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

The National Snow and Ice Data Center (NSIDC) DAAC is hosted by the University of Colorado. Along with the NSIDC and the World Data Center (WDC) for Glaciology, with which it is inextricably intertwined, the DAAC manages data related to snow and ice, climate, and the cryosphere—the part of the Earth's surface that is perennially frozen. The NSIDC DAAC will also receive Moderate Resolution Imaging Spectroradiometer (MODIS) data from the GSFC DAAC and may create MODIS snow and ice products. Although no small task, particularly since the EOSDIS Core System (ECS) has not been completed, most of the EOS instruments related to the polar regions will be launched after

the AM-1 platform. Consequently, the DAAC will face its most difficult data management challenges in a few years.

The NSIDC DAAC provides an outstanding example of how good data management practices and a close relationship with researchers can help lead to scientific advances. Although no major problems were found during its site visit, the panel recommends that the NSIDC DAAC sponsor joint activities with the ASF DAAC on scientific issues pertaining to polar regions, which have not received adequate attention from ESDIS so far. The panel also recommends that the DAAC develop and implement a transition plan describing the critical path of DAAC activities prior to site acceptance of the ECS.

The National Snow and Ice Data Center DAAC was created by NASA in 1991 (Box 8.1). Its roots go back to 1957, when the World Data Center for Glaciology was established at the American Geographical Society in New York. The WDC relocated to the University of Colorado in 1976 with NOAA sponsorship, and a new data center, the NSIDC, was created in 1982. The NSIDC is by far the larger of the two organizations, and is funded by a variety of agencies, notably NOAA, NASA, and the NSF. The DAAC is larger still, and accounts for about 75% of the total operation. All three components are located within the Cooperative Institute for Research in Environmental Sciences (CIRES) at the University of Colorado.

All three components serve the cryospheric and polar science communities (Box 8.1). Although the operations and staff of the three centers are commingled, the holdings of the DAAC are distinct from those of the WDC and NSIDC. Current holdings of the DAAC include passive microwave and AVHRR products, altimetry and elevation data, and remotely sensed and in situ polar atmospheric science data.

The aggregate volume of these data sets, together with the holdings of the NSIDC and the WDC, is about 1 TB. In the EOS AM-1 era, the NSIDC DAAC will receive approximately 15–18 GB of MODIS data per day from the GSFC DAAC and will use them to produce MODIS snow and ice products. The DAAC will not be a direct recipient of high-volume Level 1 data streams until the Advanced Microwave Scanning Radiometer (AMSR) and the Geoscience Laser Altimeter instruments are launched in a few years (see Table 1.1 for a description of data processing levels).

To prepare for the upcoming missions, the DAAC is developing new products, testing MODIS algorithms, and working on the ECS release B testbed. All of its baseline hardware is in place, and the DAAC is reconfiguring hardware and installing the ECS commercial-off-the-shelf (COTS) software to prepare for Version 2. In terms of readiness for the EOS data streams, the greatest challenges that the DAAC faces are staffing up in time and planning the near-term transition to

the ECS. With regard to the first, the DAAC currently has funding for 33 FTEs, although several positions are vacant, and the DAAC will have to add 20 more FTEs over the next few years.

The Panel to Review the NSIDC DAAC held its site visit on March 4–5, 1998. The following report is based on the results of the site visit and on subsequent e-mail discussions with the DAAC manager in June through September 1998.

The snow and ice data archived and distributed by the DAAC (see Box 8.2) are a critical resource for the cryosphere research community. The need for

remote sensing data on snow and ice is likely to increase over the years as programs emerge addressing gaps in the predictive knowledge due to inadequate understanding of feedbacks related to the cryosphere. Large expanses of the Earth that have a permanent ice cover are remote and inaccessible, and remote sensing is the most, if not the only, effective tool for data gathering. This applies equally to glacial ice and sea ice. The need to detect changes over time will be a critical element. For example, the extent of snow cover in winter is an indicator of climatic conditions over the land areas in the northern hemisphere and is considered an index of warming conditions (e.g., Figure 8.1). A time series is therefore valuable for global change studies. Examples of other high-priority areas include the mass balance of glacial ice, which has implications for freshwater input into the marine environment, which in turn determines sea level rise and influences oceanic water column structure. Ice cover also affects surface albedo, modifying the energy balance of polar regions, thereby affecting global climate. These are only a few examples of why observations from space of snow and ice are critical.

