Front page Table of Contents Abstract Chapters 1 2 3 4 5 6 7 8 9 Bibliography
For over a decade, the U.S. Department of Energy’s Bonneville Power Administration has worked with state, tribal, and federal agencies to harmonize fisheries, wildlife, and hydroelectric facilities in the Pacific Northwest region. This region is linked geographically and economically by the Columbia River basin, which covers an area roughly the size of France (Figure 6-1).
To support these activities, these agencies built two collections of digital data on stream-related fisheries and wildlife, the Northwest Environmental Database (NED) and the Coordinated Information System (CIS). NED was a collection of geographic datasets on fisheries, wildlife, cultural resources, dams, and other features throughout the Pacific Northwest. CIS was a tabular database focused on anadromous fisheries (i.e. ocean-migrating species such as salmon and steelhead) within the Columbia River basin.Figure 6-1. Pacific Northwest States and Columbia River Basin
Data for both systems came from many different sources and were shared via state and tribal coordinators using a common geographic reference, the River Reach File identifiers devised by the U.S. Environmental Protection Agency (EPA) and the U.S. Geological Survey. Both systems were funded by the Bonneville Power Administration and guided by the Northwest Power Planning Council, a regional agency representing the four Northwest states. Yet despite their similarities, NED and CIS were conceived and built separately: the former to plan hydropower development and protect fish and wildlife; the latter to monitor the progress of anadromous fisheries programs. They took quite different approaches to building a regional information sharing infrastructure. NED consisted of topologically complete geographic data residing in and maintained by individual state agencies, loosely integrated by means of the River Reach identifiers and a set of cross-reference tables. CIS followed a more traditional approach, with centralized, tabular data maintained and disseminated by a regional consortium, the Pacific States Marine Fisheries Commission.
Both systems facilitated complex collaborative environmental management in the region. NED supported habitat assessment for endangered species, hydroelectric permitting, and a variety of other activities; CIS supported the management, and rehabilitation of salmon and steelhead species in the Columbia River and its tributaries, several of which have become severely threatened in recent years.
After several years of parallel existence, NED and CIS were merged in early 1996 into a regional "aquatic resource information network," dubbed StreamNet. The initial focus of this new information system is the assessment and protection of anadromous fisheries; ultimately, its founders hope to build it into a general-purpose information resource. Meanwhile, as the salmon crisis and other environmental controversies unfold in the Pacific Northwest, dozens of private, state, federal, and tribal groups continue trying to protect and restore a valuable ecosystem permanently altered by human activity. StreamNet may have a pivotal role to play in these efforts by maintaining regional consistency, facilitating consensus, and infusing controversial decisions and complex research with the best possible environmental information.
StreamNet and its predecessors offer useful lessons for the design of modern information-sharing infrastructures because of their decentralized data storage and control, their relative longevity, and the clear impacts they have had on tasks and processes of environmental management in the region. This study sketches the contexts in which these systems were built, the choices made in their design, the impacts they achieved, and the challenges they encountered, in order to draw general lessons on the sharing of geographic information.
In December 1980, the U.S. Congress passed the Northwest Power Act (U.S. Congress, 1980), calling for electric-power conservation and planning in the Pacific Northwest region. Recognizing the adverse ecological impacts of hydroelectric power, the Act also included measures to "protect, mitigate, and enhance the fish and wildlife, including related spawning grounds and habitat, of the Columbia River and its tributaries" (§4(h)(1)(A)). The Act created a regional inter-state compact, the Northwest Power Planning Council, with a mandate to draw up a long-term Electric Power Plan for the four-state region, and a Fish and Wildlife Program for the Columbia River basin; and to oversee their implementation with funds from the Bonneville Power Administration. The Northwest Power Act was intended as a compromise between utilities’ energy needs and the public’s call for energy conservation environmental protection. However, beginning in the early 1980s, the demand for power dropped, and the Northwest Power Act ended up being "the blueprint for a laboratory of energy and environmental conservation" in the region (National Research Council, 1995).
In its call for a regional Fish and Wildlife Program, the Northwest Power Act spelled out three requirements in particular:
Accordingly, Bonneville and the Power Planning Council began the Fish and Wildlife Program with a regional, systematic survey of stream-related natural resources, the Hydro Assessment Study. This was conducted in two parts, with quite different methods and outcomes: a detailed study of anadromous fisheries and Indian cultural values, and the Pacific Northwest Rivers Study, which examined everything else, as follows.§4(h)(6). The Council shall include in the program measures which (...) will§4(h)(6)(A). complement the existing and future activities of the Federal and the region’s State fish and wildlife agencies and appropriate Indian tribes;§4(h)(1)(A). The program shall, to the extent possible, be designed to deal with [the Columbia] river and its tributaries as a system.§4(h)(6)(B). be based on, and supported by, the best available scientific knowledge; (...)
b. Pacific Northwest Rivers Study and Northwest Environmental Database
For the Pacific Northwest Rivers Study, Bonneville contracted with the states of Idaho, Montana, Oregon, and Washington, together with the region’s Native American tribes and Federal land management agencies, to co-write assessment guidelines; these agencies then spent over two years (1984-86) collecting a broad range of data on the region’s rivers and streams: fisheries, wildlife, botanical and geological features, recreation, historic and archeological sites, and legal constraints. The data were qualitative in nature, describing a stream’s significance (high / medium / low) for each of the resource categories, as defined by panels of local specialists in each area. In keeping with the Electric Power Plan, the study’s scope went beyond the Columbia River basin, to encompass the entire four-state region.
Midway through the Rivers Study, Bonneville convened a "Geographic Information System Task Force" to design and plan for an information system which would integrate the data from the four state Rivers Studies, including anadromous fisheries, for a variety of purposes. This data system consisted of digital geographic "layers" of stream-quality rankings tied to spatial features through the Environmental Protection Agency’s nascent River Reach Files (BPA, 1985) at a scale of 1:250,000. Bonneville and the Council later added hydroelectric site information, and dubbed the resulting collection of data layers the Northwest Environmental Database, or NED.
NED was envisioned as an evolving set of databases maintained in the four states, with summary information transmitted biannually back to Bonneville. It was also seen as geographic, rather than tabular, in nature: "Digital spatial analysis is a desire of both [Bonneville] and the Council," reported the GIS Task Force in 1985. For the most part, this vision came true: each of the four states distributed its portion of NED as a microcomputer-based "Rivers Information System," and all four states integrated these data to varying degrees into their own geographic data collections. Bonneville would extract summaries from the state databases, and incorporate them into a region-wide set of GIS data layers, while maintaining its own database of hydroelectric sites obtained from the US Army Corps of Engineers.
