Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
40 OCEAN-ATMOSPHERE OBSERVATIONS measure from buoys all the variables (u', v', w', T', Q', radiation terms) required to make direct, nonbulk-formula determination of the surface fluxes of momentum, heat, and moisture. At present, the combination of using bulk formulas and the large errors inherent in standard shipboard surface meteorological observations make the errors in flux determinations so large that prediction models tend to compute the fluxes instead of assimilating flux data or using them to validate predictions. Sets of accurate, directly calculated flux data from moored stations would be powerful constraints for future models, certainly for verification and perhaps eventually for initial- ization and assimilation. Ship differential Global Positioning System heading sensors have reached the point at which upper-ocean acoustic Doppler current profilers could be operated on VOSs without unacceptably large errors in cross-track currents due to heading errors. While it may not be necessary to operate such equipment on all VOSs, consideration should be given to outfitting those that regularly cross the major tropical near-surface currents as one more means of obtaining ocean circulation measurements for model validation. These are examples of promising technologies and approaches that will be pursued under research auspices in the coming years and that may well lead to new climatically important data streams, just as the invention of the XBT did less than three decades ago. The assessment of which methods are most useful for climate prediction purposes and most appropriate for incor- poration into operational systems is business that must be left for the future, as the technical development successes become clearer and the models tell us more about which kinds of observations are most valuable and which methods are most cost effective. Here lies another good reason why this panel can only do an incomplete evaluation of such questions, at one point in time, and thus why ongoing means of dealing with these questions are needed.
5 Additional Issues COMMUNICATIONS Global deployments of meteorological and oceanographic instruments are increasingly hampered by inadequate systems for receiving and trans- mitting data from these instruments. It is obvious that to be of any use the data stream from the observing system must reach the end users in a "timely" manner. Also, the data stream must contain "adequate" resolution of sig- nals relevant to climate. The end-user community for the observing system is already broad and will continue to grow. Thus, the definition of what is timely and adequate in terms of data distribution varies considerably de- pending on the use. For some purposes the present systems are already inadequate, and without deliberate planning and action the situation will worsen rapidly. Operational numerical weather prediction centers require rapid transmission of a relatively coarse set of observations. Climate as- sessment and longer time-scale prediction centers may require higher-reso- lution data sets (to avoid aliasing) but may tolerate some delay in data receipt. Process studies require the highest resolution possible from the observations and, in the past, could tolerate lengthy delays for receipt of information. However, the issue of timeliness is no longer obvious for process studies; given adaptive sampling in near-real-time, large process studies [e.g., the TOGA Coupled Ocean-Atmosphere Response Experiment (COARE)] have considerable requirements for rapid dissemination of high- resolution data sets. There are three elements of the data communications infrastructure that 41
42 OCEAN-ATMOSPHERE OBSERVATIONS must be addressed: (1) the in situ data acquisition/transmission system, (2) the broadband data dissemination network, and (3) the interface between these two systems. Presently, the in situ data transmission systems rely primarily on polar- orbiting satellites through Service Argos and on geostationary satellites. Argos works reasonably well (e.g., surface drifting buoys), and the equip- ment demands are not great. It is somewhat expensive in terms of data transmission and processing and can only accept fairly low data rates. Most importantly, timeliness of data transmission is a problem, especially in the tropics, due to the frequency of satellite overpasses. Finally, there are fundamental limitations on the magnitude of the data stream through Argos that will become important as the number of observing platforms increases. These latter two limitations cannot be overcome without the addition of further satellites to the system. Geostationary satellites provide a means for timely data transmission from regions that are within the useful footprint of the satellites (e.g., TOGA sea-level stations). There are severe bandwidth limitations on these satel- lites; in addition, there are various governmental restrictions on the use of particular satellites. More sophisticated data transmission equipment is re- quired to make use of geostationary satellites. The present data distribution network used by the climate community is the Global Telecommunications System (GTS). The GTS consists of a mixture of very old and new hardware and software and contains links for the high-speed transfer of large volumes of data as well as 50 baud tele- graph circuits and radio teletype broadcasts. The elements of this system are provided by many nations and involve a complex mix of