Administrator of NOAA was designated as the director of the overall research program.
In view of the complexity and extent of the 10-year National Acid Precipitation Assessment Program (NAPAP) study and the limitations of the committee's time to devote to this task, the committee decided to concentrate its review efforts on data management aspects of the Aquatic Processes and Effects portion of the total program. In addition to the documents specifically cited below, the committee reviewed a number of other publications relevant to this case study (NAPAP, 1990a,b,c,d,e,f,g; 1991a,b; Oversight Review Board of the NAPAP, 1991; Rubin, 1991; and Rubin et al., 1992.)
VARIABLES MEASURED AND SOURCES OF DATA FOR THE AQUATIC PROCESSES AND EFFECTS PORTION OF NAPAP
The highly diverse data and information needs for the Aquatic Processes and Effects part of the total study are summarized in the National Acid Precipitation Assessment Plan under two topics (Interagency Task Force on Acid Precipitation, 1982):
first, the chemical alteration of water quality, including ground water, drinking water supplies, streams, and lakes; and secondly, the effects on the species and populations that make up biologically productive components of aquatic ecosystems. The information needs on water quality effects from acid precipitation concern regional trends, factors affecting watershed tolerances, the chemistry of metal mobilization, modeling, and related dose/response relationships for watersheds, lakes, and streams, and the risk associated with effects on drinking water.
Research components designed to obtain the needed information are presented under the following headings:
Monitoring National and Regional Water. "In addition to the water chemistry, factors to be documented should include: weather and acid deposition records; air trajectory data and the frequency of lightning (a natural nitrate production mechanism); soils, geology, and land use in the watershed and upwind areas; and watershed management trends that could affect the acid neutralizing and buffering capacity of the vegetation and soil."
Determining Factors That Control Lake Susceptibility. Analyses of "lake/environment relationships will indicate the relative importance of hydrogen ions from precipitation and dry deposition, relative proportions of nitrate and sulfate inputs, soil-chemical processes, predominant vegetation, and bottom sediment characteristics."
Determining the Relative Contribution of Nitric and Sulfuric Acid Inputs. "Studies will be undertaken to determine the relative contribution of nitrogen and sulfur from acid deposition to the productivity and/or acidification of aquatic ecosystems."
Evaluating the Significance of Mobilization of Toxic Metals. "Analyses will be made of the extent to which metal contamination in drinking water, food crops, and fish is due to acid deposition and subsequent leaching and mobilization of metals."
Modeling Watershed Dose/Response Relationship. "Attempts will be made to develop simple empirical models relating the readily measured chemical characteristics of lakes and streams to atmospheric deposition." "Relatively detailed simulation models of the acidification process and its effects will be developed and evaluated." "The goal of this research will be to have the most complete, quantitative long-term dose/response models evaluated fully and compared with the more empirical field relationships now in use."
Studying Acidification of Drinking Water Sources. "Analyses will be made of historical records and current data from public drinking water systems, whether using ground water or surface water reservoirs, to determine whether pH or potentially significant metal concentrations have changed during the past 10 to 30 years. Where acidification is found, the chemistry of water supply lines will be studied and estimates will be made of the possible impact on human and livestock populations."
Monitoring Drinking Water and Evaluating Treatment Methods. "Investigations will be made of how much effect chemical treatments, such as lime or other alkaline solutions, have on the acidity of surface water or ground water sources of drinking water. The possible short- and long-term usefulness of this ameliorative approach on human health will also be determined."
Monitoring Regional Trends in Biological Effects. "Scientists will seek to identify lakes and streams believed to have been affected by or apparently tolerant to acid deposition. Information on fish-eating birds of prey and furbearing mammals also will be sought."
Studying Watershed Productivity. "Measurements will be made of progressive changes in: (1) the chemistry of the open-water system and sediments; (2) the types and numbers of surface, subsurface, and bottom-dwelling insects, plants, animals and micro-organisms; and (3) terrestrial productivity (using predictive models when necessary). Efforts will be
made to establish correlations between the chemical properties of the water or lake sediments and the populations and reproductive success of the various organisms.''
Identifying Vulnerable Growth Stages. "Field and laboratory experiments will be conducted with aquatic animals, plants, and micro-organisms to identify times of reproduction and stages of growth that coincide with episodes of strong acid inputs."
Studying Metal Contamination of Fish. "Analyses will be made of historical records, fish samples, and trophy fish to determine if concentrations of toxic metals in fish have changed over time."
