Future of the National Weather Service Cooperative Observer Network (1998)

In this chapter, the panel outlines a realistic approach to improving the Coop Network to meet current and future needs. Conclusions and recommendations are based on analysis of data and evidence in this report. The heart of this chapter is a detailed blueprint for a sustainable Coop Network. The panel envisions that the Coop Network will continue to rely fundamentally on volunteer observers and will incorporate strategic upgrades that are feasible with current technology. This upgraded and strengthened network will be a component of a larger national weather observing system that will evolve to meet the nation's needs. The growing demand for accurate and timely cooperative data will probably determine the requirements for the system. The Coop Network exists in the context of other national, regional, and local networks that serve a variety of functions. In some cases, cooperative observing sites are part of those other networks, and vice versa.

To determine the overall design of a surface observing system like the Coop Network, the following questions should be considered: How many sites will be required? What other networks, if any, should be incorporated? What variables should be measured? Which components should be automated? How often should data be sampled? How rapidly should data be made available to users? What tradeoffs of cost against quality and performance should be made? On what schedule should the system design or redesign be implemented? Several of these questions are addressed in this chapter. However, a thorough, detailed system design is a complex undertaking that must take into account the needs of a wide range of users, technical and budgetary trade-offs, and many other factors, and is beyond the scope of this report.

The rapidly growing commercial demand for coop data (especially by resource managers, attorneys, insurance com-parties, and consultants), the growing number of individual users reported by NCDC, and greater use of the coop data for climatological research have a number of direct implications. Perhaps the most pressing need of the greatest number of users is for faster access to data. In response to this demand, limited, but rapidly increasing, amounts of current coop data are now distributed on the Internet through NCDC, the RCCs, the National Centers for Environmental Prediction, and WFOs.

With the increasing demand for weather and climate information and the rapid development of new technologies, mechanisms should be established to provide feedback between the users and the producers of coop data. NCDC provides and receives feedback through regular contacts with clients and in-house climate researchers. The NWS has no feedback mechanism for climate data.

Conclusion. In response to the changing nature of users and applications, the NWS needs more interaction with users of the data and derived products. Links between NWS and users should be formalized organizationally to provide NWS and users, as well as coop observers, with consistent, orderly feedback.

Before going into the subject of changing the Coop Network, the panel would like to stress the importance of carefully managing the introduction of new instruments and procedures. Consistency of measurements has been a critical—perhaps even a unique—attribute of the Coop Network, one that has made it indispensable for assessing long-term climate changes and determining the frequency distribution of climate elements. There are several reasons for that consistency:

• Instrument types are changed only rarely and gradually.
• Many stations have been making observations for long periods of time (20 to 50 years) under constant conditions (i.e., the same site location and equipment).

• Standards for exposure of the instruments have remained constant over the life of the network.

It is vital that this consistency be preserved. For many purposes, consistency is as important as accuracy. For example, even data that are consistently biased because of the poor siting of an instrument shelter can be used to reconstruct past climate fluctuations if the biases are known. Potential biases in the climate record should be evaluated and accommodated to ensure that changes in equipment and/or procedures do not compromise consistency.

Conclusion. Procedures for assessing and accounting for biases introduced by changes in station location, instrumentation, and time of observation (for daily data) are essential. Appendix E lists a number of recommendations made by the panel's parent committee in an earlier report (NRC, 1992) regarding the appropriate procedures for implementing changes in the Cooperative Observer Program. The recommendations are still valid and should be followed.

At the same time, flexibility for expansion and/or modification of the system must be built in. Although volunteer observers will continue to be a necessary and central element of the system, in the network of the future fewer observing stations are likely to remain in one location for many years.

Conclusion. The future Coop Network will have to accommodate a mix of stations with varying levels of automation and sensors contributing observations at different times, as well as stations from other networks.

Recommendation. The National Oceanic and Atmospheric Administration, in cooperation with other agencies, should conduct an analysis of requirements for surface observations, with periodic follow-ups to develop requirements and specifications for a strong and viable surface observing system. The goal should be to develop and implement, over time, a comprehensive system planning architecture that ensures the effectiveness of the Cooperative Observer Network as part of a composite national surface observing system. This system architecture should be fully integrated with the other components of the overall National Weather System.