Most of the data sets currently held by the DAAC are received as Level 1 products, then processed into higher-level products. The panel notes that infor-

FIGURE 8.1. Trend of snow-covered area for the northern hemisphere from 1978 to 1996. Above: visible band-derived (NOAA) snow-covered area departures from monthly means. Below: passive microwave-derived (SMMR and SSM/I) snowcovered area departures from monthly means. SOURCE: NSIDC DAAC.

mation on lower-level versions of these data sets is not included in the DAAC's catalog. Some scientists need access to these data sets, sometimes in real time, and it is important for all the data holdings to be made visible in the catalog.

For the upcoming missions, NASA is considering transferring responsibility for data processing from the DAACs to the instrument teams on an instrument-by-instrument basis. The possible change in plans is being driven by new delays in the ECS. If the original processing plans are adhered to, the NSIDC DAAC will distribute and archive Level 1 AMSR data and will provide NSIDC glacier

inventory information to validate ASTER data. It will also archive the derived Level 4 products from the ASTER Global Land Ice Monitoring System, although the ASTER images themselves will be made available by the EDC DAAC. Finally, the NSIDC DAAC will receive Level 2 MODIS snow and ice products from the GSFC DAAC and will create Level 3 gridded daily and composite products, which are not covered in the current MODIS instrument team processing proposals.

To prepare for MODIS product production, the DAAC has tested early versions of three out of four of the Product Generation Executables (PGEs). Drop 4 of the ECS software was delivered to the DAAC in April 1998, and the DAAC is currently testing the next version PGEs with the new software. None of the tests have used a simultaneous processing configuration, which is useful for determining whether the capacity of the system is sufficient.

Most users will not be able to manage the large files that will result from the EOS instruments. Subsetting will make the data more accessible to users and, in the DAAC's view, will probably be the biggest factor in increasing the size of the user community. Nevertheless, subsetting capabilities are not scheduled to be included in the ECS until 1999, after MODIS has been launched. Consequently, the DAAC cosponsored a workshop in July 1998 to discuss, among other things, the results of several prototype efforts for creating Hierarchical Data Format (HDF)-EOS subsetting tools. As a result of the workshop, the DAAC plans to test a subsetting tool developed at the University of Alabama, Huntsville.

NASA and NOAA are currently negotiating plans to transition data from the DAACs to NOAA archives. The NSIDC has both funding and organizational links to NOAA, in the latter case, through a Cooperative Agreement between CIRES-University of Colorado and NOAA's National Geophysical Data Center (NGDC). Consequently, if NGDC becomes the designated archive, the transfer of data from the NSIDC DAAC to the NGDC should be one of the easiest to plan among all the DAACs.

The NSIDC DAAC tracks its customers by means of several criteria, including the number of users by category (e.g., university, federal employee, commercial sector), the number of data sets distributed, and the monthly volume of data

downloaded from the Web. The NSIDC-WDC-DAAC complex also maintains an impressive list of several hundred journal papers than have used its data. The users of data and data products are multinational and diverse in discipline and technological sophistication.

High priorities of the DAAC include determining the scientific impact of data sets and ensuring customer satisfaction. A needs and requirements database that keeps track of user satisfaction is essential for the latter. Such a database extends beyond the DAAC's present tracking system, and the panel believes that the DAAC can and should be more proactive in its assessments of the uses of its data sets.