Analyzing Mitigation Strategies for Acidified Lakes. "Experiments will include the application of various types of acid-neutralizing materials, such as powdered lime, rock limestone, and organic or inorganic materials that would bind or inactivate toxic metal ions."
The coordinating agency for this research was EPA. Other participating agencies included the Department of the Interior, the Department of Agriculture, and the Tennessee Valley Authority. Of the above research subject areas, numbers 1, 2, 3, 5, 6, 8, 9, and 10 were accorded priority 1, and the remainder (4, 7, 11, and 12) were priority 2.
As documented in the National Acid Precipitation Assessment Plan, the task force agreed to the following criteria for assigning research task priorities:
Priority 1—Urgently needed research of the highest priority. Timely conduct of these research tasks is necessary to answer critical scientific questions concerning acid deposition. Each task investigates a crucial question where no or inadequate similar research is underway. The economic or social value of the potentially affected resources is high and the geographical area of investigation is highly sensitive to or heavily affected by acid deposition.
Priority 2—Research that addresses an important information need but is less urgent than Priority 1. The phenomenon or geographical area to be investigated is believed to be moderately sensitive to acid deposition. The economic or social value of the affected resource is high.
Using these definitions, the coordinating agencies together with the participating agencies recommended, and the task force approved, the research priorities identified above.
EPA divided its Aquatic Processes and Effects portion of NAPAP into three major projects, entitled the National Surface Water Survey (NSWS), the Direct/Delayed Response Project (DDRP), and the Episodic Response Project (ERP). The committee's review of data interfacing activities focused primarily on the NSWS and DDRP. NSWS included these elements:
A survey of water chemistry in a statistical sample of almost 3,000 lakes and streams representing a population of 28,000 lakes and 200,000 km of streams in acid-sensitive regions of the United States.
Studies of watershed geochemical processes, deposition rates, fish toxicity, and temporal variation in lake and stream chemistry.
Analysis of long-term chemical data, fishery records, and lake sediments to document historical changes in surface water chemistry.
DDRP's overall purpose was to characterize geographic regions by predicting the long-term response of watersheds and surface waters to acid deposition. The regions selected for study were chosen from regions with surface water that have low acid-neutralizing capacity and that exhibit a wide contrast, both in soil and watershed characteristics and in levels of acid deposition.
An additional biological assessment, the Episodic Response Project, was subsequently incorporated into the Aquatic Processes and Effects portion of NAPAP. This occurred well after the design of both NSWS and DDRP, when it became apparent that additional biological measurements would be necessary to achieve the NAPAP goal in this research area.
MAJOR CONSIDERATIONS IN EVALUATING THE DATA MANAGEMENT ACTIVITIES OF THE AQUATIC PROCESSES AND EFFECTS PORTION OF NAPAP
Two major issues emerged under users' needs: identifying the users at the inception of the research and monitoring project, and understanding users' requirements.
The identification of the primary users was clear for the National Surface Water Survey (NSWS) because the whole project stemmed from a question asked by then EPA Administrator William Ruckelshaus regarding the status of acid-sensitive surface waters in the United States. Therefore, from the inception of the project, the principal user was clearly the administration of EPA at a very high level. Other portions of NAPAP included: terrestrial effects, effects on materials and cultural resources, visibility effects, economics, and atmospheric transport and deposition. Of course, the output of the Aquatic Processes and Effects portion of the program was also planned to be an input to the overall NAPAP integration synthesis.
For the lake survey portion of NSWS, research managers were particularly effective in maintaining good communication with the primary
user identified at the beginning of the project. Although there was a notable exception, as discussed below, the designer/manager of the project focused only on identifying or answering the general question asked by EPA. Because some opposition to this approach was voiced, additional sharply defined questions were asked by the EPA Administrator: how many acid-sensitive lakes and streams are there, how sensitive are they, and where are they located? In contrast, some scientist wanted more longer-term process data and more detail on single ecosystems.
The committee concludes that the NSWS portion of NAPAP was successful in answering these questions and in providing useful data and information to the primary user. The NSWS director avoided vague mandates and tried to be as specific as possible in defining goals. In addition, desirable interactions between scientists and policymakers were maintained throughout most of the program. One great strength was the continuity of scientific project leadership throughout the program—the project leaders knew the program goals and stayed with them. Perhaps even more important was the relatively stable, high-level support for NSWS within EPA.