The present Coop Network consists of more than 11,800 observing sites, of which about 8,750 are "published stations" whose reports appear regularly in the NCDC summary of the day, hourly precipitation data, and other reports. But how many coop stations will be needed in the future? A complete answer to this question is beyond the scope of this report, but the panel has made several observations. First, the number of current stations is close to, but still below, the NWS goal (consistent with World Meteorological Organization standards) of one station per 625 square miles of area. The Coop Network density is based on a design formulated in an NWS document that states,

Second, the density of observations depends very much on the purposes for which the observations will be used. Different users of cooperative data have different needs. NWS officials told the panel that a density of about one site per county (or around 3,300 sites) would be sufficient to support county forecast and warning programs, provided the sites were largely automated and that the NWS had access to the data on a 24-hour per day basis. The USDA estimates that at least 10,000 stations will be needed in agricultural areas. The U.S. Army Corps of Engineers estimates that 1,100 cooperative observer sites located in critical watersheds, in addition to their automated river gauges, could meet their needs for flood control. Other agencies also have specific requirements. Because a single station could meet the needs of more than one agency, there would be considerable overlap. Climate description/climate reference and climate change data would require a very different density and distribution of stations.

Conclusion. Determining an appropriate size for the Coop Network will involve determining the minimum number of stations that would meet all anticipated needs of major long-term users, taking into account adequate station density and appropriate distribution. This analysis should be part of the comprehensive observing system recommended earlier. (Appendix F presents some general guidelines, developed by NCDC, for this analysis.)

Recommendation. As a first approximation, the panel recommends that the network support the following needs:

• climate change (requires 1,200 suitable observing sites, i.e., the number of stations in the Historical Climate Network)
• operational weather support (requires approximately 3,300 sites, i.e., one station per county)

• climate reference and description (requires at least 5,000 sites, i.e., the total number of A stations plus sites in the Reference Climate Network, part of a global network of stations selected as the ''best'' climate stations)
• agricultural weather, hydrology, and other applications (requires 10,000 sites, i.e., the number of stations, including B stations, required by USDA)

Because many stations will meet two or more of these needs, determining the required number of stations will entail a detailed analysis incorporating input from all categories of users.

Data from approximately 3,000 of the 11,800 stations in the current Coop Network are not included in NCDC publications, although the raw data forms are archived. Although the data from some of these stations may be used periodically by NWS forecast offices and other customers, the panel was informed by NCDC that the main reason these stations are unpublished is that their records are incomplete or of questionable quality.

Conclusion. Given the shortage of staff for managing the Coop Network, the excess of paper forms clogging the system, the stringency of the program's budget, and the need to improve overall efficiency, it is difficult to justify maintaining a large number of unpublished stations.

Recommendation. The National Weather Service, in coordination with the National Climatic Data Center, should evaluate the roughly 3,000 unpublished stations in the Cooperative Observer Network and determine, on a case-by-case basis, if they should be retained. If data from a given location are necessary, either the cooperative observer should be assisted to meet the network standards or another station should be established nearby.

A key goal of modernizing the Coop Network is to standardize observation times. Daily records of maximum and minimum temperature, precipitation, snowfall, snow on the ground, and any other available observations should, if at all possible, be collected at midnight (where instruments are automated) or in the early morning (where manual observations are made).

Conclusion. Standardizing observation times will facilitate the evaluation and interpretation of data in the short-term and the long-term. Whatever the observation time, it is critical that observers maintain a consistent time of observation and, if a change is absolutely necessary, that they inform the NWS official(s) responsible for managing their station.

Recommendation. Automated and manual observations of temperature and precipitation should be recorded and reported at standardized times.

In addition to daily extremes and totals, frequent readings of accumulated precipitation and temperature should be gathered from all continuous sensors and retained in an on-site data logger. Hourly precipitation data are planned to help adjust radar estimates of precipitation amounts used to validate NEXRAD estimates, for example, and more frequent data are used to calibrate radar. If two-way communication with the station logger is possible, the station should be equipped to gather data from sensors at different rates (more frequent precipitation data, for instance, when flash floods are imminent).

To maintain an unbroken record of observations, adequate battery backup should be provided for automated sensors, and precipitation gauges should be able to measure frozen precipitation. A means of taking manual observations and entering them into a data logger should be standard in case the instrument-logger interface fails. Furthermore, to allow for interruptions in communication, sufficient memory must be built into the data logger so that at least a month of data (a minimum requirement for the climate record) can be retained.

Recommendation. Automated instruments should be equipped to gather and transmit observations to a data logger frequently, flexibly, and without interruption. Loggers should be able to accommodate manual inputs and store at least a month of data.

