Front page Table of Contents Abstract Chapters 1 2 3 4 5 6 7 8 9 Bibliography
The Gulf of Maine geography is a semi-enclosed sea in the northwest Atlantic, whose rich coastal and marine ecology sharply defines the culture and economy of three U.S. states and two Canadian provinces. For several years, these five governments have worked on joint management of the Gulf of Maine’s natural resources through a bilateral Council on the Marine Environment. One of the Council’s stated goals is communication and information inter-change: at its outset in 1990, it created a Data and Information Management Committee (DIMC) to oversee communications and information interchange.
One of this Committee’s main accomplishments was a clearinghouse for data held by the region’s various public agencies and research institutions: the Gulf of Maine Environmental Data and Information Management System (EDIMS). The EDIMS system was intended as a distributed information resource, which would draw on autonomous data sites throughout the region via a common network. This was an ambitious design for the time — especially since a basic networking infra-structure to support it did not become widespread until the recent growth of the Internet’s World Wide Web. Although EDIMS has thus far had only a minor impact on the Council’s activities, it has served as a rallying point for the Council’s data-related activities, and has begun to see some use in supporting communication and collaboration across the region.
Several barriers have kept EDIMS from playing a more significant role: the Council’s weak institutional position and insecure funding, the difficulties of coordinating participants with vastly different levels of technical experience, the absence of an agreed-upon networking infrastructure, the lack of a clear shared objective, and little or no sense of inter-dependency among its participants. This complex brew of factors seems thus far to have kept EDIMS from growing beyond an initial centralized proof-of-concept into a distributed and widely-used tool for regional planning and policy-making. In the second half of its ten-year Gulf of Maine Action Plan, the Council on the Marine Environment is beginning to set forth a few clearly shared, information-intensive goals. Time will tell whether this shift can provide the momentum, resources, and direction EDIMS needs to fulfill its promise as an advanced, multi-organizational information infrastructure.
In late 1989, the states of Massachusetts, New Hampshire, and Maine, and the provinces of New Brunswick and Nova Scotia, formed the Gulf of Maine Council on the Marine Environment, with a mandate to maintain and enhance the quality of the Gulf of Maine’s coastal and marine resources. One of the Council’s focus areas was the interchange of information among its participants: in its ten-year Action Plan (1991-2000), it called for "methods to ensure that Gulf environmental databases are compatible," and for the development of a "common regional protocol allowing for the transfer (…) and periodic updating of data and information." To guide these activities, it assembled a Data and Information Management Committee (DIMC) to oversee an Environmental Data and Information Management System (EDIMS) for the Gulf of Maine. This Committee was initially chaired by a member of the Massachusetts Department of Coastal Zone Management, who secured a grant from the U.S. National Oceanographic and Atmospheric Administration (NOAA) for the Committee to compile a directory of coastal and marine data for the region through a survey of state, province, and federal agencies. Another Committee member, representing the U.S. Environmental Protection Agency, packaged this directory as a Clipper database application; it was then placed on a Unix server at the University of New Hampshire’s Ocean Process Analysis Lab, under the supervision of the Lab’s Director (who took over as co-chair of the Committee when the founding chairman had to resign by request of his home agency in Massachusetts). At the University of New Hampshire, EDIMS was coordinated by a Research Associate in the Lab, who converted the directory to an Oracle database and provided a simple forms-based interface with remote "telnet" and "ftp" access. In late 1993, however, outside project funding ran out, bringing data development to a halt (and slowing the Council’s other activities) until late 1995, when a new grant from the U.S. Environmental Protection Agency revived the Council’s activities and its interest in shared information systems. In the nearly two year hiatus, the EDIMS Coordinator and a handful of students at the University of New Hampshire built a simple World Wide Web interface to EDIMS, which gradually attained greater visibility as state and provincial agencies gained Internet access. However, after being "dormant" for such a long and crucial time, EDIMS (and the Council) seem to have had difficulty picking up where they left off: many agencies had made decisions on computing and networking without EDIMS’ input, and their concern for regional Gulf of Maine issues had waned. As a result, in mid-1996, participation in the 15-member EDIMS Committee was low (attendance overlap between consecutive meetings was small; people didn’t know each other’s names), coordination between American and Canadian participants was limited, and funding was sufficient for little more than maintenance. In early to mid-1996, EDIMS seemed to be struggling to define its role in a decentralized, complex technical and institutional environment.
The following sections present a bit more detail on the context EDIMS has fit into, the choices that have shaped it, and the measures of size and quality it has attained.