The above comments concerning the continuity of project leadership and the stable high-level EPA support apply equally well to the follow-on study on critical watersheds, the DDRP. The primary user for this study was the relevant program office in EPA. The EPA staff assured the committee that the question that DDRP was trying to answer came from the EPA Administrator Ruckelshaus and his concern for what the future would bring; namely, what types of systems were vulnerable, where did they occur, and how would they respond under various emission control scenarios (including a no-action option)?
These DDRP questions were translated into watershed-level and process studies by the EPA research and design team. This was a very complex set of objectives. The success of the DDRP in answering these questions was less obvious to the committee than in the case of the NSWS. The users here included both some scientists working on the project and the policy- or decision-making staff of EPA. Although there may have been some conflicts between EPA policymakers and the scientists, it seems clear that the overall study was designed so that the primary data users would be policymakers at EPA.
There are two key areas related to study design: conceptual models and methodological considerations. In general, these two areas cannot be independent and must be mutually supportive.
In the initial NSWS sampling, no conceptual model for the ecosystem was apparent. The choice to measure acid neutralizing capacity (ANC) was based on a model of how lake chemistry works, and expert groups were used to determine variables that would be measured. Although major cations and anions were analyzed, ANC turned out to be the key variable not only in the survey of lake sensitivity, but also of the overall NAPAP, for several reasons. First, although ANC is an aquatic ecosystem variable, it integrates conditions in the watershed and is itself a function of various terrestrial processes (including processes of the soils, biota, and parent bedrock). Second, focusing on a single variable aided in briefing policymakers because they could understand and use the data relatively easily and were therefore likely to continue their support of the program. Third, ANC is now universally recognized as the key variable indicating acid sensitivity for aquatic ecosystems. It is thus to EPA's credit that it recognized early the importance of ANC.
The success enjoyed with ANC may not be easily translated into lessons learned for other complex programs, however. Multiple pollutants and impacts from various pathways may preclude an easy focus on a single ecological parameter.
Identification of a planned statistical analysis seemed to be the first priority in the NSWS project's experimental design. This approach was successful because the questions asked by the user and the background knowledge of the design team meshed well. The resultant probability sampling for NSWS was considered to be the most desirable approach by the policymakers at EPA, as well as by the committee. This approach facilitated follow-on programs, such as EPA's Environmental Monitoring and Assessment Program, and helped to maintain political support for the NSWS program. Two aspects of the NSWS project design process are worthy of note: (1) the data users' views were sought early in the design of the study, and (2) policymakers were strongly influencing the direction for this program and future EPA programs.
The use of conceptual models in the design of Aquatic Processes and Effects watershed-intensive studies was much more evident. Here the questions were more complex, and the designers took a much longer time to review existing conceptual models and develop new ones. Reliance on a dispersed stochastic sampling design was not feasible, and the designers relied much more heavily on the use of deterministic and conceptual models, both in the design and in the interpretation of the collected data. In addition, expert groups were used to help select where and what actual measuring points needed to be sampled. It seems to the committee that many of these experts relied heavily on historical data sets. Because it was not possible to monitor every water body, models were required that
would permit statements to be made about many water bodies on the basis of limited data.
In the watershed parts of the Aquatic Processes and Effects studies, EPA researchers did have to integrate multimedia data, including data on atmospheric deposition, watershed, soils, and surface water chemistry. Because of this need, they also were more dependent on a good conceptual model, not only in the planning stages but, more importantly, in their final assessments, which made projections and predictions on a regional basis. The committee concludes that the technical aspects of these parts of the project were performed successfully.
There was a fortuitous aspect to the watershed parts of the studies that was key to the successful turnaround and interpretation of data. Sulfate concentrations did not vary appreciably with time (seasonally), and so extensive spatial data could be used in assessments without expensive temporal characterization. It was to EPA's credit to realize (and document) this circumstance early on and to take advantage of it in both the conceptual and the practical design of the project.