Thanks to the Northwest Rivers Study and the parallel anadromous fish study (detailed below), the Council was able to amend its Fish and Wildlife Program in 1988, declaring 44,000 miles of streams "Protected Areas," off-limits to all hydroelectric development. An excerpt from the Amendment shows how much it relied on the Rivers Study information:
The Council’s Protected Areas amendment, combined with an unexpectedly low energy demand through most of the 1980s, brought hydroelectric development almost to a standstill in the region. As a result, after 1988, NED saw little use for system-wide facility siting or regional power planning. Instead, it served a variety of other purposes, such as assessing species health and habitat quality, or preparing maps and guides. Meanwhile, as NED’s state partners became increasingly focused on site-specific environmental assessment tasks, they found NED’s qualitative measures and its coarse-grained geographic scale increasingly ill-suited to their needs. Thus, faced with competing priorities, many of these agencies allowed their portion of NED to fall out of date, and out of consistency with other NED components in the region.Beginning in 1983, the Council directed extensive studies of existing habitat and has analyzed alternative means of protection. (...) The Council, relying on these studies, designated certain river reaches in the Basin as "protected areas," where the Council believes hydroelectric development would have unacceptable risk of loss to fish and wildlife species of concern, their productive capacity, or their habitat. (NWPPC, 1988)
c. Anadromous fisheries and the Coordinated Information System.
As the Northwest Rivers Study was underway, in a separate part of the Hydro Assessment Study, the Northwest Power Planning Council was collecting detailed, quantitative data on anadromous fisheries. For several years, beginning in 1983, the Columbia River Inter-Tribal Fish Commission (CRITFC), along with Bonneville, the Council, and other state, federal, and tribal organizations, debated how to coordinate, maintain, and disseminate the resulting anadromous fisheries data. This was a slow process: Bonneville first designated funds for a Coordinated Information System in 1988; actual databases began to be constructed in 1992; and the Pacific States Marine Fisheries Commission assumed responsibility for the project in 1993. CIS grew to be a large, complex set of data compiled from many published and unpublished sources. State, federal, and tribal fisheries managers would send information to the Pacific States Marine Fisheries Commission, which distributed it periodically to users via diskettes, reports, and an electronic bulletin board.
CIS was funded by Bonneville and built on the same unit of analysis as NED (the stream reach)—closely related to and in many ways an offshoot of NED. "I view CIS as conceived out of "NED envy,"’ recalls one NED participant, "I have always seen them as two parts of the same puzzle. [Bonneville was] hornswoggled from day one to pay twice for the same work!" Yet CIS’s proponents emphasized its detailed, quantitative, purely tabular information, specifically focused on anadromous fisheries (distributions, habitat, life stages, and genetics): "it’s quantitative data, which means you can work with it." Moreover, as one member of the CIS steering committee pointed out, "We don’t do analysis here": CIS emphasized raw numbers as more "neutral" data than NED’s qualitative judgments, more acceptable to all parties involved. One final difference: in keeping with the Fish and Wildlife portions of the Northwest Power Act, CIS data covered only the Idaho, Oregon, and Washington portions of the Columbia River basin (Montana has no anadromous fish).
Thus, no-one saw the two studies, or the data they produced, as co-equal. Many considered anadromous fish to be the true focus of the Hydro Assessment Study, with geographic references and additional data categories added in on a secondary basis. Others felt that the anadromous fish study should have been included in the Rivers Study’s comprehensive, distributed, geographical approach. The two views are summarized by the following exchange, overheard at a Council meeting: "CIS is the tree, and NED is a branch on that tree. –No, NED is the tree, and CIS is the branch." This difference of views persisted even as Bonneville was merging the two programs (into StreamNet) in 1996: maintainers of CIS simply renamed their stand-alone database "StreamNet (formerly CIS)," and put a hypertext link to it on the StreamNet website, alongside interactive map catalogs and a wide range of stream-related links.
3. Context of the infrastructure
The context of these two data sharing efforts helps to explain the choices made in their design and implementation; changes in their context help to explain why some approaches may have been difficult to sustain.
i. Many conflicting stakeholdersBoth NED and CIS arose in a context of antagonism between developers, protectors, and competing users of natural resources. As Dietrich (1995), Lee (1993), and others have detailed, the institutions concerned with the Columbia River Basin form a complex, overlapping web of regional, state, federal, tribal, and private groups, all with different, often conflicting objectives and views of the river. In its recent "Salmon Governance" report to Congress, the Council referred to "a constellation of agencies, courts and other entities" that have shaped the development and management of the Columbia River (Northwest Power Planning Council, 1996).
Nearly all inter-agency cooperation in the Columbia River region has been mandated by court orders or federal legislation (e.g. the Pacific Salmon Treaty, the Salmon and Steelhead Conservation Act), and often facilitated by improved data and analyses that helped competing users to agree on basic assumptions. Yet despite federal legislation and the promise of improved decision-making information, cooperative projects such as NED and CIS often had to adopt complex arrangements to work around sharp disagreements between agencies. For instance, as one CIS participant recalls, Idaho’s Fish and Game Department and the Shoshone-Bannock Tribes would not contract with the Columbia River Inter-Tribal Fisheries Commission due to ongoing court battles; so the contract for their CIS work was made through a third party, the Columbia River Fish and Wildlife Authority.
ii. Other regional data sharing efforts: reinforcement and competitionWhen NED and CIS first got underway in the mid-1980s, many inter-agency efforts had been formed over the years ( "the Columbia River Basin is the capital of inter-agency fishery committees," one long-time observer told me); but paradoxically, information sharing between agencies had remained rare. The Northwest Power Act and its resulting Hydro Assessment Study were seen as a pioneering experiment in inter-state and state-federal cooperation: one Council member considers that one of the greatest accomplishments of the Northwest Rivers Study was simply persuading states and tribes to release their data outside their jurisdiction. In subsequent years, the region saw several other efforts in regional cooperation and data sharing. These encouraged agencies to coordinate their information and activities and expanded the channels of communication among them—while also complicating the choices among competing standards and affiliations.