The panel suggests the following as possible ways to improve the assessments:

• identify and track electronic downloads of NSIDC DAAC data;

• determine users' satisfaction not only with the DAAC's provision of the data set, but also with the data set itself (e.g., What were the primary limitations the user encountered with the data set?); and

• assess the scientific impacts of the papers that resulted from the data set.

The last could be done by DAAC scientists and/or outside experts. This more extensive documentation would provide a basis for setting internal priorities on how much effort to place on particular data sets, and for general informational purposes for funding agencies, for example. By synthesizing information on data set utilization and impacts, the NSIDC DAAC can play a valuable role in evaluating the data priorities of the snow and ice community. As the voluminous and expensive products of EOS come on-line, there may be a much greater need than in the past for this type of function. The panel notes that the NSIDC DAAC is well positioned in this regard and would be missing a potentially important opportunity if it does not enhance its documentation and assessment of data set utilization.

The DAAC has a good relationship with its Polar DAAC User Working Group (PoDAG). The PoDAG meets every eight to nine months and helps the DAAC decide such issues as which nonstandard data products should be developed. The DAAC is responsive to PoDAG recommendations, and PoDAG members feel effective and productive. PoDAG members stated that the NSIDC DAAC is well integrated into the scientific community and is not simply a number-crunching data center. This characteristic arises in part from its cooperation with and proximity to scientific research programs (see below).

Because of their overlapping interests in the polar regions, the ASF and NSIDC DAACs had a joint User Working Group for the first few years of their

existence. Apparently the joint group was not as effective as either the ASF or the NSIDC DAAC might have wished, and separate user working groups for the two DAACs were formed. The outcome, in the panel's view, was that the NSIDC DAAC gained an effective User Working Group, but at the cost of losing synergy with the ASF DAAC.

The NSIDC-WDC-DAAC complex has a long and impressive history of responding to the needs of snow and ice researchers. Active involvement on the part of technical personnel in the acquisition and development of data products, and the close juxtaposition of the external support function with active faculty and student in-house research, have resulted in an understanding of the modus operandi of scientific research on the part of the technical staff and in a proactive attitude. The panel notes that this cooperative and proactive attitude is a strong positive attribute and that the in-house scientific competence adds value to the data sets. The cryospheric and polar science research communities will continue to increase their use of satellite remotely sensed data, as scientists who are not currently among the remote sensing specialists recognize the value of the data products and become users. Such an expansion of the number and diversity of users can well be accommodated by the DAAC.

Although the DAAC clearly has an excellent relationship with its scientific users, the review showed that visitors and outside collaborators play a relatively small role in the DAAC's operations. Among the reasons given were a lack of available space and computer equipment for visitors, as well as a tendency for in-house scientific staff to do the ''value-added'' work such as (1) synthesizing data sets, especially data sets of the same variable but from different time periods and/or regions; (2) reformatting and gridding data sets to facilitate user access; and (3) providing quality control and producing "clean" versions of data sets that were originally contaminated. The HARA Arctic upper-air sounding data set is an example of an NSIDC-enhanced data set that benefited a large segment of users. At present, the NSIDC tends to do this type of work when an in-house researcher needs the value-added data set. Hence, this work is often driven by external funding of in-house projects. This does not guarantee that the tasks undertaken are those most needed by the community. Visitors at the DAAC, who could contribute to such efforts, would bring a nice "external drive" into this process.

The DAAC would benefit in several ways from a more active visitor program. First, it would be able to tap into the scientific expertise and data set usage experi-

ence of the broader community, which is more diverse than even a high-quality in-house staff can possibly be. Feedback from external scientists working at the DAAC will almost certainly be more substantive than feedback from remote users. Second, a strong visitor program would foster the community's stake in the DAAC and would counter any perception of its being a "closed shop" or competitor. Finally, visitors would disseminate first-hand information about the DAAC, its holdings, and its products, thereby enhancing the visibility of the DAAC.