Unfortunately, the biological design portion of the Aquatic Processes and Effects part of NAPAP had more difficulty from a conceptual and budgetary standpoint than the physical/chemical sampling and measurement tasks. For example, some scientists who helped plan the biological sampling efforts believed that too much emphasis was placed on the physical/chemical parameters in the initial design and not enough on the biological needs. In particular, scant attention was paid to the selection of chemical parameters that were most important for understanding biological impact, and funds were limited for implementing biological measurements. While NAPAP had excellent experimental data for fish response to acidification (a well-focused impact) with insights on interactions with calcium and pH, there was an insufficient effort to collect new biological data from the field that were integrated with concurrent measures of physical/chemical parameters deemed important in the earlier aspects of the aquatic studies. Apparently, some adjustments for this deficiency were made midway through the biological design portion.
The methodological considerations reflected some of the strengths and weaknesses associated with the conceptual framework for the three parts of the Aquatic Processes and Effects portion of NAPAP reviewed by the committee. For the NSWS, the method of ANC determination in terms of sample collection, preservation, and laboratory analysis had to be developed and tested; the method is now consistent, accurate, precise, and regularly applicable to a wide range of aquatic ecosystems and habitats. Thus, from a procedural view, the focus on ANC in the NSWS was a considerable asset in terms of managing data of known quality and for later data integration activities.
In contrast, the methods for biological assessment frequently were a liability to the success of the program. The hydrogen ion and metal concentrations (especially aluminum) of surface waters are more relevant to determining acid rain impacts to fisheries. The pH is difficult to measure in either the field or the laboratory for water of low ionic strength (typical of acid-sensitive ecosystems). Methods of measuring metals have to take into account chemical speciation and dissolved versus particulate fractions; both the instrumentation and the methods for metal determinations are more complex and costly than for ANC. The most significant methodological problem, however, is in the sampling of fish populations. A variety of different methods, including the use of seines, nets, electroshocking gear, and poisons (e.g., rotenone), are typically employed for fisheries work. The efficiency of each method may differ with fish species, age class, habitat, and the field personnel; further, there is no consistent approach or regular coordination from study to study in the use of these various methods. Given NAPAP's dependence on existing fish data from state management agencies, coupled with these methodological problems (and the lack of good methods documentation), it is not surprising that data management and integration were problematic.
The effect of methods on the watershed part of the NSWS might be viewed as being somewhere between these two extremes. In essence, both a hydrological and a chemical budget (e.g., inputs, transformations, outputs) had to be measured for a given subcatchment for this part of the study. Although many more variables were measured than for the lake survey component of NSWS, both field and laboratory methods were more standardized than they were for biological measurements, or at least could be agreed upon; also, it appeared that some complex processes, such as water movement or ion exchange in soils that are difficult to measure, could be simplified for assessment purposes.
DATA MANAGEMENT AND INTERFACING IN THE AQUATIC PROCESSES AND EFFECTS PORTION OF NAPAP
In the following discussion, no attempt is made to treat systematically the specifics of all the various types of data collected in the Aquatic Processes and Effects program, or the specifics of the relationships between or among the data types. Similarly, the specifics of the data management system, which was an ad hoc system assembled by the contractor to provide verified data summaries in forms most available and helpful to the researchers, are not described in detail. Instead, the committee provides a summary of its most important observations dealing with problems related to data management and interfacing.
The committee identified many data management and integration
issues associated with this complex research project. Among the most significant related to the actual procedures of how and by whom the data were stored and manipulated. The data management was subcontracted at a different site. There the data were verified and placed in a data management system and then distributed to the researchers.
Several issues arose with this arrangement. First, there was conflict over "ownership" of the data and timely return of the data to the scientific and technical team in EPA, most members of which were located at a different site and institution than the data management team. Briefings by both teams provided the committee with a number of insights that could help in the design and management of future interdisciplinary projects. The scientific and technical team left the impression with the data management team that the latter team's primary role was to provide a workable database from which the scientific and technical team and its subcontractors could do the actual data analysis and interpretation. However, data management team members considered themselves scientists as well as data managers and wanted the opportunity to interpret the data also. At stake was the issue of scientific recognition and credibility. According to the scientific and technical team, this issue had been equitably resolved by a series of early agreements governing data use and publication rights. Nevertheless, it was apparent that this matter was not viewed in the same light by the data management team even several years after the program was over. Without having the opportunity to work with the data, the data managers considered themselves inhibited in their ability to write and produce peer-reviewed publications describing their work.
Second, there were inordinately long delays in acquiring data from the chemical analysis laboratories. Such delays are a chronic problem in large-scale environmental sampling efforts that depend heavily on chemical analysis. EPA was able to correct this problem toward the end of the project by using a management tracking system and having much of the data verification done by the data management team. Long delays complicated matters because analytical problems could continue unchecked for quite some time, thus compromising data quality. EPA researchers eventually automated their audit program (e.g., checking variances and means plus outliers) and could notify analytical laboratories quickly to correct problems.
