James J. Anderson
Center for Quantitative Science and School of Fisheries
University of Washington
May 13, 1995

Web groups in adaptive management

Management of the Columbia River fisheries resource is now employing the World Wide Web (or Web). A coordinated group of Web sites exchange information on the status of the Federal Columbia River Power System (FCRPS) and fish migration through the river. The Web is evolving quickly with traditional concepts for data management being replaced by the emerging multimedia hypertext capabilities. This is nothing short of a revolution. It is fast moving and far reaching. In particular, the Web is a powerful tool to apply principles of adaptive management. In adaptive management, actions are taken under scientific uncertainty and are subsequently monitored and scientifically evaluated. Through this coupling of actions and evaluation knowledge is gained for better management.

Many protocols and strategies for serving data through the Web are being developed. Some of these will disappear, others emerge, as the community gains experience in using the Web. To guide this evolution we must understand how information is developed and used. For Columbia River management several Web sites are working in a structured and interactive manner. Together this group of sites comprises the rudiments a hydrosystem management system. This concept of a "Web Group" naturally evolves from the Web structure. A "Web Group" by its nature has a central purpose, is distributed and non-hierarchical. This document outlines fundamental processes of the Web Group being developed for adaptive management of salmon in the FCRPS.

Data in management

Data for resource management is distilled and synthesized from its initial collection as observations on a resource to its use by decisions makers to manage the resource. In Columbia River fisheries management, part of this information is handled in nearly real-time with daily collected data being archived and analyzed for weekly decision making on hydrosystem operations. In addition to the real-time exchange, historical data is required for comparisons to the current conditions, for analysis of the ecology, and for calibration and validation of predictive models. The Web Group coordinates these data distribution and transformation functions.

Data is the currency of resource management, and the data distribution system is ultimately driven by the uses of the data in an adaptive management cycle (Fig. 1). Information on resource status is input into primary data sites that archive real-time and historical data. Secondary database sites, accessing the primary sites, add value to the data by analyzing the information with a variety of modeling, statistical and graphical tools. The secondary site information, filtered through the political process, is then used to develop actions for resource use. The effect of the actions on the resource are monitored and the new information is input back into the system completing a cycle in the adaptive management process.

This cyclical process contains four essential elements or activities: resource monitoring and experiments, data archiving, data analysis, and a decision process. The Web becomes the structure to exchange information between these elements. Each element may involve a number of interactive and competing activities. Moving through the information flow from monitoring to decisions, the information is distilled and predictions on the consequences of actions are formulated. With the predictions comes uncertainty and disagreement. This conflict is implicit in resource management. The goal of the Web Group is to clearly articulate differences in and justifications for alternative management strategies by providing users tools to trace through the decisions processes in an ordered and scientific fashion.

Fig. 1 Information flow in adaptive management of Columbia River resources.

Interaction of primary and secondary database sites

By the flexible nature of Web tools, a number of primary and secondary database sites can provide many of the functions and features of data handling, analysis and display. What distinguishes sites is their essential function. Primary sites obtain data from field monitoring and are responsible for maintaining data integrity and serving raw data to other sites. Secondary sites add value to the data through statistical analysis and modeling activities. Data maintained by the secondary sites includes the results of the analysis and the analysis programs themselves. This broader definition of data originates from the concepts of object oriented programming, which itself is a revolution in programming. Secondary sites can be categorized in terms of who accesses them. "Analysis" type secondary sites are used by scientists who perform statistical analysis and make predictions on the outcomes of resource management actions. "Decision" type secondary sites further structure and distill the information for use by decision makers. The public, government officials, and decisions makers are the main users of decision sites.

For example in the FCRPS Web, the Army Corps of Engineers is a primary site that collects and maintains river environmental data. The University of Washington is a secondary database site that develops and maintains models for the analysis of fish responses to flows. As a result of a variety of analysis activities and types of monitoring data, there are a number of primary and secondary sites in a Web Group.

Both primary and secondary sites may maintain similar databases in-house. The choice of where information resides depends on the speed at which data can be accessed and the availability of technical help and software to the users. Routine database requests can be handled from a centralized database server, but for more specialized uses of data it is preferable to keep databases at the site where the analysis is being conducted. As the World Wide Web activity increases redundant database services have evolved. For example, the popular Web page maintained by the Louvre in Paris has expanded into a worldwide WebMuseum network with twelve redundant sites. The important issue is not whether sites have redundant information. What is important is that each site provide the Web Group with the most updated version of its portion of the data, be it raw data, programs to analyze data, or public comments on proposed management. Thus, a site's purpose is not defined by the data on hand but by its responsibility in maintaining data within a Web Group.