From the mid-1980s to the early 1990s, Bonneville took part in the Northwest Land Information Systems Network (NWLISN), an interagency committee that built a geographic data clearinghouse for Washington and Oregon, the Spatial Data indeX (SDX), maintained by a research group at Portland State University (Dueker and Vrana, 1995). Despite an enthusiastic start, the Network’s informal structure proved hard to sustain based only on a shared need, in the absence of dedicated funds or signatory commitments. In the words of one NWLISN participant, "There was no real meat involved: we just passed the hat whenever data needed to be collected." Nonetheless, in 1986 the Network led the US Geological Survey to begin upgrading the River Reach files to the 1:100 000 scale in the four-state region (Fisher, 1993), a move which finally came to fruition, largely through NED, in the mid-1990s as the four states finalized the Reach Files and upgraded their rivers data to this level of detail.
Also at the federal level, President Clinton’s 1992 Northwest Forest Plan (EOP, 1993) responded to a long-standing controversy over old-growth forests by calling for regional ecosystem management in the US Forest Service and the Bureau of Land Management. As detailed by Yaffee (1994), this forced the Forest Service in particular to undergo deep changes in its goals and modus operandi. In response, these two agencies undertook major efforts in data collection and standardization efforts, and in sharing information among their own units and with outside agencies (including several NED and CIS participants). The first of these programs, known as the Inter-organization Resource Information Coordinating Council (IRICC), covered western portions of Washington, Oregon, and northern California, with a strong initial focus on the range of the northern spotted owl. A second program, known as the Eastside Assessment, studied forests in eastern Washington, Oregon, and Idaho. A third program, known as PACFISH, focused on Pacific salmon and steelhead habitat from California to Alaska. These programs brought together many resource data managers in the Pacific Northwest region; they funded some NED-related data development (the Eastside Assessment helped fund a 1994 update of the Idaho Rivers Information System); they led to detailed data standards (such as PACFISH stream habitat criteria), and contributed detailed information (the present-day StreamNet credits IRICC for its aquatic habitat data).
iii. Increasingly mature & formal state data sharingOver the course of NED’s and CIS’ existence, each of the region’s four states independently defined increasingly detailed standards and procedures for data exchange. Each defined a "lead" agency for GIS, a Geographic Information Council, and written guidelines for inter-agency cooperation and data exchange. Under funding from IRICC, Oregon and Washington made formal agreements (known, respectively, as "Data ’96" and "Baseline ’97") with state and federal agencies for long-term cooperative base data development. Thus, regional data sharing efforts such as NED or CIS had to take into account increasingly complex underlying institutions and relationships in their contracting and funding.
iv. Informal communitiesIn addition to formal guidelines and institutional agreements, increasing use of computers and geographic information systems led each of the states to set up technical working groups to facilitate communication and coordination. One regional forum among state GIS specialists came about unexpectedly: when the U.S. Geological Survey released the new 1:100,000 River Reach files in 1990, state agencies found them riddled with errors; a "user’s revolt" ensued (in the words of one state GIS manager), and Bonneville contracted with the states to finish the job, which took until mid-1995. In so doing, the states traded information and learned from each other in the area of data management and GIS. These informal technical relationships were further strengthened by mechanisms such as the annual Northwest Arc/Info Conference. As a result, state GIS specialists in Idaho, for instance, know far more about their counterparts in Montana or Oregon than those in neighboring Nevada, Utah, or Wyoming.
i. Widely varying degrees of experienceIn the early 1980s, as the statewide Rivers Studies got underway, geographic information systems (GIS) were beginning to shape the work of a few agencies in the region, including Bonneville and the US Geological Survey. In assembling information from the Rivers Studies, however, Bonneville’s approach had to be very flexible, not only because of conflicts and mistrust between organizations, as noted above, but also because expertise varied widely across the region’s federal, state, and tribal organizations.
Many agencies had little or no experience with digital mapping or data management: in 1985, for instance, Bonneville’s GIS Task Force doubted that there was enough GIS capacity within the states or federal agencies to attempt cooperative digitizing of paper maps (BPA, 1985). End-user computing was also quite new to some agencies: the Northwest Rivers Study supplied Idaho’s Department of Fish and Game with its first personal computer. Thus, NED and CIS couldn’t count on a lot of technical input from Bonneville’s partners; part of their work consisted of equipping state and tribal agencies with hardware, software, and training.
On the other hand, a few state agencies had years of experience with GIS and databases. Washington’s Department of Natural Resources and Department of Wildlife were customers #3 and #19 of the Environmental Systems Research Institute (ESRI), maker of Arc/Info GIS software (now the leading "industrial-strength" GIS system). Montana’s Department of Fish, Wildlife, and Parks began rating its stream resources systematically in the 1950s, and had maintained an extensive Interagency Stream Fisheries Database, initially on mainframe punch-cards, since 1979. Not surprisingly, these agencies were sometimes reluctant to conform to NED’s specifications, which they saw as "a step down" from their own sophisticated data systems: "what they don’t get from us in consistency, they get in information," said one NED participant. Even as GIS use increased throughout the region, these wide differences in expertise persisted and continued to complicate any coordination.
ii. Increasing prominence of GIS standards and networkingBy the mid-1990s, all four states had built up their geographic data collections significantly, with large datasets at scales of 1:100,000 or 1:24,000; and each state’s geographic information council had begun to set standards for information exchange within the state. Given their increasing sophistication, states became less likely to worry about conforming with NED in their data management decisions, given its coarse scale (1:250,000) and the "archaic" form of each state’s Rivers Information System (generally a stand-alone database with a DOS menu interface). In most cases, the states incorporated NED’s rivers data into their state GIS collections, modifying them with little concern for regional consistency. Their data management choices mostly followed their own needs in-state, with some adjustments to suit regional projects funded by the Forest Service and Bureau of Land Management: for instance, in 1993 IRICC’s GIS team recommended a standard map scale of 1:24,000.
Furthermore, although the design and implementation of NED and CIS did not depend on a physical data network, several of their participating agencies began to explore the use of an increasingly ubiquitous Internet in sharing natural resource information. By the mid-1990s, some had come to rely on wide-area networks for their basic functions: Washington’s Department of Natural Resources used the Internet to share data volumes with the Department of Fish and Wildlife across town, and to link its own 14 district offices across the state to large shared computing resources. Montana’s State Library became one of the first nodes on the National Geospatial Data Clearinghouse, and led the move to build an Internet "backbone" linking state offices in Montana’s capital.
Thus, regional data initiatives such as NED or CIS had to take into account increasingly complex data systems, rather than just building them from the ground up as they could in the early 1980s. In 1995, NED and CIS managers began to experiment with Internet data distribution, which led them to design StreamNet as a fully Internet-based information service, with distributed databases and maps based on the World Wide Web.