Ideally, all stations in the modernized Coop Network will be equipped to transmit daily observations electronically to WFOs (ROSA is an example of a similar system for transmitting manual observations). The system would include on-the-spot quality control of the data and would provide feedback to the observer if a data entry is suspect. It would also allow observers to transmit special reports at any time of the day and, perhaps, ask observers for reports as the need arises.

Recommendation. Data loggers for all automated instruments should be interfaced with communications equipment that can transfer observations to weather forecast offices at prescribed times (i.e., hourly, every six hours, or daily). Weather forecast offices should also be able to interrogate data loggers on demand.

Although stability will always be a core strength of the Coop Network, as phased upgrades in technology are made

a mix of different sensor types will coexist. Hardware standards will be necessary to avoid low-quality data and high failure rates. Reasonable cost/benefit trade-offs can be made with commercially available hardware. Once a threshold of accuracy has been reached, it is often more important to have more samples rather than fewer higher-quality samples. Maintenance standards are necessary to ensure that calibration and accuracy requirements are met.

Conclusion. Achievable standards for sensor performance, maintenance, and calibration must be established.

Recommendation. The National Oceanic and Atmospheric Administration, together with other user agencies, should develop standards for sensor performance, maintenance, and calibration based on reasonable trade-offs between accuracy and cost. Data from instruments that meet technical standards should qualify as "official."

Up to now, various elements of metadata (site information) have not been available to most users of coop data, or even to most NWS staff. This shortcoming has introduced an element of uncertainty into the interpretation and analysis of long-term climate data. In addition, management of the Coop Program has been hampered by a lack of tools for effective oversight of the network. As a result, the enforcement of standards has been limited.

Conclusion. New tools, such as the Internet, GPS, and digital cameras, promise to improve the collection, storage, and dissemination of metadata and strengthen NWS management and oversight of the Coop Network. Stronger management will be particularly important if stations from mesonets and other networks are used to augment Coop Network stations.

Recommendation. The modernized Cooperative Observer Network should adopt the oversight practices made possible by new information technology so that all site information is available in on-line computer files. Each site should be located with global positioning system technology, and digital site photos should be placed in on-line files. Siting standards should be reviewed, updated, and applied consistently.

Human observers will continue to play an important role in the collection of data at most cooperative stations. Even at stations where much of the instrumentation is automated, human observers are needed to monitor and maintain the equipment on a daily to monthly basis and to provide backup observations when necessary. Observers also record rainfall, snowfall, and snow on the ground, as well as weather events, such as hail, thunder, freezing rain, sleet, and high winds.

Conclusion. Automation should be added when and where it allows the NWS to reduce the burden on observers, reduce errors, control the time of observations, and gather more data (such as hourly precipitation) at more sites. Automation will be successful only if it does not increase the observers' burden and if the communication interfaces are very simple.

Recommendation. When automation at any level is introduced at a station, the system and procedures should be thoroughly explained to the observer. The observer should be reassured that his or her role will not be made more difficult or less important.

An important factor in ensuring the consistency of observations, and thus the high-quality of data, is proper training. Modernization, including automation, will provide an opportunity to bring new knowledge and skills to coop observers and to review their previous training.

Conclusion. Personal, hands-on training is an effective way for observers to learn; however, training videos can also be used between visits or if visits are not possible. In addition, training and program updates on the Internet will be increasingly feasible as a larger proportion of the observer population acquires the capability to go on-line. However, experience has shown that there is no substitute for personal visits to stations, twice a year if possible.

Automation will probably require some specialized skills on the part of NWS personnel, as well as additional maintenance of equipment.

Conclusion. NWS's training of Coop Program personnel will have to include new skill requirements, such as first-level maintenance for data loggers or automated sensors.

Recommendation. Observers should receive on-site, personal training from qualified National Weather Service staff at least once a year. In addition, training videos should be produced that can be played on a standard VCR. The National Weather Service should begin planning to offer observer training on the Internet, especially supplemental training and updates. The same mechanisms, in addition to formal training, can be used to train National Weather Service personnel in the new skills required by the automation of cooperative stations.

Presently, there are a number of excellent mesonets at the local, state, and regional levels that gather high-quality climatic data. In some cases the density of a mesonet exceeds that of the Coop Network in the area, and mesonet stations are often located where there are gaps in the Coop Network.