Getting on the Web

User access to the Web Group is as essential as the Group itself. Several levels of access exist depending on the hardware and Internet connections available to the uses. PC users, at a minimum, must have Microsoft Windows, a 386 computer with 4 MB of RAM, a hard drive and a modem. Mac users minimum requirements include System 7, 6 MB RAM, a hard drive and a modem. Access is through a commercial Internet provider such as America Online (AOL) or MCI. With the minimum configuration, users can access multimedia hypertext functions of the World Wide Web. It is also possible to run models remotely from server computers. This remote client server protocol requires the user to run X-terminal emulation software, which is also available for free on the Internet.

Columbia River Technical Management Team Web Group

The Columbia River operations for fisheries management currently involves a number of primary and secondary database sites that make up the Columbia River Technical Management Team Web Group (Fig. 2). The immediate purpose of the group is to assist river managers in making weekly decisions on operating the FCRPS to improve fish passage. The Web Group also serves to distribute other types of information on the Columbia River. This secondary task at the present time falls under the responsibilities of the Coordinated Information System (CIS) which is not currently on the Internet.

Primary database sites

Three primary databases receive data from monitoring programs. Non standard data collected under individual research projects are not generally input to primary databases, although the CIS is intended to track such data. For management of juvenile salmon migration in the Columbia River three primary databases sites now exist:

• Army Corps CROMS database provides hydrological information. The CROMS database is accessible through the Internet but it is not supported on the Web. Data is downloaded daily into the secondary site at the University of Washington.
• Pacific States Marine Fish Commission (PSMFC) provides information on individual smolt passage through dams from the PIT tag data system and information on adult fish catches from the CWT database system. The PSMFC site is connected to the Internet but does not currently maintain a Web site.
• Fish Passage Center (FPC) provides information on the adults counts, total dissolved gas information, juvenile passage indexes and some information on fish condition. The FPC is accessible via a modem bulletin board.

Secondary database sites

Secondary database sites in the Columbia River Web Group provide analysis software, analyze data and present results to users, primarily the hydrosystem managers. Traditionally results from secondary sites are made available through publications, newsletters and journal articles. With the Web, results are available across the Internet. Three prototype secondary sites are currently in the Columbia River Web Group:

• Pacific Northwest Laboratory (PNL) maintains a homepage of information exchange for the Columbia/Snake River Interagency Technical Management Team (TMT). The purpose is to facilitate communication among participants in the TMT through timely access to data and information.
• University of Washington Center for Quantitative Science (UW) maintains a homepage that serves information on fish and the river environment, models for predicting fish migration and survival through the river system, and models for managing salmon harvest.
• Bonneville Power Administration (BPA) maintains a homepage providing information on flow projections and analysis of alternative FCRPS operation.

Modeling groups

National Marine Fisheries Service and the Northwest Power Planing Council have identified the need to integrate existing models into a common framework to enhance their utility as analytical tools. To achieve this integration they propose to develop a regional center for biological analysis and modeling. Although a regional center is appealing, it is not practical since modeling expertise is distributed throughout the region. Model integration is better achieved through a Web Group devoted to modeling efforts. The existing Columbia River Web Group has envisioned this capability through the links between the PNL and UW Web pages. However, an effective and functional modeling Web Group must involve more than simple Web links. To coordinate modeling efforts free and detailed exchange of information, calibrated and validated models, modeling code, and executable programs must be made available to all members of the group. To the degree possible, models should share common modules including database access software, computation algorithms, graphical output tools and standardization in documentation.

Although model building is a flexible activity with differing goals there are well defined steps in developing models. These include identifying the relevant ecological interactions, expressing these interactions in equations, coding the equations, determining model sensitivity, calibrating the model, validating the model, documenting the model and finally training others to use the model for management purposes. Not all models will evolve through all steps. Prototype models, constructed to study ecological interactions, are at one end of the spectrum. At the other end are resource management models that have gone through all steps and are used for making decisions on resource allocations. Both types of modeling activity need to be supported and shared in the modeling Web Group.

Fig. 2 Current configuration of the Federal Columbia River Power System World Wide Web Group. Sites with Web homepages are shaded.