Third, the spatial location for the sampling data (deposition, soils, watershed, water chemistry, fishes) that were to be integrated for assessments illustrates another concern. In NAPAP, it appears that multimedia data were collected in relative physical proximity to each other, thus facilitating integration for a site or ecosystem. Designers of new research and monitoring programs such as EPA's Environmental Monitoring and
Assessment Program should be aware that if different media are randomly sampled independently of each other (different laboratories or institutions may have different media and may randomly select their own independent sites), data analysis and integration across media may be complicated.
Despite the difficulties summarized above, the database management system for the NSWS appears to have been well planned. EPA especially wanted it to be of a known quality. Both NSWS and DDRP had extensive quality control. The quality control measures included traditional quality control methods and appeared to be applied also to the metadata collected. The Quality Assurance Program was peer-reviewed, the database was verified (poor data were flagged), quality assurance (QA) data and metadata were included, and a database dictionary was put together. A number of peer-reviewed articles were published from NSWS, and the data sets were made readily available on diskette to a variety of other users, who have made numerous requests.
The database management team appears to have grappled with the question of how much QA is enough. Although they indicated that dedicating about 10 percent of fiscal resources may have been reasonable, in reality they may have spent about 20 to 30 percent of their funding on QA in some cases, based on some committee members' direct experience with NAPAP protocols. The database management team indicated that good QA early on is important, and they seemed to have pioneered new ground in this area, by solving the problem of high nitrogen in field blanks (from washing filters with nitric acid!) and developing natural audit samples that were much more useful for problems in limits of detection. In addition, by maintaining flexibility in QA, they were able to identify problems as they arose and to deal with them effectively.
It was more difficult to evaluate data management for DDRP because the final product was not yet available at the time that the committee conducted its case study. In general, the DDRP database is being developed along the same lines as the NSWS model, although with more complicated statistical analyses and including a data dictionary. The same contractor was used for managing both DDRP and NSWS data. The advantage was that the DDRP had the benefit of experience from NSWS, which provided good continuity in the program. The disadvantage was that logistics were complicated. Sites were in the East, project management was at Corvallis, Oregon, soil analysis was conducted by contract laboratories, and the data managers were at the Oak Ridge National Laboratory in Tennessee. Under this arrangement, communications were extremely difficult. A major decision for future studies was to conduct data management "in-house" to facilitate logistics and communications if the staff and hardware were available.
The data manager whom the committee interviewed provided some practical views on aspects of data management within the aquatic parts of NAPAP. He emphasized that data management should be about 10 percent of the total project budget, and this guideline apparently has been followed. His comparison of NSWS and DDRP was insightful: raw versus synthetic data; differences in management style; homogeneous, focused data sets versus complex multimedia data; and straightforward reporting of results (ANC emphasis) versus complicated analyses and predictions. Also, for DDRP, many data had to be carefully evaluated (e.g., geology and soils) because the format or units used were not consistent across states. For various reasons, including a mid-course expansion of the project, it took 3 years longer to get all of the data together than originally thought.
These differences highlight several issues relevant to the successful interfacing of data. The NSWS data sets required less documentation of metadata and were available relatively quickly to policymakers, agency administrators, and scientists both inside and outside of the agency. Because the NSWS data were more descriptive in nature, focused on a single medium, and required less preliminary processing, there seemed to be fewer organizational barriers to sharing and integrating data. With fewer variables in NSWS, it was easier to agree on and use standard formats and data conventions, which resulted in fewer data incompatibilities. Also, there was only a limited temporal component in NSWS, and all data were collected on the same spatial scale within a consistently used sampling design, a situation that facilitated data interfacing within various statistical analyses.
The multimedia database in DDRP was still unavailable at the time of this writing because of the greater degree of complexity in data integration. Scientists in a wider range of disciplines had different attitudes and approaches to sharing data, providing metadata, and using various types of software and hardware. Also, there was more uncertainty in the nature of the research, and so early results, especially experimental outcomes, resulted in shifting data requirements. For example, some pieces of field equipment did not work as well as expected or needed.
