Review of NASA's Distributed Active Archive Centers (1999)

The objective of NASA's Earth Science Enterprise (ESE) is to obtain a broad-based understanding of the functioning of the Earth as a system, with particular reference to global environmental change. The data needed to study these processes are necessarily diverse, comprising observations from a wide variety of remote sensing and in situ instruments and experiments, collected by

several government agencies and academic institutions, at different temporal and spatial scales. NASA's Earth Observing System Data and Information System was built to provide a means for scientists to integrate disparate data types collected by NASA and to study earth processes in a more comprehensive manner than was possible before.

EOSDIS includes many players—the EOSDIS Core System (ECS) contractor, the science and instrument teams, the DAACs, and the Earth Science Data and Information System (ESDIS) Project—each with well-defined roles. As originally conceived, these roles were as follows:

• the ECS contractor designs the EOSDIS Core System to capture, process, and distribute data from the EOS instruments and provides the necessary hardware and software to the DAACs;

• the science and instrument teams develop algorithms for creating data products;

• the seven DAACs process and disseminate remote sensing and in situ data and data products (land, atmosphere, ocean), and provide services to a wide variety of users (primarily scientists and NASA's partner agencies); and

• NASA's ESDIS Project sets the program requirements and provides funding and system-wide coordination.

Technical problems leading to delays in the ECS, however, have led NASA to rethink these roles. NASA's current plans are to make EOSDIS a more distributed system by having the science and instrument teams, rather than the DAACs, do much of the data processing. In addition, some of the DAACs will be permitted to develop and use their own systems, rather than the ECS, for managing data. Indeed, DAAC-unique information systems are already being used to process and distribute data from the first EOS-related mission, the Tropical Rainfall Measuring Mission (TRMM), which was launched in November 1997. By involving a broader array of constituencies in EOSDIS, and employing diverse approaches to providing data services, NASA is taking another step toward creating an EOSDIS federation.

In the new EOSDIS model, the role of the DAACs may be strengthened or diminished, depending on how much flexibility and authority NASA management is prepared to give and how much initiative the DAACs are prepared to take. However, serving the needs of their users will remain their most important task. Based on the site visit reports of the seven DAAC review panels and a user survey, the Committee on Geophysical and Environmental Data (CGED) concludes that most of the DAACs are serving their scientific user community well. (Indeed, the Physical Oceanography [PO.DAAC] and the National Snow and Ice Data Center [NSIDC] DAACs are model in this regard, although neither has to face the immediate challenge of handling the enormous data streams of the AM-

1 platform.) Moreover, each DAAC occupies a unique, scientifically important niche in the Earth Science Enterprise; closing any of them would reduce NASA's ability to meet its scientific objectives. Consequently, even DAACs with significant problems in fulfilling their missions (i.e., the Alaska Synthetic Aperture Radar [SAR] Facility [ASF] DAAC and, to a lesser extent, the Earth Resources Observation Systems [EROS] Data Center [EDC] and the Oak Ridge National Laboratory [ORNL] DAACs) should be nurtured by NASA and the ESDIS Project so that they succeed.

The individual panel reports provide detailed recommendations on improving the operations of each DAAC. However, all of the reports have three themes in common: (1) the scientific need for a coherent system of DAACs; (2) the importance of strategic planning in routine data center operations; and (3) the need for flexibility, vision, and leadership as EOSDIS evolves. The following recommendations, which focus on these themes, were based on analysis of the panel reports, the user survey, discussions with NASA managers, and the committee's experience with world and national data center reviews.

EOSDIS is more than the hardware and software needed to link the DAACs; it is a tool for achieving the science goals of the Earth Science Enterprise. The objectives of EOSDIS include the following:

• facilitating the creation of standard data products, thereby permitting the immediate scientific goals of the science teams to be realized;

• catalyzing the preparation of a wide range of secondary data sets and information products that combine information from different satellites and in situ sources, thereby stimulating collaborative, multidisciplinary research;

• making such products readily accessible to the broader scientific community; and

• preserving them in usable form for future generations of scientists.

The DAACs have a key role to play in meeting each of these objectives. First, by working with the science and instrument teams (as the Langley Research Center [LaRC] DAAC has on Earth Radiation Budget experiments), the DAACs can help generate data products and record metadata that might be viewed as irrelevant or common knowledge by science and instrument team members. The latter is vital if the data and data products are to retain their scientific value over several decades. Second, by working with the broader scientific user community, the DAACs can build a mutually beneficial relationship that results in (1) the development of useful tools and services, (2) the acquisition of scientifically important data sets, and (3) better answers to complex user queries. In practice,

however, only a few DAACs (namely the PO.DAAC and the NSIDC DAAC) have a strong relationship with the scientific community, despite their generally good interaction with User Working Groups and their collocation with scientists with expertise in the DAAC holdings.