3. Context of the infrastructure
To understand EDIMS, it’s important to know its institutional and technical surroundings, both past and present. Indeed, the Committee that oversees EDIMS interacts with other parts of the Gulf of Maine Council, and with the region’s states, provinces, and oceanographic research centers. Furthermore, EDIMS is one of several users and providers of ocean-related information, with which it maintains varying degrees of collaboration.
The Gulf of Maine Council on the Marine Environment is a coalition of state and provincial governments: Massachusetts, New Hampshire, Maine, New Brunswick, and Nova Scotia. Federal agencies in both countries, while not formally Council members, play an influential observer role on many of its committees, and provide much of its funding (state and province contributions have been mostly in-kind).
The Council’s role in state and provincial governments is only advisory, rather than legislative or executive: it "encourages" its partner agencies to carry out the various elements of its Action Plan. According to the Canadian co-chair of the Committee that oversees EDIMS, this informal relationship lets Council committees define shared initiatives without much administrative baggage or political opposition. However, it seems to have limited the Council’s influence on the agencies it’s supposed to represent, and it may have kept the Council’s Committees overly dependent on the strengths and interests of their individual participants (rather than founded on a clear charter shared and supported by partner agencies).
The Committee that oversees EDIMS interacts regularly with the Council’s other Committees. In particular, it has worked for several years with the Council’s Monitoring Committee to publish coastal ecosystem indicators online. Also, wider public access to the Web in recent months has prompted interactions with the Council’s Public Participation Committee.
EDIMS also has relationships with several major research centers in the Gulf of Maine region, including the Bedford Institute of Oceanography in Nova Scotia and its American counterpart, the Woods Hole Oceanographic Institution in Massachusetts. The Council is also affiliated with two groups of researchers on the Gulf of Maine, the Gulf of Maine Regional Marine Research Board and the Regional Association for Research on the Gulf of Maine. Their relation to EDIMS is strengthened by the participation of the University of New Hampshire’s Ocean Process Analysis Lab in both groups.
Other groups related to EDIMS include the Canadian Atlantic Coastal Zone Information Steering Committee (ACZISC), which coordinates the production and interchange of coastal and ocean data in the Canadian Atlantic provinces (Nova Scotia, New Brunswick, Prince Edward Island and Newfoundland and Labrador). ACZISC has a Database Directory Working Group, which maintains and regularly updates a directory of datasets on Atlantic Canada’s coastal zone. The Canadian portion of the EDIMS dataset directory was drawn from this directory. The chair of ACZISC, and the chair of the Database Directory Working Group, both sit on the Committee that oversees EDIMS.
Finally, besides resource managers or researchers, several political associations are part of the EDIMS context, including the New England and Maritime Governor’s Conference and the Council of Maritime Premiers. Their influence on EDIMS is indirect: their overall policy recommendations influence the Council’s priorities and thus its demands on EDIMS.
In addition to its institutional setting, EDIMS is tied to an increasingly complex technological context. For several years, one impediment to its growth was that most of its intended audience (resource managers and planners outside the university research setting) had no access to its information resources. Most state and provincial agencies in the region did not have Internet access, or adopted it in a hesitant, piecemeal fashion. (For instance, the first chair of the EDIMS Committee had to rely on a personal Internet mailbox for professional purposes.) The advent of Internet access in a majority of the region’s agencies represented a boon for EDIMS, enabling its wider use and adoption among resource managers and state planners. On the other hand, now that these people have so many other sites on the Web to compare it with, they may demand a lot more from it. Already, members of the Public Participation Committee are demanding a "more interesting" front page for EDIMS.
Although Internet access is taking hold overall, the pattern varies a lot across the region. At one end of the spectrum, Nova Scotia’s provincial agencies and Environment Canada have used the Internet for several years to communicate and share information. At the other end, some key agencies in New Hampshire remain hesitant or even unwilling to obtain basic Internet access. Major networking initiatives are underway in most of the states and provinces: highlights include Maine’s NYNEX-funded network linking state agencies, schools, and libraries, Nova Scotia’s province-wide fiber-optic network, and New Hampshire’s ResourceNet, linking its state planning office and geographic data center.
The states and provinces each have a center for Geographic Information Systems (GIS) which archives digital information on land and natural resources and provides technical assistance to partner agencies. As of mid-1996, all of these GIS centers are looking into some form of online distribution of data and metadata. Their relationship to EDIMS ranges from close collaboration (the Committee’s Canadian co-chair represents the Nova Scotia Land Use Committee, which coordinates all of Nova Scotia’s GIS work) to mere acquaintance (New Hampshire’s GIS center has had almost no interaction with EDIMS—despite being housed in the same building at the University of New Hampshire). Even though these GIS data centers emphasize land, not ocean, resources, most involved agree that their overlap with EDIMS would be fruitful.