Possible vehicles for enhancing external collaboration include dedicated resources (e.g., workstations, space, travel funds) for visitors and a wide solicitation of visitors. In addition, collaborative ties could be fostered through joint research proposals by NSIDC DAAC personnel and outside scientists. The latter strategy represents a significant step beyond the practice of sending representatives to a workshop and submitting a proposal to serve as the data archive for a particular program (e.g., NSF's Arctic System Science initiative).

The panel was impressed with the high level of user services that the DAAC provides to its customers. The balanced suite of analyzed products and data sets offered by the DAAC is much to the benefit of its clients. It also has an active CD-ROM publications program and services a significant number of regular subscribers. Subscribers to a series generally receive multiple products during a year. This is counted as a single request.

The method of counting requests is realistic in that the DAAC counts only requests for which it does work and supplies something to the user. It does not justify itself on soft figures such as hits on a Web site or requests that result in a simple referral to another center. Given the method of counting, the servicing of 1,500 to 2,000 requests a year (including more than 500 subscriptions) indicates that the DAAC is heavily used.

DAAC policy is that all requests receive a response within 24 hours. This does not mean that requests are always completed within 24 hours but that the request is acknowledged and the user receives information on how and when the data or information will be provided. This policy has contributed to a very positive relationship between the DAAC and its users.

The ECS software employs rectangular projections for gridding data. Rect-

angular projections, however, are inappropriate for high-latitude regions because of the distortion in horizontal distance, which becomes increasingly severe toward the poles. At the pole, the calculations fail. Consequently, polar projections are essential to the cryospheric and polar science communities. The National Snow and Ice Data Center has developed the Equal Area Scalable Earth grid, which has been adopted by the community for the Polar Pathfinder data sets. However, if EOS data are not translated into polar grids, the DAAC (and NASA) will completely fail its user community.

At the time of the review, the DAAC had tried unsuccessfully for five years to make support of polar grids a requirement of the ECS. An instrument team had also submitted a request to the ESDIS Resource Allocation Board for funding to develop a polar grid. A subsequent discussion between the panel and the ECS contractor revealed that ECS contractors have begun to address the problem, and the panel urges ESDIS to ensure that this critical functionality is developed and fully tested before launch.

The NSIDC DAAC seeks to balance the risks of prematurely adopting "cutting-edge" technology with the benefits that more modern systems provide. The DAAC therefore tries to be "sufficiently modern" with regard to hardware. The panel agreed that this is a sensible approach.

The network topology and communication infrastructure seem sufficiently modern to meet the needs of the DAAC for the foreseeable future. The storage capacity of the Storage Tek PowderHorn (300 TB) for on-line storage and archive is more than sufficient to handle the legacy data (1 TB) and the new data expected over the 10-year life-cycle. The server and workstation configuration are currently adequate to meet the needs of the DAAC, although the number of workstations available for non-DAAC use (i.e., visiting scholars) was raised as an issue (see Recommendation 2).

At the time of the site visit, the initial delivery of operational ECS software had not been made, and the DAAC was still working with Version 0. The panel was concerned about the configuration management of the software baseline and delivery process, given the number of sites and the mixture of core functionality and DAAC-specific functionality required. How will DAAC-unique functionality and "patches" be controlled, and how will they be integrated with future baseline upgrades and deliveries? The configuration management problem could be exacerbated if the DAAC runs Version 0 in parallel with Version 2 of the ECS. Maintaining multiple versions of the system software would also increase operations and maintenance costs.

Part of the problem is related to the lack of baselined standardized Application Program Interface (API) documentation, which was supposed to facilitate the development of DAAC-unique functionality and buffer the changes from the core functionality. To mitigate risk, the panel suggests that the ECS contractor work with the DAACs to set priorities for standardized API development and documentation, based on which functions are most critical, which will be used the most, and which are most feasible in terms of cost and schedule.