civilian and military communications links. The contents of the data stream from the GTS vary depending on the node that is tapped. At a minimum, this implies that internal gateways between various elements are not operating efficiently. The GTS relies on relatively fixed formats that are specified for particular observing systems. This presents a major obstacle to the addition of a new generation of oceanic monitoring systems. It takes many years to adopt new formats for use on the GTS. Additionally, the expansion of the number of observing systems (and the associated increase in data volume) presents the real potential for slowdowns in the arrival of critical data at end-user sites. It may be that the overall architecture of the GTS provides fundamen- tal limitations on its usefulness as a climate data distribution system. The interfacing of in situ observational systems to the GTS has pre- sented a major obstacle to successful data transmission to end users. While there have been recent improvements to this interface for the TOGA TAO by Service Argos and the National Oceanic and Atmospheric Administra- tion (NOAA)/National Ocean Service, an unacceptably large loss of data still occurs. Much of this can be attributed to problems intrinsic to the
ADDITIONAL ISSUES 43 GTS. Future problems with the interface will occur with the development of new observing systems. As the data communications system is not specific to TOGA or its successor, the problems and inherent limitations of the system must be addressed by a broader community. It is reasonable to insist that the over- haul of the system incorporate specifications that anticipate large growth of the volume as well as an increase in the variety of data transmitted. We can request that the agencies that have a stake in the transmission of data from in situ systems form cooperative efforts to fund research devoted to devel- oping new technologies for data transmission and distribution. Opportunities for advantageous use of commercially developed data collection and transmission systems should also be explored. For example, a global cellular telephone system is proposed. Which of these alternative systems will afford significant improvements in terms of coverage, cost, data rates, and equipment simplicity relative to the existing situation re- mains to be seen. Agencies and program managers can assist progress in several ways. By funding experimental efforts aimed at trying to use new systems, they accelerate the acquisition of knowledge about the advantages and disadvantages of these systems. By collecting and organizing knowl- edge about the growth of global instrument deployments, they can keep the scientific community and the commercial data system architects aware of the potential in this slice of the "market" for new collection and transmis- sion systems. If and when one or more new systems reach viability, there may be ways in which agencies can broker consolidated deals for economi- cal system use by scientific programs. Similar consolidations have been done with Argos. At the very least, encouraging new systems in these ways may help 'to bring competitive pressure to bear on existing systems like Argos. PROCESS STUDIES Although the panel was created to consider operational observations, it is important to take note of the intimate relationship between focused pro- cess studies justified on research grounds and the evolution of operational observing schemes. Divorcing the two is a recipe for ineffectiveness. In the present context of climate predictions, there will always be new ideas about the relevant machinery of the ocean and atmosphere and about dis- crete experiments that could be mounted to test these ideas. TOGA COARE, which has just completed its field phase, is an excellent case in point. Its goals are to describe and understand (1) the principal processes responsible for the coupling of the ocean and atmosphere in the western Pacific warm pool system, (2) the principal atmospheric processes that organize convec- tion in the warm pool region, (3) the oceanic response to the combined
44 OCEAN-ATMOSPHERE OBSERVATIONS buoyancy and wind stress forcing in the western Pacific warm pool region, and (4) the multiple-scale interactions that extend the oceanic and atmo- spheric influence of the western Pacific warm pool system to other regions, and vice versa. These objectives and the finite-term field experiment that was executed to address them constitute a clear process study. But it is to be expected that from TOGA COARE results will eventually flow sugges- tions about sensible improvements of ongoing (operational) observations in the warm pool region and perhaps beyond. COARE and other such process experiments, far from being unrelated to or even in competition with opera- tional concerns, are the key means by which new information can be brought forward to improve the effectiveness of operational observations. TRANSITIONS TO OPERATIONAL SUPPORT In the, foregoing we used the term "operational" to refer to observations that clearly should be sustained for the long term and also in a more restric- tive sense to refer to observations that should be undertaken by an opera- tional entity of government, as opposed to a research program. In the following paragraphs we take up the latter meaning. We will discuss a particularly important set of issues that revolve around establishing proper and mutually agreeable roles for both operational agencies of government and for the research