Given the goals set forth for EDIMS, and its institutional and technological context, it becomes possible to evaluate the choices—some organizational, others technical—made in its design and implementation.
The most significant organizational choice in building EDIMS was probably the decision to host it at the Ocean Process Analysis Laboratory at the University of New Hampshire. This ensured easy access to advanced computing resources (such as a Unix server) and to the research community. The university setting also allowed EDIMS to survive an unfunded downtime: between late 1993 and mid-1995, University of New Hampshire staff and students not only kept it online, but turned it into a Web site and added a few information sources to it. However, a Committee member in Nova Scotia suggested that low participation on the Canadian side may have been due to a perception of EDIMS as "a UNH research project, down there in New Hampshire."
The first two years of the EDIMS Committee’s existence were busy ones. It commissioned an in-depth survey of user needs, compiled a regional directory of coastal and marine datasets, and built the prototype telnet interface to this directory and ftp access to a few datasets. This directory was a centerpiece of the EDIMS design: it described data formats, dates, sizes, and distribution contact points. The Canadian portion of this directory was drawn from the database directory compiled by the Canadian Atlantic Coastal Zone Information Steering Committee (ACZISC).
Although the Council’s intent for EDIMS was to link distributed environmental information resources through a "common regional protocol," the initial EDIMS prototype was housed on a single server at the University of New Hampshire. Through the database directory, EDIMS was to evolve from a centralized archive into a distributed system that would draw on data suppliers throughout the region. However, few keepers of related data rose to the task of putting their data online, so EDIMS has remained a central archive. The delay in moving to a distributed data model has caused some tension with strong proponents of decentralized systems. The Canadian provinces, in particular, had major concerns about creating a centralized database. These concerns may have arisen from their Land Records Information System, a regional initiative to create a centralized database some 20 years earlier. This was eventually abandoned in favor of a cooperative effort to create, coordinate, and maintain a center for "primary data" (geodetic control, basemaps, parcel maps), which split along provincial lines after a decade of data development due to a variety of technical and organizational difficulties. (As a single organization, it could not maintain the expertise needed to oversee all scientific and socio-economic land-related data for three provinces; furthermore, the provinces were reluctant to give up control over their information to this "outside" data center.) Some may have perceived EDIMS as another such centralized archiving effort, and felt that it should be distributed even in its prototype phase.
A third highlight of EDIMS’ technical design was the choice of networking protocols — or lack of choice. The team at the University of New Hampshire advocated using the Internet Protocol (IP); but other members of the Committee felt this would exclude government agencies, and maintained that other forms of networking (Banyan, modem) were needed. Some on the Committee would say that "the EDIMS concept was a bit ahead of its time; suddenly the Internet made it all possible." In late 1993, IP began its explosive expansion through every sector of business and government—but by then EDIMS had gone "dormant" for lack of funds. In essence, the EDIMS Committee seems to have missed the opportunity to build an advanced information service when finances and support were relatively plentiful (though not all on the Committee would agree that funding was ever adequate for the EDIMS goal); it now finds itself merely maintaining a service whose functionality (by the World Wide Web’s rapidly rising standards) may seem limited.
5. Information sharing characteristics
The EDIMS effort can be evaluated in several ways. The following paragraphs outline several measures of the size and "quality" of EDIMS as an information sharing infrastructure, beginning with the kinds of information sharing it provides.
a. Forms of information sharing
EDIMS is primarily a repository and clearinghouse for coastal and ocean information, accessed through the Internet’s World Wide Web protocol. The target audience is researchers and resource managers in the Gulf of Maine (though actual usage patterns may be quite different—see below).
A main feature of the EDIMS Web site is the directory of coastal and marine datasets, stored in a relational database and queried through a World Wide Web forms interface. Users can search on keywords, geographic coordinates, and date ranges. The EDIMS front page also points to several sets of data archived locally, including (i) sea-surface temperature maps (several images a day since mid-1993); (ii) hydrographic data from the Atlantic Fisheries and Adjustment Program (AFAP), from 1993; (iii) oceanographic indicators in the Massachusetts and Cape Cod bays, compiled in 1990-91; and other bathymetric, meteorological, and ecological datasets. EDIMS also supports the Council’s administrative communication needs through several mailing lists, electronic-mail forums, and bulletin-board systems.