In response to their context, NED and CIS made distinct choices in building an information sharing infrastructure, including both institutional arrangements (responsibilities and decision-making powers) and technical design choices (degrees of decentralization, standards, and geographic scale).
Both NED and CIS were funded by Bonneville and guided by the Northwest Power Planning Council, but they adopted quite different organizational structures.
i. Northwest Environmental Database: stakeholder cooperation and interdependenceNED’s institutional relationships were designed to involve a wide range of interested parties while keeping a fairly simple structure. In cooperation with the Northwest Power Planning Council, Bonneville funded state agencies in Montana, Idaho, Washington, and Oregon to collect and then maintain statewide rivers data. These counterparts had some discretion as to the data format, pace, and even scope of their portion of the study, and coordinated state, federal, and tribal sources of resource information within each state. NED brought together, in some cases for the first time, information and experts from quite different areas, with quite fruitful results: as one participant recalls, Montana’s Department of Fish, Wildlife, and Parks provided initial data-management concepts in 1985; Oregon’s Department of Energy then wrote up a data management plan and built a DB2 interface; the Washington State Energy Office proposed tying data to stream reaches; Idaho’s Department of Fish and Game added color to Oregon’s data interface; and thus NED gradually took shape. Several years later, Idaho’s and Oregon’s Water Resources Departments led the move to complete the 1:100,000-scale River Reach files. Data maintenance was entrusted to the state energy departments and fish and wildlife agencies, with regional coordination through quarterly reports and biannual meetings. Thus, to enlist the support of the major stakeholders, NED used interdependence and cooperation between federal and state government agencies, and between energy production and environmental protection.
ii. Coordinated Information System: neutrality through isolation from stakeholdersCIS’ organizational relationships were a bit more complex. The Columbia River Inter-Tribal Fish Commission (CRITFC) developed the concept in 1984 and built institutional partnerships over several years. CIS began as a forum for exchanging salmon data between policy-level fisheries specialists; but it gradually turned into a database management project, and in 1993 CRITFC transferred its technical and administrative responsibility to the Pacific States Marine Fisheries Commission (PSMFC), which had several years of experience with managing large fisheries databases (PSMFC, 1996). CIS built institutional support through bimonthly meetings of its Steering Committee, which represented Bonneville, the Council, CRITFC, state and tribal fish and wildlife agencies, and the National Marine Fisheries Service, which played an important role after 1992 under the Endangered Species Act. Thus, to encourage support from all sides of the salmon controversy, CIS positioned itself as a repository of "uninterpreted," "neutral" data on anadromous fisheries, housed within a well-respected inter-state Commission with no regulatory or management authority.
The new StreamNet embodies both of these approaches, contradictory though they may be (one NED coordinator described it as "one of the continuing thorns in my side"). To emphasize StreamNet’s independence from the region’s stakeholders, a new Internet domain was created for it, streamnet.org, whose Web-based information services are developed largely on Bonneville computers but reside on PSMFC computers.
Here again, NED and CIS represent quite different choices. NED’s focus was broad, qualitative data, carefully tied to geographic data features; whereas CIS emphasized quantitative detail and bibliographic references, with only implicit links to geographic features.
i. Northwest Environmental Database: decentralized, qualitative, geographicInitially, NED was not intended as a distributed database; but because the Council’s efforts were to complement the region’s existing fish and wildlife activities, NED was designed to strengthen, expand, and integrate, but not replace, existing data systems. This was accomplished through a distributed database model. This was an ambitious choice for 1985: Bonneville and the Council let each state determine most data management specifics, and in order to integrate and compare between states, they spent about a year building cross-reference tables between each state’s river coding systems and the EPA Reach File identifiers. NED also emphasized spatial analysis and a comprehensive view of the ecosystem through use of GIS. To obtain a regional overview quickly, Bonneville and the Council opted for a geographic scale of 1:250,000, and chose to record only qualitative judgments (such as good / fair / poor habitat) from panels of biologists, rather than quantitative data (such as population counts for particular species).
ii. Coordinated Information System: centralized, quantitative, tabularCIS took a more centralized approach than NED: the Pacific States Marine Fisheries Commission (PSMFC) managed a tightly integrated set of data tables, for which it collected data mostly by traditional methods such as reading printed reports (though some contributors did begin to shift to electronic data submissions). River Reach numbers, at a scale of 1:250,000, served as identifiers, but their geographic role in CIS’ tabular data was only implicit. Although essentially a centralized system, CIS was seen as "distributed" in that it could be packaged and sent out on diskettes for local use. In mid-1995, PSMFC staff began to explore options for networked data distribution, both through a dial-up bulletin-board system and the World Wide Web—"I don’t like to put all my eggs in one basket," said one data manager at the time. Another difference was CIS’ emphasis on quantitative detail: this design choice was partly a response to states’ disappointment with NED’s broad, qualitative information.
As a successor to both of these systems, StreamNet supplies both tabular and graphic information through Web-based query tools; its long-term goal is to link together decentralized information maintained by agencies throughout the region, but for now all of its information resides on a single server at www.streamnet.org.
5. Information sharing characteristics
NED and CIS can be characterized by the kinds of information sharing they supported, the size of each information infrastructure, and the quality of information sharing achieved.
a. Forms of information sharing
Information sharing in NED and CIS took a few, mostly simple forms. Many NED users obtained hard-copy tables and maps, or digital database or GIS extracts, from the state coordinators, Bonneville, or the Council. Many CIS users obtained digital or hard-copy database extracts from the Pacific States Marine Fisheries Commission or members of the CIS steering committee. NED and CIS were also distributed on diskettes as standalone microcomputer database applications: NED consisted of four distinct state Rivers Information Systems, a hydroelectric site database, and the Council’s Protected Areas database; CIS came as a single, large information package, the "CIS Distributed System."
Many users of NED and CIS were within state and tribal fish and wildlife agencies; others were in state energy offices and water resource departments, county planning offices, and federal agencies such as the Forest Service, the Bureau of Land Management, the National Marine Fisheries Service, and the Fish and Wildlife Service. Additional users included academics, private consultants, non-profit groups, and utilities.
Generally, few of these users demanded more advanced forms of sharing, such as client-server architectures or interoperability with existing software. However, anticipating a more highly networked future, StreamNet represents a major effort at providing advanced Internet access to the region’s resource information.