Operational testing of the baseline software in end-to-end functional tests is a critical step in determining whether the system is ready for production. Once this level of testing is completed (as part of the site acceptance tests), a detailed evaluation of the test reports, of outstanding hardware and software defects, of workarounds, and so forth, must follow as part of the assessment of operational readiness. At the time of the review, installation, checkout, and testing of the hardware were being performed by ECS contractors, and operational testing had not yet begun. In view of the timetable for the upcoming missions, the panel viewed this as an area of risk.

Although distribution of data on media will always be important, the DAAC regards Web distribution of its data sets and the development of home pages as major accomplishments. Over the past two to three years, ftp and Web access have increased sharply. The panel found the DAAC's Web site to be functional and useful, with many data sets that can be downloaded from the site.

The panel met with two managers at the site visit—Ron Weaver, manager of the NSIDC DAAC, and Roger Barry, director of the NSIDC and the VVDC for Glaciology. Barry obtained funding in the early 1990s to create the DAAC and serves as the DAAC scientist. He also orchestrated a user community letter-writing campaign, which probably saved the DAAC from closure in 1994. The panel views Barry's involvement as a tremendous strength of the DAAC, not only because of his strong connections to the scientific community and his leadership abilities, but because he is a faculty member. The latter is particularly important because of the clout it brings with the host university.

The DAAC leverages considerable scientific and technical expertise through its association with the University of Colorado, its role as a World Data Center for Glaciology, and its many connections with international science programs such as the World Climate Research Program. These diverse and important associations ensure not only enhanced scientific and technical guidance, but also the broadest global view of the user community. This broad view of user needs is much to the benefit of the Earth Science Enterprise.

The DAAC has adopted a "product team" approach to the development and management of its operations. A product team is responsible for a data set. The team is generally led by a scientist with expertise relevant to the data set, and other members are drawn from operations, user services, scientific programming, technical writing, and database administration. This approach helps ensure that all aspects of the life-cycle of the data set are covered. To further guide the development and management of the data sets, data set production and documentation checklists have been implemented.

The panel considered this approach to operations and development of data sets to be effective. The lessons learned by staff working on one data set are transmitted immediately to the other product teams on which a member works. Methodologies are well established and understood. The procedures necessary to ensure that the breadth of expertise to build and manage end-to-end systems for complex, interrelated scientific data sets have been designed and formalized. Coupled with the checklists to ensure compliance, there is clearly a good management system in place to keep operations on track and on schedule.

All DAAC staff, including the ECS contractors, are university employees. This further blurs the distinction between the centers and the university and be-

tween the DAAC and the ECS contractors. It also contributes to a strong sense of team membership. The staff have a mixture of expertise that is well suited to the mandate of the DAAC. In particular, there is strong scientific capability and leadership in the organization. This depth of scientific expertise has allowed the DAAC to develop a program and services that meet the real needs of its user base. In addition, the technical staff responsible for day-to-day operations, budgeting, and data set development were found to be knowledgeable, capable, and highly motivated. Considerable interest in the review was demonstrated by a large attendance at the open sessions with the panel. There was broad participation by all in the discussions, and contributions by the staff were relevant, positive, and to the point.

In general, the DAAC does not have a problem retaining staff. Turnover is about 8%, which is well below the rate of 30% in Colorado for all industries. It appears that the NSIDC complex offers an interesting and challenging working environment that results in the low turnover. Staff recruiting, on the other hand, appears to be more problematic. At the time of the review, there were several key vacancies, such as a system test engineer, at the DAAC. The system test engineer should have been in place before the hardware installation began. Filling this and other positions will help ensure that the DAAC is ready for the EOS data streams.

The NSIDC DAAC's FY 1998 budget is $4.1 million, about 10% of which is ECS-supplied hardware, software, and personnel (Table 8.1). About half of the resources are applied to development of new data sets and services, and about half to routine processing and operations. The panel views this as a healthy balance, particularly since science is moving toward multidisciplinary global activities that

require the production of a broad suite of data and information products, rather than just the provision of data sets. It also allows the scientists who use the center to do science without having to do the data processing and analysis first.