Conclusion. It is reasonable to use mesonet stations to supplement or augment the current Coop Network, as long as these stations measure the proper weather variables, meet or exceed equipment and exposure requirements, and agree to participate in the Coop Program. Mesonet managers would have to agree to station maintenance, data formats, and instrumentation acceptable to the Cooperative Observer Program for fixed periods of time. In turn, NOAA could provide support for the quality control, archiving, and dissemination of mesonet data. Mesonets have not existed long enough to prove the long-term reliability of their data.

Recommendation. Mesonet stations that meet or exceed equipment and exposure requirements should be considered as supplements to, but not replacements for, the Cooperative Observer Network stations. The National Oceanic and Atmospheric Administration should establish a mechanism for evaluating the performance and set instrumentation and data standards for mesonet stations. The National Oceanic and Atmospheric Administration should establish cooperative agreements with states and other mesonet operating authorities. Mesonet operators who wish to associate their networks with the Coop Program should be required to commit to maintaining stations, data formats, and instrumentation that meet the standards of the Cooperative Observer Program for a fixed period of time. In return, the National Oceanic and Atmospheric Administration should provide quality control, archiving, and dissemination of selected data from mesonet stations.

Systematic quality control, from data collection to archiving and dissemination, is very important for an effective data collection and distribution system. Quality control should be based on a solid understanding of the sources of error and the different types of errors. A detailed data flow analysis can identify potential sources of error at various points in the data collection and transmission process. Quality control is critical in a network as large and complex as the Coop Network. For automated observations transmitted in real-time, well tested quality control procedures that have already been developed for existing networks can be readily adapted to the Coop Network. These largely automated procedures can alert network technicians to potential problems.

NWS and NCDC, RCCs, and state climatologists have accumulated a wealth of experience about the types of errors made by individuals as they observe, record, and transmit observations of daily temperature, precipitation, and snowfall on paper forms and via telephone. Data entry via a computer or touch pad creates additional potential problems but also offers potential solutions to those problems. As cooperative observers increasingly use automated data entry, immediate checking for errors becomes possible; creative graphical-visual quality control tools are available.

Conclusion. Testing and evaluation of new procedures with selected cooperative observers will be very useful in developing a single-step, friendly procedure for data entry and transmission. It is important that this procedure be as simple, efficient, and satisfying as possible in light of the time constraints on volunteer observers. Problems with instruments or observers should elicit prompt response from the local WFO.

Conclusion. The most effective way to ensure the quality of manual observations of selected climatic elements, such as precipitation and snowfall, may be through initial training and continuing education provided by the personnel at each WFO. The investment of a small amount of time initially could save a great deal of time later. With well trained observers and effective on-site quality control, NCDC should be able to substantially reduce the amount of editing. An added benefit would be a faster turnaround time from the time data are received to the time they are suitable for dissemination.

Recommendation. The modernized Cooperative Observer Network should identify problems as early as possible in the data collection process and provide prompt feedback to both observers and network technicians in order to improve the overall quality of data and reduce the costs of quality control and data turnaround time at National Climatic Data Center.

The demand for rapid and timely weather and climate data is growing. Ideally, users should be able to obtain the data from a single source. Ensuring the consistency and reliability of data means that the quality control, archiving, and first point of dissemination of validated data should be done by a single organization. The NCDC, by virtue of its facilities, experience, and expertise, is the organization best suited to manage these functions. However, the RCCs and state climatologists are well suited to assist with the dissemination of climate data to the public.

Conclusion. All cooperative data should be routed through NCDC for inspection, quality control, database development, the calculation of indices, and the production of publications. NCDC should also continue to develop the distribution of data via the Internet to all interested parties. Some means of cost recovery for Internet requests would help to defray operating costs at NCDC that are currently only partly covered by revenue from mail and fax requests.

Recommendation. The National Climatic Data Center should continue to be the focal point for archiving and disseminating cooperative data and should work with regional climate centers and state climatologists to disseminate data to all interested parties, making databases available in a timely manner. The National Environmental Satellite, Data,

and Information Service should make every effort to recover its costs for processing, copying, and providing data over the Internet.

On-site observations and data communications to the local NWS forecast office presently require manual techniques, which involve reading one or more sensors once a day, recording the data on a paper form, and mailing the form to the NWS at the end of the month. A subset of Coop Network stations have automated rain gauges and can transfer daily observations digitally via specialized devices connected to touch-tone phones.