Another measure of these data sharing efforts
was their size and traffic volume. Both data systems were large enough
to warrant a detailed data-management policy and standards, yet small enough
for fairly simple methods of data collection and distribution. CIS had
80,000 records, describing about 36,000 reaches. Bonneville’s summary version
of NED described about 34,000 reaches in all, representing 135,000 miles
of streams; the individual state Rivers Information Systems varied widely
by state and by scale (Table 6-2).
|
River reaches
|
Stream miles
|
|
Source | |
|
Montana
|
4,000
|
?
|
|
(Montana DFWP, 1993?) |
|
8,000
|
?
|
|
(Montana DFWP, 1994; NRIS, 1996) | |
|
Idaho
|
4,500
|
28,000
|
|
(IDFG, 1994) |
|
?
|
103,000
|
|
||
|
Washington
|
1,500
|
?
|
|
(Knudsen et al., 1992) |
|
26,000
|
31,000
|
|
||
|
Oregon
|
4,000
|
?
|
|
(Forsberg, 1992) |
|
14,000
|
45,000
|
|
One last measure: According to Bonneville’s contracting officer, the yearly budget for NED was $400-$440 thousand, not including $100-$150 thousand per year to develop the River Reach File system; Bonneville’s yearly budget for CIS averaged about $500 thousand over the 1988-1995 period. The combined StreamNet began with a 1996 budget of about $1.7 million.
c. Quality of the shared information
From the user’s perspective, NED and CIS were only as good the information they provided: (i) their precision and accuracy in relation to user needs, (ii) their update frequency relative to master data sources (concurrency) and to the user’s time requirements (timeliness); and (iii) their ease of use or encapsulation.
i. Precision, accuracy, and scaleNED’s lack of quantitative precision and especially its geographic scale were its chief criticisms in its latter years. NED’s initial choice of geographic scale (1:250,000) and its "soft" data (based on the judgments of local resource specialists) were fully adequate for its initial policy objectives, Protected Areas and regional hydropower planning. Designers of the Northwest Rivers Study and NED acknowledged the limitations of these choices from the outset—the 1985 GIS Task Force warned against misuse, and recommended a 1:100,000-scale update within 5 to 10 years (BPA, 1985)—although one NED coordinator saw the scale question quite differently:
NED’s goal was not to supply full local detail, but to provide a consistent, comprehensive view of the region’s natural resources (which had never been done before). Thus, Bonneville and its state, tribal, and federal partners spent a lot of time jointly drawing up a set of assessment guidelines. As a result, NED’s comprehensive "presence/absence" data on anadromous fisheries continued to prove valuable even a decade later to the National Marine Fisheries Service in evaluating the endangered status of salmon stocks in coastal Washington and Oregon.No research (to my knowledge) shows that [1:100:000 or even 1:250,000 scale data omit ecologically significant streams]. At times, I'm afraid, that excuse is used to inhibit action and change to benefit the environment. We will never know all that there is to know about our lands and waters—this should not, in my view, be used as an excuse for inaction. (...) Indeed, many of the region's old growth forests have been chopped down without the use of ANY data at all!
However, as the region shifted its attention away from regional energy policy in the early 1990s, many state and tribal agencies embarked on GIS-assisted projects aimed at site-level resource management, and found NED’s scale and precision inadequate. A good example of this was Washington’s Surface Water Identification System, a separate set of stream codes at a scale of 1:24,000 (Vanzwol, 1995). In anticipation of these uses, Bonneville invested a steady stream of funds, starting in 1986, into upgrading the River Reach files to the 1:100,000 scale; but conflating data across geographic scales proved harder in practice than on paper. As one member of the conflation team wrote in 1993, "the result was a project that took three times longer, and cost four times more than originally estimated" (Fisher, 1993). As a result, the 1:100,000-scale products weren’t complete until nearly 10 years later—by which time several states had embarked on building their own 1:100,000 or 1:24,000 scale data, with little concern for consistency beyond state boundaries. As a result, according to a tribal fisheries coordinator, several tribes in Western Washington refused to use even 1:100,000 data, claiming that it omitted 40% of the streams. This was a matter of some controversy: according to one federal fisheries manager, the 1:24,000 dataset was so dense as to be nearly unusable, featuring some streams no larger than roadside ditches. To further complicate the problem, part of the River Reach File identifier was based on a linear "River Mile" referencing system—these numbers became meaningless when transferred to a different geographic scale.
CIS’ focus on tabular, quantitative data reflected some of these criticisms of NED. First, CIS was based on the same River Reach identifiers as NED, and thus corresponded to the same geographic scale, 1:250,000; but its purely tabular structure allowed easier insertion of data for smaller stream reaches. Thus, it circumvented many of the difficulties of spatial accuracy and topology implied by a GIS. Tying these newer data back to a consistent geographic framework was not a high priority for CIS’ managers: they felt that GIS was beyond the reach of most of their audience, and rarely used for analytical purposes anyway. A second way in which CIS reflected NED’s criticisms was its focus on collecting and providing quantitative data, instead of value-laden analyses or assessments. However, one NED participant offered a different perspective:
The new StreamNet project adopted the 1:100,000 River Reach system as its primary geographic reference. A 1995 StreamNet proposal indicated that this referencing system would be enhanced in the future by "appropriate links to other scales." It also chose to focus its data collection efforts in the first few years, to provide greater detail on fish distributions and habitat before branching out to broad ecosystem data.There is always more quantitative data on anadromous fish because that’s always where the money has been. BUT, don’t be deceived. Much of the ‘science’ on anadromous fish is soft at best—why else the 15 year argument over how to bring them back?
ii. Timeliness and concurrencyFish and wildlife phenomena in the Columbia River follow a seasonal frequency, as the various fish runs make their way through the river system to and from the ocean. CIS was updated once or twice a year, as figures came in from state and tribal coordinators and other sources at the end of each fishing season. This was the frequency needed for CIS’ primary use: assessing the progress of the Council’s Fish and Wildlife Program. For ecological assessment, long-term trends also needed to be distinguished from short-term fluctuations, and current conditions compared with historical figures. Both CIS and NED supported this category of use, mostly at the policy level: characterizing species habitats or development suitability for the long term, with reassessments every few years. Some of NED’s data categories were updated as funds and cooperative relationships allowed, and served these purposes well.