Perhaps the most significant difference between DDRP and NSWS was the requirement in DDRP of relatively complex models for data integration. Two such models were the Electric Power Research Institute's Integrated Lake-Watershed Acidification Study (ILWAS) model (Gherini et al., 1985) and the Model of Acidification of Groundwater in Catchments (MAGIC) from the University of Virginia (Cosby et al., 1984). Both models are driven by precipitation rates and rainfall chemistry (inputs) and are capable of predicting future (50 to 200 years) rates of surface water acidification (output) in terms of changes in pH and ANC. ILWAS, considered
to be more realistic by some researchers, is complex and requires concurrent hydrologic and chemical process data on similar scales, especially in regard to soils that may be highly heterogeneous. In contrast, MAGIC is based on more simplified assumptions about hydrology and data for averaged soil parameters; thus, data integration is an easier task, but this benefit comes at the expense of less realism about the ecosystem.
Another degree of complexity in data interfacing develops when atmospheric fate and transport models are coupled with watershed models such as these. Clearly, coupled models are needed to produce the final product (future scenarios of acidification), and as such they determine the requirements of these models in terms of data preparation, quality control, data compatibility, and data management.
With regard to the biological assessment portion of the Aquatic Processes and Effects part of NAPAP, the committee found it difficult to understand why many aspects of that research, including data management in particular, deteriorated over time. Although NAPAP managers improved their efforts based on their experiences with the program, they were unsuccessful in obtaining sufficient funds to implement fully the biological component as originally proposed. Thus, the biological assessments portion, especially near the end of the program, seemed to have experienced the most problems: the database was not maintained nor accessible; no consideration was given to distributing or archiving the data; and there was poor planning (or no planning) early on concerning data needs, compatibility, and integration. It is important to note that policymakers were convinced from the total integrated results of the studies, including those on biological indicators, surface water chemistry, and paleolimnological studies (not described here), that adverse impacts on water bodies are occurring due to anthropogenic causes and that further research is indicated.
Although important difficulties with the aquatic parts of NAPAP have been identified, most of the individual components seemed to complement the others, and the database management system appeared to facilitate necessary interactions, by easing the exchange and application of data collected in one part of the program to another. Specifically, the tiered approach was thought to be successful: NSWS was an extensive survey of a more focused nature (fewer parameters and questions addressed), DDRP was regional and predictive in nature and integrated multimedia data, and the ERP focused on process-related research at fewer selective sites, but did integrate physical and biotic databases. This tiered approach was a deliberate part of the design of the project. Consideration was given to different spatial and temporal scales, and there was a balance between monitoring, assessment, and research to address process questions within a freshwater ecosystem perspective. This foresight
appeared to be based on effective planning done by strong project management.
Finally, the committee was informed by the executive director of NAPAP, Derek Winstanley, that the future of the databases generated for the Aquatic Processes and Effects portion of NAPAP (and perhaps all of NAPAP) is uncertain. These data will be crucial to developing other programs, such as the Environmental Monitoring and Assessment Program, but no consideration was given to long-term maintenance of the data in the original goals, despite the long-term support of NAPAP. This situation should not be allowed to occur in future environmental monitoring programs.
The lessons learned are organized so as to emphasize that data integration in interdisciplinary studies cannot be viewed as a separate and distinct entity. On the contrary, data integration is inextricably linked to program planning and objectives, all aspects of sampling and analysis, and the various methods and procedures employed in the analysis and interpretation of the resulting data. Accordingly, even though the principal topic of this study is data integration, the committee found it necessary to expand its review somewhat, in order to place data integration in the proper context.
This section summarizes the major lessons learned by the committee in its review of this interdisciplinary, multimedia, and exceedingly complex research program. It should be emphasized that NAPAP was a policy-oriented activity, and the potential difficulty in transferring the lessons from this to other types of studies is recognized. The major deficiencies in data management and integration were related principally to inadequacies in the program planning component.
An organization and management structure setting forth roles, responsibilities, and authorities of all cooperating agencies should be established at the outset. This should include the designation of a program director with appropriate authority over all participants.
A detailed experimental protocol should be developed and approved for the total project. This protocol should identify research needs, priorities, milestones, and a phased or tiered approach to completing the entire program. All foreseeable interdisciplinary research requirements and supporting data management provisions should be addressed at this time.
Multiyear commitments for required levels of resource support should be made and vigorously supported by each participating agency.