In the panel's view, the budget and number of staff (33 FTEs) are reasonable, given the scope of responsibilities of the NSIDC DAAC. The cost of the DAAC and the split between salary and operations are in proportion compared to some other international data centers of similar size that deliver similar services. Lever-aging off the activities of the NSIDC and WDC for Glaciology is also highly cost-effective, but the intermingling of these activities makes it difficult to determine the cost-effectiveness of the DAAC component by itself. In general, the DAAC would benefit from developing quantitative measures to show its cost-effectiveness, particularly given community concerns about the apparent high cost of the DAACs.

In the event of an ECS failure, the NSIDC DAAC has developed a contingency plan for MODIS that will cost a few million dollars a year for seven years. The MODIS instrument team has also developed a contingency plan, although details on the plan and its cost were not available to the panel. A final decision on which plan would be adopted had not yet been made. In either case, the DAAC risks failing its user community if sufficient funds are not available to produce the snow and ice products, or if implementing the contingency plans leads to a reduction in the number of geographic areas that can be processed—cryospheric science requires global coverage.

Future broad directions for the DAAC include delivering near-real-time data, creating a relationship with the modeling and climate assessment communities, and developing MODIS products. The first two were not discussed in detail at the site visit, and the panel assumes the DAAC has a detailed strategy for implement-ing them. The third relies, in part, on the technological transition from Version 0 to the ECS. The DAAC was concerned about setting priorities and managing competing demands for space and resources during this transition. It was not apparent in the review that realistic tests dealing with the anticipated large volumes of data and information, and with a possible large increase in the number of simultaneous user requests, have been attempted or that plans for such tests had been devised. The performance requirements developed by ESDIS, which are vague and will be difficult to verify, should be made more explicit. In addition, the panel suggests that as part of site acceptance testing, the DAAC run stress test procedures for average and peak usage to identify system bottlenecks.

The director of CIRES, Susan Avery, recognizes that having the NSIDC-WDC-DAAC complex embedded in the university, yields benefits to both the university and the data center complex. In general, the university does not go out of its way to nurture the complex, but it does provide a reasonable level of infrastructure, including space, telephone and networking services, and contract administration. It also maintains the Internet connection to the DAAC. The University of Colorado's commitment to the DAAC will end when external funding ends.

The University of Colorado has provided university affiliation for DAAC technical personnel, and the administrative placement of the DAAC within CIRES has worked well. Although the intermeshing of the DAAC with its host institute creates complexities in financial and time budgeting, it leads to a high level of flexibility. The university environment and benefits are attractive to the staff, and the DAAC benefits in many ways from, and makes use of, its position within the university. In the panel's opinion, a tremendous strength of the NSIDC DAAC is the leadership of a faculty scientist who has a voice in university affairs. Consequently, the DAAC gains visibility and clout within the university.

The National Snow and Ice Data Center is administered as a component of NOAA's National Geophysical Data Center complex of data centers and is located in a building adjacent to NGDC. Although NOAA does not have responsibility for the NSIDC DAAC, the proximity and administrative links between the centers present an opportunity for the DAAC and the Earth Science Enterprise. NGDC and the two other NOAA data centers (the National Oceanographic Data Center and the National Climatic Data Center) are the major archives of environmental in situ data in the United States. These data have enormous potential value to the EOS program, since they could and should serve to calibrate and validate satellite-derived data. NSIDC data will be used for ASTER validation purposes, but there is an opportunity for NGDC to use its links to NSIDC to play a larger positive role in the EOS mission. For that reason, the panel was disappointed that the connection between the NSIDC and the NGDC seemed pro forma.

The DAAC has had mixed results from its interactions with ESDIS. For example, ESDIS helped broker an agreement between the DAACs and the ECS contractor that will enable the DAAC to obtain the ECS source code before launch. On the other hand, it took several years for the DAAC to convince ESDIS of its need for polar grids. With the departure of Gregory Hunolt at ESDIS and Dixon Butler at NASA Headquarters, the DAAC is concerned that the idea of an integrated data system for science might be lost.