Bonneville did continue funding data-management activities in the states and tribes; but as region-wide concerns faded from view, states and tribes turned their attention to local priorities, as noted above, and left many of NED’s data categories unchanged after the 1984-86 Rivers Studies. Thus, much of NED gradually became obsolete for most ecological assessments: in the eyes of many potential users, this was NED’s third major limitation (after its scale and qualitative nature). The lack of updates was closely related to the first two limitations: states saw little use for NED in their site-specific work, so they saw little value in updating it; and Bonneville never directly required them to do so. One of StreamNet’s initial objectives was to update several key NED categories, such as anadromous fish distribution.
iii. Usability and encapsulationThe most common mode of use for both NED and CIS was a telephone call to one of the state, tribal, or regional coordinators, who would look up the information to answer a question, or prepare a custom list or map—so measures of usability or encapsulation were not all that meaningful. NED and CIS were also distributed as stand-alone microcomputer database applications, directly usable by most recipients upon installation on a local microcomputer. More advanced uses, such as linking to local GIS systems, were not supported, though advanced users with geographic River Reach Files could fairly easily tie data records to spatial features. In late 1995, the US Geological Survey’s Oregon District office began to facilitate this use by putting 1:100,000-scale River Reach files in Arc/Info format on its Web site. Around that time, the new StreamNet system also went online, with the primary goal of providing a variety of forms of Internet access to geographic resource data, in particular through Web-based mapping and querying tools.
d. Quality of the information infrastructure
In contrast with the users of an information sharing infrastructure, its maintainers generally have quite different ways to assess its quality. Their focus would be on aspects of the infrastructure itself: (i) its capacity for two-way information sharing (reciprocity), (ii) the flexibility to grow the infrastructure without overloading data management procedures (scalability), and (iii) its non-intrusiveness, the degree to which information sharing can adapt to participants’ various procedures and standards.
i. ReciprocityNED’s and CIS’ design depended on several data providers feeding back information to a central "hub" (Bonneville or the Pacific States Marine Fisheries Commission), for use there and redistribution to users throughout the region. In practice, because of its infrequent updates, NED saw few opportunities for data transfer between Bonneville and state coordinators, and so its reciprocity was only potential. In contrast, CIS was updated about twice a year, thanks to a small team at the hub which collected and integrated data from steering committee members, mostly by manual means. In 1994, the CIS team began to streamline the data integration process by means of a simple dial-in facility and standard formats for data interchange. It also began to distribute CIS data on diskettes (the "Distributed System"), and in late 1995 devised a simple Web-based query form, a precursor to the query tool now on the StreamNet Web site. Many of CIS’ users were its own data providers; therefore some described CIS as a "crossroads" allowing fisheries specialists to share information crucial to their work. Around that time also, the USGS’ Oregon District office put its River Reach File Clearinghouse on the Web, with provisions for coordinating updates and corrections suggested by Internet users—thus further blurring the distinction between "user" and "provider."
ii. ScalabilityNED was a mostly fixed set of information and participants. But by delegating nearly all data responsibilities to the state coordinators, NED’s design lessened the data-management burden at the Bonneville "hub," thus enabling some scalable growth to occur. The bottleneck in its design was the cross-reference tables that linked River Reach File numbers to the river codes used by the four states. It took one person about a year to build these cross-reference tables—and they soon became obsolete as state identifiers changed.
CIS was a rapidly growing resource, as its data managers added data from each fishing season and related them to existing records to form time series. Yet because CIS was centrally managed, its growth required a proportional increase in the data management effort at the PSMFC "hub." By 1995, CIS had grown to about 80,000 records, and its managers were devising "filter programs" to extract data from state databases, encouraging electronic data submissions in a common exchange format, and even standardizing data collection methods. These developments, still in progress when CIS became part of StreamNet, were no doubt continued as part of that project.
iii. Non-intrusivenessIn principle at least, participants in both NED and CIS were expected to follow certain conventions in managing their data, but were not required to restructure it all. As mentioned earlier, in keeping with the Northwest Power Act, NED was a loosely-integrated regional database that complemented, but did not replace, existing state activities. In particular, states could continue using the river coding schemes to which they were accustomed; cross-referencing with River Reach File numbers made it possible to assemble a consistent regional view when needed. However, once hydroelectric development became less of a crisis issue, NED’s very hands-off policy made regional consistency difficult to maintain: update schedules, data formats, definitions, and stream identifiers diverged as each state pursued its own priorities with little or no common direction. Bonneville staff would reconcile the data manually in order to produce regional maps and summaries used mostly by federal agencies such as the US Forest Service and the National Marine Fisheries Service, who continued to find NED "a gold mine," in the words of a NED coordinator, for their assessments of spotted-owl and salmon habitats. However, the states and tribes, in many ways the backbone of NED’s design, didn’t value these course-scale, qualitative, outdated data nearly as highly; and NED gradually became simply a program whereby Bonneville funded the data management activities of fish and wildlife agencies. As one long-time observer put it in 1996, "Bonneville could take a greater role in defining data standards; but right now it just writes checks."
CIS began in an even more non-intrusive fashion than NED: it would accept anadromous fisheries data in any form, and counted on a central data management staff to reconcile it all. This policy was devised in the early days of CIS, because several key agencies (most notably the Fish Passage Center and Idaho’s Department of Fish and Game) were reluctant to share information or to change how they kept their data. As a result, when CIS was transferred to the Pacific States Marine Fisheries Commission in 1993, it consisted of an assemblage of completely disparate data. Over time, CIS staff tied these data to stream reaches and began to define standards for data definitions, identifiers, and collection methods, thus demanding some changes on the part of data suppliers.
A major challenge for the new StreamNet is to define flexible standards that will let data contributors manage their data in ways useful to them, while still enabling summaries and comparisons across the region. One change that will facilitate this process: StreamNet is specifically targeted at meeting regional data needs in support of the Council’s Fish and Wildlife Program. Compared with NED, therefore, it has less of a need to appease the four states: indeed, its steering committee features federal and tribal agencies on an equal footing with the four states.
The final measure of an infrastructure’s quality is a pragmatic one: what has it accomplished? NED and CIS provide interesting examples of impacts, from enabling consistent region-wide strategies; to enhancing public participation and inter-state collaboration, to enhancing state and regional data management.
a. Consistent, region-wide strategies
NED was built specifically for a region-wide assessment of environmental quality and hydroelectric capacity. Thanks to the Rivers Study, the Council was able to persuade the region’s public and private stakeholders to set aside 44,000 miles of streams (that is, 15% to 20% of the total) as "Protected Areas," and to redirect federal hydroelectric development to the remaining 200,000 stream miles. To some of those involved, this achievement was unthinkable before the states and tribes pooled their information consistently and systematically through NED. NED data and the Protected Areas decision also led to the first-ever estimate of future developable hydroelectric power resources, or "hydro supply curves," in the 1986 Power Plan (BPA, 1993). In speaking with those involved, it seemed that the promise of the Protected Areas amendment, and its aftermath, provided much of NED’s early momentum and support.