Methods, techniques, and procedures should be established in the planning process for moving the program forward successfully. Requirements for databases and data management to include acceptable data and metadata characteristics, formats, data ownership, and accessibility to users should be established. Such planning requires early consideration of how the data produced by each cooperating agency will be meshed to provide an integrated assessment of causes and effects of acid precipitation.
Provisions should be made for a periodic planning review process (e.g., every 6 months) to assess progress of the study and make any indicated mid-course corrections in the total experimental plan as well as in the technical work plans of the participating agencies.
During program implementation, there should be timely exchange among all cooperating agencies of all research results and technical reports and presentations. This should involve frequent scientist-to-scientist technical exchanges, with special attention given to interfaces or boundaries among the research projects being carried out by the various cooperating agencies.
There should be periodic (e.g., yearly) internal and external quality assurance audits of all aspects of the total program. In addition, periodic external peer review of the total program implementation should be conducted.
Comprehensive information meetings involving all cooperating agencies should be conducted on at least an annual basis, with appropriate emphasis given to the interdisciplinary research issues and related data management and integration activities.
Program Completion, Evaluation, and Future Activities
Continuous feedback should be obtained from users of the resulting databases with regard to their accessibility, utility, and any problems encountered. Appropriate changes should be made as required.
Databases should be maintained in readily available form and updated as necessary. Resources required to accomplish this maintenance should be identified and obtained. Based on this case study, allocating at least 10 percent of the total program budget for data management would not be unreasonable.
Depending on research objectives, the level of effort and the scale of the required biological data should be carefully matched with those of the required geophysical and geochemical data during the planning phase of the total program.
Data managers and scientists should be located at the same site, if at all possible, to facilitate effective interaction and cooperative team efforts.
Provision should be made for periodic (e.g., once every 5 years, or more often) information meetings or symposia to review the scientific state of the art. Following each information meeting, a proceedings volume should be prepared, including a description of current interdisciplinary research needs and related data management priorities to support those needs.
Cosby, B.J., R.F. Wright, G.M. Hornberger, and J.N. Galloway. 1984. Model of Acidification of Groundwater in Catchments. Draft Users Manual. EPA/NCSU Acid Precipitation Program, North Carolina State University, Raleigh, N.C. 246 pp.
Gherini, S.A., L. Mok, R.J. Hudson, G.F. Davis, C.W. Chen, and R.A. Goldstein. 1985. The ILWAS model: Formulation and application. Water, Air, Soil Pollut. 26: 425–459.
Interagency Task Force on Acid Precipitation. 1982. National Acid Precipitation Assessment Plan. Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1990a. Current Status of Surface Water Acid-Base Chemistry. Acidic Deposition: State of Science and Technology, Rep. 9. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1990b. Watershed and Lake Processes Affecting Surface Water Acid-Base Chemistry. Acidic Deposition: State of Science and Technology, Rep. 10. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1990c. Historical Changes in Surface Water Acid-Base Chemistry in Response to Acidic Deposition. Acidic Deposition: State of Science and Technology, Rep. 11. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1990d. Episodic Acidification of Surface Waters Due to Acidic Deposition. Acidic Deposition: State of Science and Technology, Rep. 12. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1990e. Biological Effects of Changes in Surface Water Acid-Base Chemistry. Acidic Deposition: State of Science and Technology, Rep. 13. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1990f. Methods for Projecting Future Changes in Surface Water Acid-Base Chemistry. Acidic Deposition: State of Science and Technology, Rep. 14. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1990g. Liming Acidic Surface Waters. Acidic Deposition: State of Science and Technology, Rep. 15. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1991a. Mission, Goals, and Program Plan Post 1990. Public Review Draft. NAPAP Office of the Director, Washington, D.C.
National Acid Precipitation Assessment Program (NAPAP). 1991b. The U.S. National Acid Precipitation Assessment Program 1990 Integrated Assessment Report. NAPAP Office of the Director, Washington, D.C.
Oversight Review Board of the National Acid Precipitation Assessment Program. 1991. The Experience and Legacy of NAPAP. NAPAP Office of the Director, Washington, D.C.
Rubin, E.S. 1991. Benefit-cost implications of acid rain controls: An evaluation of the NAPAP integrated assessment. J. Air Waste Manage. Assoc. 41(7): 914–921.
Rubin, E.S., L.B. Lave, and M.G. Morgan. 1992. Keeping climate research relevant. Issues Sci. Technol. VIII(2): 47–55.