The NSIDC DAAC interacts regularly with three other DAACs. It shares MODIS data product interdependencies with the GSFC and EDC DAACs, and it used to share the same User Working Group with the ASF DAAC, which also deals with polar region data. Most of the collaboration exists at the working level because the DAACs need to solve problems in common. At the management level, however, the relationship with the other DAACs is weak. Weaver told the panel that the DAAC managers worked together only because they had a common funding source and a strong manager at ESDIS (Hunolt).

In the past, the DAAC communicated with the ECS contractor through ESDIS, an arrangement that led to frustrations on both sides. The DAAC is now communicating directly with the ECS contractor through its ECS liaison. The ECS science liaison came from the Raytheon ECS facility in Landover, Maryland, and has been able to open the lines of communication from the DAAC to the ECS contractor, although communication from Landover back to the DAAC is only beginning to be effective. Two-way communication will presumably help solve some of the frustrations that the DAAC has had with the ECS contractor, such as receiving incomplete and inconsistent documentation. It may also help reduce conflicts arising from the DAAC's desire to customize the ECS and the ECS contractor's desire to maintain a standard ECS configuration at all sites.

The DAAC is concerned about its limited ability to influence the interdisciplinary science teams and instrument teams. The DAAC staff believes it is essential to be involved with field programs if they are to do the data management well, but they have found that it is difficult to convince the scientists. Although six to eight DAAC staff are designated as mission data coordinators to the EOS instru-

ment and interdisciplinary science teams and attend team meetings, the DAAC feels it does not have much influence over the data management plans.

The NSIDC DAAC contributes well to the strategic goals of the Earth Science Enterprise by facilitating research that will lead to fundamental contributions to cryospheric and polar science. Two important factors have helped the DAAC implement its mission: (1) it has a strong vision of serving its science community; (2) it is collocated with two other major cryospheric-polar data centers, the NSIDC and the WDC for Glaciology. With regard to the first, the DAAC has an excellent relationship with its cryospheric and polar science user communities. Indeed, the DAAC is embedded in the science community, an accomplishment that few data centers are able to achieve. Its success in this regard comes in part from understanding who its users are and what data they need. A better understanding of how the data are used, however, would further improve the DAAC's service to its customers. In addition, joint activities with the ASF DAAC would help the NSIDC DAAC develop closer ties with scientists who use synthetic aperture radar data in the polar regions.

The second factor for the DAAC's success, the collocation and intermingling of DAAC operations with the NSIDC and the WDC for Glaciology, not only leverages NASA's investment but also has practical benefits for the DAAC. The DAAC can take advantage of lessons learned by the older data centers, and it has access to a wide variety of ancillary data and in-house scientific expertise. Such in-house resources, however, come with the risk of becoming a closed shop, and the DAAC and its sister centers would benefit from a strong visiting scientist program.

The DAAC's relationship with its host institution, CIRES, is also good, in part because of the strong involvement of a faculty member, Roger Barry, in the DAAC. The DAAC's relationship with NOAA's National Geophysical Data Center, on the other hand, seems pro forma and the potential for a stronger, beneficial relationship has yet to be realized. A better relationship with NOAA could help smooth the transition for the long-term archive of NSIDC DAAC data.

One of the greatest challenges facing the DAAC is accommodating limitations in the ECS that will make it difficult for the DAAC to satisfy the needs of the polar science community. For example, the latest version of the ECS does not provide subsetting capabilities or polar grids, although the DAAC and the ECS contractor are taking steps to remedy these deficiencies. Although the DAAC appears to know what needs to be done to be ready for launch, a better near-term plan and schedule to transition from Version 0 to the ECS would help ensure the DAAC's readiness for the AM-1 platform.