community (governmental and academic) in conducting and guiding long-term observations for climate studies. Especially for ocean observations (but not exclusively-see comments about TOGA-initiated meteorological stations above), systems and time se- ries tend to arise in the research community as exploratory measurements, as observations begun in a process study, etc. After a time it becomes clear that the observations have long-term value as a time series for climate studies; how should they be carried forward? There are often good reasons to consider having some operational entity of government "take over" the responsibility. Purely research funding (year-to-year National Science Foundation grants) may be uncertain, the research principal investigator(s) may be less than eager to assume operational responsibility with all the repetition and attendant chores of data distribution, and so on. But rigid insistence on transfer of responsibility at this point can lead to bad results and lack of data continuity. Several factors need to be studied first: 1. Responsibility. Does the target government agency have real re- sponsibility for this set of observations AS CLIMATE TIME SERIES built into its charter or set of priorities? Undertaking the maintenance of a climate time series implies concerns about accuracy, about biases that can result from changes in method, and about the need to sustain the integrity of the data set when changes of method or sampling pattern are proposed,
ADDITIONAL ISSUES 45 concerns that may be minimal with regard to data sets used for other pur- poses. This point and the following one lead directly to Conclusion 6. 2. Technical and operational readiness. The transfer of an "up and running" observing system involves much more than a change from "re- search" to "operational" funding. If the recipients of the system have not established technical understanding of the system and the operational abil- ity to deploy it, maintain it, deal with necessary infrastructures (volunteer observing ships, foreign sites, communication networks, etc.), acquire, check, and distribute the data, any transfer is premature. 3. Ongoing research involvement. Operational systems are meant to be long term and relatively stable, but they are not static. The future will bring new awareness of better sampling plans, better methods, trade-offs between data of various kinds and between data and models, and the like. Continued research with and scrutiny of the data will reveal errors, gaps, new applica- tions of the data, and ways to improve the ongoing work. The sustained involvement of the research community (academic and governmental) is essential in this context. Mechanisms must be put in place to foster and encourage this "hands-on" involvement, not merely to give it grudging ac- ceptance. Examples of WOCE and TOGA data assembly centers located at research institutions are useful paradigms for what can and should be done in this regard. End-to-end involvement of research scientists at all stages, from raw data collection through quality control to standard maps and data products and on to research papers, is the best guarantee of overall data integrity and usefulness for the long term. (See Conclusion 5.) This close relationship to the research community also builds a better- informed and more active constituency for the data system. When, as is inevitable, pressures for cutbacks or abandonment arise, this larger agency/ research community constituency will be better able to address the pres- sures and to help resist them if that is warranted. To tum this comment around, the housing of a set of observations under operational auspices is not in and of itself full insurance against future collapse of the observations. One has only to look at the present deteriorating state of the World Weather Watch network discussed above to see this. Having both research commu- nity and operational agency advocacy for a set of observations is certainly a better situation than having only one type of support. 4. Role for institutional effons. Some observing system elements may be better off remaining at research institutions for a variety of reasons. Sometimes the real expertise is there; sometimes having the institution in a leading role leads to essential nongovernment funding for the work. The California Cooperative Ocean Fisheries Investigations data series in the eastern boundary current zone of the North Pacific is the premier extant example of a long ocean time series. It has survived for over 40 years as a federal-state-university partnership involving a research entity (the Marine
46 OCEAN-ATMOSPHERE OBSERVATIONS Life Research Group of Scripps Institution of Oceanography) and as an operational one (NOAA's National Marine Fisheries Service Southwest Fishery Laboratory). While there may be good arguments to modify the arrange- ments as to which participants carry out which aspects of the work, there seems to be no reason to alter the basic pattern of shared institutional re- sponsibility for the overall program. The point is that not every observing system reaching operational maturity is necessarily better positioned com- pletely within a federal agency. Each case needs to be looked at individu- ally. The key concern should always be the long-term health of the data series as a set of observations for climate study and research. If this health is better secured in a nonfederal or partly nonfederal setting, that is where the series should be housed and run.