After 1992, the National Marine Fisheries Service (NMFS) used both CIS and NED in applying the Endangered Species Act to anadromous fish in the region. According to one NMFS analyst, this Act requires assessing stream habitat and fish species by "Evolutionarily Significant Unit" (ESU), that is, a land unit supporting a genetically distinct and significant population. Compared with studying either individual salmon runs or the Pacific salmon genus as a whole, analysis at the intermediate ESU scale required extensive data processing and regional comparison, for which CIS and NED data proved valuable in determining the boundaries and health of species. Even though it was several years out of date, NED played a prominent role in assessing critical salmon runs in coastal Oregon and Washington, outside the Columbia River basin (CIS’ geographic area).
Of course, regional summaries and joint projects are nothing new; but CIS and NED were infrastructures that allowed for more than a one-time effort and served multiple purposes. The Council’s Northwest Power Plan and Fish and Wildlife Program have five-year update cycles; Endangered Species listings are reviewed periodically; all of these studies will be easier to repeat thanks to the regional consistency afforded by StreamNet’s common reference system and its shared assessment guidelines. These regional accomplishments are good examples of "superordinate goals" or "killer applications" that often foster information sharing. NED’s experience also illustrates what can happen in the absence of such a shared goal: although it was designed as a general-purpose system, and many agreed that it could be useful for other purposes, its regional focus became diffuse soon after energy development slowed down.
b. Enhanced public access to information
According to some participants, a key benefit of these infrastructures was better public access to information. In particular, simply having region-wide information organized and accessible in one place helped state and federal groups comply with public-review requirements. Open access to information was a high priority for the Northwest Power Planning Council in particular; but given public disclosure laws, many (especially federal) agencies saw it as a requirement; and others considered it a wise precaution. As one tribal fisheries coordinator put it, "Information—if you try to keep it behind closed doors, it gets pried out in court!" Increasingly widespread Internet access in the 1990s brought greatly enhanced opportunities to release information to the public.
For the most part, pricing and access restrictions were not a major issue in releasing information to the public: most of the region’s government agencies had grown to consider data and products they built to be public goods, and freely gave them to anyone who asked. Montana and Oregon put their Rivers Information Systems verbatim on the Web; the US Geological Survey did the same with the entire set of River Reach Files. There were exceptions, to be sure: some agencies were reluctant to post data on the Web that cost a few dollars in paper form; some tribes were unwilling to release their seasonal harvest figures; and some agencies took a "fair market value" view of data distribution. Washington’s Department of Natural Resources (DNR), in particular, considered its extensive geographic data collections to be exempt from public disclosure laws, and distributed its data on a for-profit basis. This policy was widely criticized; but DNR saw it as essential to its stewardship of state trust funds, and set up elaborate written procedures for determining fair market value and allocating funds from data sales. In the early 1990s, however, Washington’s Commissioner of Public Lands called on DNR to "accept our trust mandate as broader than ‘maximizing income’ and recognize the value of ecosystems to trust beneficiaries." DNR’s GIS staff began letting more data recipients "pay" for data in a variety of ways, such as helping to enhance a regional data layer by adding local detail, or paying only token amounts—in short, anything that might benefit DNR and seal a relationship. This new data policy, emphasizing broad inter-agency partnerships, came about in part through the leadership of DNR’s GIS manager, who transposed this partnership model to the national level as chair of the National Research Council’s Mapping Science Committee (National Research Council, 1994). In short, this "fair market value" policy stood out from the "free and open access" policy of most of the region’s agencies, and generated controversy because of DNR’s prominence in the region’s GIS scene. But with the rise of "ecosystem thinking" in the region, fostered by projects such as NED and CIS, and by issues such as endangered species, even giant DNR moved closer to a policy of free and open data sharing.
c. Improved inter-state collaboration
These sharing infrastructures primarily benefited regional agencies: the inter-state Northwest Power Planning Council, the federal National Marine Fisheries Service, and others. To a lesser degree, they also enabled groups of people in the states and tribes to collaborate and to coordinate their work. At the policy level, the Northwest Rivers Study provided a mechanism for the states to wield a lot of power by working together. At the management level, CIS allowed upstream agencies in Idaho’s Snake River basin to track salmon populations on the lower Columbia on occasion as an input to local regulatory decisions. Interstate collaboration may prove critical to the long-term success of the new StreamNet, to ensure ongoing support and input from the region’s state and tribal agencies.
d. Enhanced data management: CIS and other NED offshoots
According to some NED participants, its primary impact was that it allowed people to concentrate on data management, thus enabling many other data-related projects. One data manager in Idaho stressed the learning opportunities provided by NED despite its infrequent update schedule in the 1990s:
For instance, most of the state agencies that participated in NED adopted its River Reach data structure to build datasets of their own. In fact, CIS itself, with its River Reach-based data architecture, was a spin-off of NED. The new StreamNet is yet another stage in the learning process, deriving from both NED and CIS and adding significant networked functionality.NED has indeed floundered in terms of its output. However, (...) what appears to be stagnation has really been a learning process for GIS analysts, biologists, and database programmers (like myself). (...) A gestation period was necessary, and NED has been lucky enough to have one.
Both NED and CIS faced several important challenges, and learned important lessons, in the areas of decentralized coordination, data connectivity, and organizational teamwork.
a. Decentralized vs. centralized coordination
A key challenge to any partnership is maintaining a coordinated overall focus despite differing agendas and styles among participants. NED’s decentralized coordination, mandated by the Northwest Power Act, worked well prior to the Council’s Protected Areas amendment, which was a regional concern of direct interest to each state and tribe. With the Protected Areas milestone behind them, Bonneville and the Northwest Power Planning Council never did follow up with another compelling shared objective for NED. This failing was related to funding: NED’s funds came in large part through the Council’s Power Planning budget. As the region’s focus on hydropower planning dwindled in the 1990s, so did Bonneville’s NED funding to states and tribes, and thus its ability to guide their data-management activities in a common regional direction. "It all comes down to money," says an Idaho GIS manager. "Bonneville paid people to cooperate. No money, no conductor." Yet there were other factors: for instance, without the severe delays in upgrading the River Reach File standard to larger scales, it may well have been easier to maintain greater data consistency among a dynamic, increasingly sophisticated set of participants.