One measure of an information sharing infrastructure is the number of sites that provide information. In the case of EDIMS, most of the data is still local at present, but a few outside servers are being linked in via hypertext: the US Geological Survey and NOAA at the Woods Hole Oceanographic Institution in Massachusetts; Environment Canada; and others. Besides those who have put their own data online, several others have contributed data to be served up on the EDIMS server. These include the Massachusetts Dept. of Coastal Zone Management, the Bedford Institute of Oceanography, and the US Fish and Wildlife Service's Gulf of Maine Program.
Another measure is the volume of data available through EDIMS. Besides its hypertext pointers to agencies and providers in the region, the EDIMS server itself provides about 170Mb of data, including sea-surface temperature maps (101 Mb); (ii) Atlantic Fisheries and Adjustment Program (AFAP) data (25 Mb); and (iii) Massachusetts and Cape Cod oceanographic data (23 Mb).
A third measure is the traffic volume on the main EDIMS server, and the number of distinct sites using it. According to its logs, in the month of July, 1996, it transferred 22 Mb (3,300 files). (By comparison, the "hub" server for the Great Lakes Information Network transferred 51,000 files totaling 485 Mb in the same month.) EDIMS’ traffic grew by about 19% per month from January to July, 1996. Usage figures for EDIMS, due to their small size, are quite sensitive both to individual users (a persistent user may add 10 or 15% to the month’s usage in a single afternoon) and to automated Web search engines (which account for 15-25% of all data requests, more than any actual user). Top users (other than search engines) include the State of Maine and Canada’s Department of Fisheries and Oceans, but also the Institute for Global Communications and the University of Washington— both well outside the "target" audience. 1996 has brought dramatic growth in the number of distinct sites accessing EDIMS, from just over 100 in January to nearly 800 in July. This is most likely due to the increased use of Web search engines, and suggests that most or all of EDIMS’ growth in traffic may be outside its intended audience.
In summary, EDIMS is still an infrastructure of modest size, growing at an average rate, with an increasingly diverse user group. Although it is being used by researchers and resource managers in the Gulf of Maine, their use has grown only slowly, and is outweighed by that of outside users.
c. Quality of the shared information
Having provided several measures of the size of this infrastructure, we now turn to an assessment of its "quality." This is in part a property of the information it provides to users: its precision, accuracy, timeliness, and usability.
i. Precision and accuracyAs mentioned earlier, one of the key features of the EDIMS server is the Gulf of Maine Dataset Directory. EDIMS Committee members generally consider this a useful resource; but it was compiled in 1991-1992 and has not been updated since. Many of these listings, especially contact names and telephone numbers, are probably out of date. Perhaps for this reason, the directory sees only light use. (A comprehensive directory update was planned beginning in mid-1996.)
Besides this directory, the actual datasets on EDIMS have only seen light use as well, so their precision or accuracy have not been a great concern so far. Nonetheless, each set of data on the EDIMS server is accompanied by a free-text description of survey methods, time frame, data formats, and often a published report that analyzes and documents the dataset. This should at least help users decide whether the precision and accuracy of the data are adequate for their purpose.
ii. Timeliness and concurrencyThese criteria measure how well the data’s update frequency matches that of the phenomenon being measured and that of the user’s analysis or decisions. These criteria have only limited applicability to EDIMS, given that most of its data are fixed measurements made at a specific date and time. One exception: the Sea Surface Temperature maps, which the EDIMS coordinator at the University of New Hampshire transfers from a NOAA server every morning. (Interestingly, one of the few instances of EDIMS user feedback came from a group of tuna fishermen, who sent electronic mail requesting more frequent updates to the sea surface temperature maps.) Another exception is a set of data on New Brunswick river flows, which until March 1996 was updated daily from a server in Saint John, NB. Here again, because of the light use of these data, their timeliness or concurrency have not been a prime concern.
iii. Usability and encapsulationAnother measure of EDIMS information: the amount of knowledge and tools needed to interpret and use it. Obviously, EDIMS’ ease of use varies among its uses and users. Administrative information, such as e-mail forums and address lists, is easily accessed through a Web browser such as Netscape. The dataset directory is easy to search through the forms interface, given some knowledge of geography and either oceanography, fisheries, or meteorology. The archived datasets are a different story: they are provided as ASCII tables and files whose correct use requires tools and expertise beyond the reach of the typical practitioner.