Multidisciplinary researchers interested in understanding earth system science processes present a special challenge to the DAACs. These researchers need to obtain useful data from more than one DAAC. Indeed, it was to serve their needs that the concept of ''one-stop shopping'' (i.e., users can access the system through any DAAC, search all the EOSDIS holdings, and obtain the relevant data) was born. With the delays in the ECS and the move toward federation, however, one-stop shopping may no longer be feasible. For EOSDIS to serve the earth system science constituency, the DAACs will have to overcome their differences and begin to act as components of a coherent system that (1) is able to manage high-volume data streams from coordinated instruments and (2) offers comparable access technologies, a consistent terminology, and unobtrusive authorization procedures. If such a system is not possible, NASA will have to provide sufficient resources (e.g., through requests for proposals to participate in NASA's federation) to enable users to find useful information from a disparate collection of DAACs. The development of Version 0 of the EOSDIS information system demonstrated the ability of the DAACs to work together. With the proper incentives from ESDIS, the DAACs should also be able to create common tools and standards that would enable users to locate, access, and combine various types of data in the EOS era.

DAACs differ from data centers by focusing on the early, most scientifically active part of the mission or experiment, rather than on long-term data stewardship. To fulfill the science objectives of the Earth Science Enterprise, however, the DAACs also have an obligation to ensure that the data in their charge remain useful for future generations of scientists studying long-term global environmental change. Consequently, the DAACs must be involved in all stages of data management—from data collection, to management of active data sets, to long-term archive. Such involvement requires strategic planning.

To date, however, few DAACs have engaged in strategic planning, in part because ESDIS and the ECS contractor have done so much of their planning for them. (The PO.DAAC is an exception.) However, now that NASA is adopting some of the DAAC fall-back systems for processing and handling data in place of the ECS, the DAACs will have a greater voice in their own evolution and in the development of EOSDIS. To take advantage of this opportunity, each DAAC should create a vision and an implementation strategy. The strategy should permeate everything the DAAC does. For example, it should describe how the DAACs will become involved in collecting the metadata and in calibrating and validating observations from the EOS instruments. The latter is particularly important for the ORNL DAAC, which has important in situ data sets, but no concrete plans for participating in calibration or validation activities with the relevant science teams. Without the integration of remote sensing and in situ observations, it is unlikely that the EOS program will ever fulfill its potential.

The vision and implementation strategy should also describe how the DAACs manage their active data sets. This is a primary focus of the DAACs and includes knowing who their current users are, how they use the data, what tools are available for serving them, and which ones must be developed in house. It also includes looking ahead to identify and provide services to potential new user communities and to keep up with rapid advances in technology, particularly those involving storage and communications. Most of the DAACs manage their active data sets well and provide a reasonably high level of service to users. The PO.DAAC, the ORNL DAAC, and the NSIDC DAAC are particularly effective in this regard, whereas the ASF and EDC DAACs have to substantially improve user services. All the DAACs, however, would serve their users better by (1) having a more detailed understanding of their user profile and (2) developing quantitative measures for tracking performance. A detailed user profile, moni-

tored over time, will enable the DAACs to provide specialized services to current and potential user communities and to expand their user base. Promoting new uses of the data, which will arise inevitably with the emergence of new user groups, will increase the taxpayer's return on investment in the EOS satellites. In addition, by developing and tracking quantitative performance measures, the DAACs will be able to assess the technical performance of individual system components, as well as the overall success of each DAAC and the entire DAAC system in meeting the needs of their users.

With the delays in the ECS, NASA has no choice now but to abandon many components of the ECS and move toward a more flexible federated structure for EOSDIS. Changing management models at this late stage poses a tremendous challenge to NASA. For the new EOSDIS paradigm to succeed, NASA must (1) empower the DAACs, science teams, and Earth Science Information Partners (ESIPs) to provide the best services and most innovative data products possible; (2) offer incentives for them to collaborate to provide users with a common look and feel to the information system; and (3) emplace safeguards to ensure that the fundamental goals of EOSDIS are met. To accomplish the first, ESDIS will have to delegate some of its authority for serving users and designing their own infor-

mation systems to the DAACs. It has done so to some extent with the Goddard Space Flight Center and LaRC DAACs, which successfully developed TRMM-specific information systems. Similar authority should be granted immediately to the ASF DAAC, which is currently powerless to guide development of the information system it is required to use (which was designed and controlled by the Jet Propulsion Laboratory). In developing their information systems, some of the DAACs may wish to use parts or all of the ECS, but this should not be a requirement for remaining within EOSDIS. In return for this newfound authority and flexibility, the DAACs must agree to act as components of a coherent system for the greater good of the Earth Science Enterprise (see "An Information System for Science," above). This is particularly important for the LaRC DAAC, which has no intention of linking its custom-built systems with EOSDIS.

In summary, every DAAC has a unique and important role in some part of EOSDIS and the Earth Science Enterprise, with little overlap or redundancy. The DAACs that will receive data from the AM-1 platform and other near-term launches are reasonably placed to address the challenges that launch will entail, given the uncertainties associated with the EOSDIS Core System. However, for the DAACs to fulfill their mission of providing seamless access to multidisciplinary data sets, and thereby promoting creative scientific analysis of the data, EOSDIS will need inspired leadership that empowers its stakeholders to fulfill their special roles within an integrated network of information centers.