CIS had a narrower scope than NED, closely related to the salmon crisis, an issue of great concern in the region. Its support and participation may have derived from its focus on an unmistakably inter-state issue, which due to the migratory patterns of anadromous fish, no one tribe or state could resolve on its own. This interdependence among states was much less strong for NED’s broader wildlife and fisheries focus. Here again, funding played an important role: CIS was funded out of the Council’s Fish and Wildlife budget. Unlike the Protected Areas and regional power planning, CIS’ raison d’être has remained prominent and will persist for a long time to come: in the words of one consultant, "Restoring endangered fish species will make the spotted-owl controversy look like child’s play."
b. Providing data connectivity among data suppliers
Whether its participants knew it or not, NED was clearly ahead of its time in the mid-1980s, seeking to reconcile distributed data sources "as is," without requiring them to conform to a single homogeneous standard. This approach has been the subject of extensive research since then, leading to modern heterogeneous multi-databases, which link component data sources via a single networked interface. NED’s cross-reference tables, linking each state’s stream identifiers with the River Reach numbers, were a simple, yet powerful form of metadata—a term which only became common parlance in database management in the mid-1990s. Thus, even without networked connectivity, NED illustrated well the potential of non-intrusive data sharing. The difficulties that it experienced - maintaining meaningful connectivity between independently changing formats, definitions, and geographic scales—remain among the key challenges of distributed database systems.
CIS was in part a response to NED’s difficulties. To facilitate comparisons, it contained quantitative data and specified how the data were measured, and established some conventions on field names and category thresholds in the data submitted from states and tribes. Furthermore, geographic scale issues were less prominent in CIS because its simple tabular structure could more easily accommodate additional river-reach records. However, with no geographic component, its use for graphic display or spatial analysis was limited to those with their own copies of the River Reach geographic database.
In merging NED and CIS into StreamNet, Bonneville, the Council, and their partner agencies sought to reap the benefits of both: NED’s decentralized structure, its links to geographic features, and its support for regional policy and planning; and CIS’ tight data integration and its capacity for scientific and management analysis.
Partnerships such as NED or CIS depend on a sense of teamwork among participants — a work relationship in which credit for accomplishments goes to the team, and not to any one agency. Yet, as one IRICC participant learned, most corporate incentives tend to favor leadership, to the detriment of teamwork: "some agencies have begun to submit solutions, rather than issues, to the team: the further we get from the President’s Forest Plan, the more the teamwork falls apart." One participant in both NED and CIS reflected that "people who mistrust others, fear change, or don’t want to share power—these are the people who undermine the collaborative process." Bonneville sought to achieve teamwork and shared ownership by giving states a lot of leeway in building NED; but as Bonneville’s "generous conductor" role diminished, the states on their own had less of a regional focus once their attention turned from NED’s ecosystem-wide concerns to specific anadromous fisheries. Also, as Bonneville’s funding diminished, face-to-face meetings among NED’s coordinators dwindled to about once a year, with quarterly written status reports. In contrast, CIS was able to maintain its "people network" through bimonthly meetings of its steering committee and frequent communications between CIS staff and other agencies (Allen et al., 1994). StreamNet has maintained this bimonthly schedule for its steering committee.
Besides coordination within NED and CIS, two key questions facing these infrastructures was whether and how to coordinate them with each other, and with related regional information sharing programs in the region (IRICC, the EastSide Assessment, and others). The new StreamNet has begun to tackle these questions by merging CIS’ and NED’s data and their approaches, and by including IRICC habitat data.
The Northwest Environmental Database (NED) and the Columbia River Coordinated Information System (CIS) provide rich illustrations of both the feasibility and the challenges of decentralized geographic and environmental information. Whereas most such systems in existence today have only begun the process of putting designs into practice, NED in particular was in place for over a decade. Their experience confirms the tangible benefits of information sharing infrastructures: declaring Protected Areas in a regionally consistent, widely acceptable manner was unthinkable prior to NED, and both NED and CIS helped to replace fruitless debates over endangered species with regionally consistent, widely accepted evaluations of species status and habitat.
NED and CIS also highlight the importance of a flexible, adaptive approach to designing, implementing, and expanding information sharing infrastructures. NED's information architecture, advanced though it was for its time, was difficult to upgrade to larger geographic scales and to increased use of advanced information systems among its state counterparts. After NED’s experience, CIS coped with its complex technological, scientific, and organizational context by centralizing data management responsibilities and by avoiding geographic information or scientific assessments. It could afford these choices because it was much more narrow in its focus, and it drew its funding and support from fish and wildlife protection budgets rather than power planning budgets. Now, with substantial new funding, StreamNet is trying to build on the experience of both CIS and NED.
NED's loss of regional consistency since the Protected Areas amendment illustrates the importance and difficulty of maintaining a compelling shared goal and an effective shared standard: by several accounts, this amendment marked the end of NED's enthusiastic participation and the beginning of its divergence into state-specific systems, accelerated by the lengthy delay in creating a new River Reach standard. CIS’ institutional support, despite its technological simplicity and high manpower requirements, suggests that a compelling, clearly shared goal can motivate quite large investments of money and time in shared information. (In fact, the comparison is a difficult one given that CIS as a data system was only around for about three years before being merged into StreamNet.) The contrast between CIS’ and NED’s pattern of growth also suggests that building a community of collaborators, as CIS did through its Steering Committee, may be important to institutional support.
On a technological note, NED confirms the feasibility of "minimal standards" in geographic information, by achieving high-quality, non-intrusive information sharing using cross-references to the EPA's River Reach Files. In its "gestation period," NED also had an unseen impact as a "seed" for new infrastructures, as illustrated by CIS and other data systems which learned from the NED experience.
In summary, both NED and CIS made impressive strides in providing regionally consistent information on the natural resources of the Columbia River Basin, and in helping to defuse a very antagonistic policy context. Their data management choices were advanced for their time, and the difficulties they encountered are still those faced by much more modern and sophisticated systems today. Moreover, their ten-year history provides a glimpse into the future of today's advanced, networked systems, and offers lessons that others ought to follow closely.
Front page Table of Contents Abstract Chapters 1 2 3 4 5 6 7 8 9 Bibliography