In its planning for distributed data, EDIMS is looking into use of a protocol devised by the international Joint Global Ocean Flux Study (JGOFS). This protocol would provide transparent access to distributed data in a variety of formats—a key ingredient of the EDIMS design. However, the JGOFS user interfaces, built to work with modeling software, are still quite complex: the preferred client among JGOFS implementers is Mathworks’ MatLab function-analysis software. Another candidate data protocol, the Distributed Ocean Data System, under development at the University of Rhode Island, has an expanded set of user-interface options, though still geared to scientists. In general, fitting a widely usable front end onto a distributed data protocol will be key to EDIMS’ data architecture, and to its wider acceptance as a useful, reliable data clearinghouse.
d. Quality of the information infrastructure
In addition to information characteristics, the quality of an information sharing infrastructure depends on aspects of the sharing mechanism itself: its flexibility for two-way interchange, its ability to grow in size, and its ease of integration with diverse local data systems.
i. ReciprocityThis aspect of the infrastructure describes how easily information receivers can also be providers. In principle, this is easy to achieve: to include new data providers, just add hypertext links to their server at the EDIMS "hub." In practice, though, maintaining a server with dynamic operational or scientific data (i.e. not just electronic "brochures") requires expertise that most EDIMS participants currently lack. EDIMS does provide a mailing-list server and a Web-based bulletin-board system, which enable two-way, free-text communication among participants. These are helping to establish EDIMS as a focal point for the Council’s (and, more generally, the Gulf of Maine’s) regional communications and data development. But overall, EDIMS is still far from its goal of linking distributed data throughout the region. Towards that end, EDIMS staff have been looking into a variety of means (including JGOFS, as mentioned above) to link in distributed data sources around the region; as these sources begin to put data online, EDIMS’ Web-based design should be easily able to link in their contributions.
ii. ScalabilityEDIMS’ Web-based design also makes it easily scaleable: if data sources are maintained on autonomous Web servers, it’s easy to provide hypertext links to them. But again, few data sources are online: most data are sent in their native format to data coordinator Karen Garrison, who manually installs them on the EDIMS web server. This method is still manageable given the small number of contributors and data sets, but is by no means scaleable: the contributors, datasets, and datatypes should grow from the present handful to dozens or hundreds, installing and updating data on the EDIMS server would become a certain bottleneck.
iii. Non-intrusivenessTo date, this aspect of the infrastructure has not been a prime concern, because few of its participants have considered how to link EDIMS to their work or their own data systems. Interestingly, the EDIMS design contains little or no discussion of standards; current designs have favored setting as few standards as possible. In particular, the EDIMS committee has only recently begun to evaluate metadata standards and reconciling the Canadian and U.S. standards. Once data sources start to come on-line, a shared data protocol such as JGOFS may prove useful as a "minimal" standard, letting data maintainers keep their information in whatever form is most useful to them locally, while still allowing outside access to it in a consistent fashion.
Perhaps the most important assessment of an information infrastructure is its impact on people’s work: their efficiency, effectiveness, or satisfaction.
To date, EDIMS has had few tangible impacts on resource management or policy decisions. The Council’s policy decisions are still generally independent of EDIMS information; and the individual states and provinces continue to manage coastal and marine resources within their political boundaries, with little attention to regional impacts or trends, and with minimal use of the information or communications provided by EDIMS.
Nonetheless, several participants in the EDIMS Committee report that it has facilitated informal communication, cooperation, and trust between organizations. Others praise EDIMS as a cheap and effective replacement for telephones or mass mailings. So, although its tangible impacts to date may be minor, EDIMS may be laying the human and organizational groundwork for future information-based collaboration.
In particular, under its revised five-year Action Plan (1996-2001), the Council is calling for better public access to ecological data, for consolidated information on toxic contaminants, and for improved communication among research and monitoring programs. These elements in the Action Plan suggest that the Council is beginning to incorporate regional environmental data into its objectives. They also indicate an emerging reliance on EDIMS as a central tool in working towards those objectives. As the Council’s priorities become more information-intensive, EDIMS may yet get to play its intended pivotal role in guiding environmental decisions in the Gulf of Maine.
As detailed earlier, EDIMS was expected to be further along than it is in coordinating information resources and supporting regional decision-making in the Gulf of Maine. EDIMS participants attribute its slow development and acceptance to a combination of four factors: cultural difficulties (researchers are slow to structure and release their data; state personnel neglect to read their e-mail), organizational inertia (the Council’s thinking and decision-making make little use of physical data), technical hurdles (there’s still no easy interface to the data sets), and financial constraints (current funding allows for little more than maintenance).
Front page Table of Contents Abstract Chapters 1 2 3 4 5 6 7 8 9 Bibliography