One obvious goal of upgrading the field sites is to automate both the data collection and transmission as much as practical. However, the upgraded network will not be entirely automated; indeed, it will still be largely manual in the sense that the degree of automation at stations will vary, and some manual observations will continue to be made at most stations. In effect, the panel envisions a multilevel network that is upgraded in accordance with three main priorities:

• maintaining the size and density of the network to satisfy all major data needs
• ensuring that the quality of data remains high
• making at least a large subset of the cooperative data available faster—preferably on a near-real-time basis—while continuing to archive all data for long-term climatological purposes

The panel reiterates that maintaining the integrity of the climatological database is imperative. Manual stations provide valuable information at low-cost and at a spatial density that has been shown to be necessary for a myriad of climatic applications.

The automation of equipment at cooperative stations could be a three-stage process. The first stage would be the automation of data transmission from cooperative observer sites to NWS forecast offices. The second stage would be automation of the data ingest process for appropriate sensors. The third stage would be a cost-effective increase in the number of sensors at given cooperative sites. The following brief discussion of these three stages is generally consistent with the NOAA Project Development Plan for Modernization of the NWS Cooperative Observer Network (NOAA, 1993).

Automated data transmission from cooperative sites to NWS forecast offices, even with manual input of the data by cooperative observers, would have several benefits. First, it would significantly reduce the time between data collection and dissemination to a wide variety of users, including the NWS, on both a daily and a monthly basis. In addition, it would permit on-site storage of data for later retrieval in the event of a communication failure. Finally, it would permit faster and better quality control, both on-site and at NWS forecast offices.

The data communications equipment at each cooperative site must have the following capabilities:

• data storage for a minimum of one month (climate record requirement) in digital format on a transferable medium (e.g., diskette)
• two-way, unattended communication with the local NWS forecast office for data transfer and system maintenance
• accommodation for the observer's comments and observations
• local (on-site) data access and display (graphical displays where feasible), for feedback as well as for the observer's personal benefit
• recognition of obvious input errors
• accurate timekeeping
• easy expansion as requirements change
• a fail-safe mechanism (batteries, solar power, etc.) in case of a power outage

These specifications can be met with current technology and at modest cost, using, for example, a low-end personal computer with a modem attached to the observer's existing telephone line. With an accurate clock, data upload to the NWS forecast office could be initiated at a preset time each day from the cooperative site. The on-site clock could be reset, as needed, as part of the daily communication session with the NWS forecast office. If bandwidth permits, the NWS might also download, as a perquisite for the observer, the latest forecast and warnings for the observer's location. As computer and communications technology improves, these tasks will become easier and more affordable.

The data communications equipment might not be a single network-wide system but could evolve through various stages of automation. However, if many different systems are in place simultaneously across the Coop Network, maintenance and replacement could be complex and expensive.

Recommendation. Automating data communications between cooperative sites and local National Weather Service forecast offices should be the first step in automating the cooperative observer sites. The goal should be to make reporting data on at least a daily basis possible at all stations, even if data are still input manually.

The second stage, automation of the data ingest process, would have the following advantages:

• elimination of manual input errors
• more frequent observations
• midnight-to-midnight data summaries
• more detailed data statistics (such as hourly means and variations)
• collection of data from a wider variety of sensors

This second stage obviously requires that the site have at least one electronic sensor. Temperature measurement is the obvious starting point. An electronic sensor would permit hourly readings, as well as daily minimum and maximum temperatures (midnight-to-midnight readings or observation time-to-observation time) to be stored for later transmission to an NWS office.

Precipitation gauges would be a logical second sensor. The automated precipitation gauges currently in use, however, are moderately expensive and often require high maintenance. New technologies being developed should be investigated before more Fischer-Porter sites are added.

Recommendation. The National Weather Service should continue to pursue alternatives to Fischer-Porter gauges for providing automated hourly measurements of precipitation. To maintain temporal consistency in the data, manual observations of precipitation should be continued in parallel wherever automated precipitation gauges are used.

Digital data ingest would require either the addition of a suitable circuit board to an on-site personal computer or a specialized data logger. This equipment is widely available with current technology at modest cost. The enhanced computer or data logger must have the following capabilities:

• continuous data collection from a variety of sensors
• operation on battery backup power (for a minimum of 10 days to cover worst-case electrical outages caused by natural disasters)

In most cases, coop stations are located at sites where individuals are available to make daily visual inspections of the equipment, measure snowfall and snow on the ground, and observe weather phenomena (e.g., wind damage, thunder, hail). Ideally, the storage device (data logger) would be interfaced with a keyboard so manual entry could be made of all observations. The user interface should be extremely user friendly.

Recommendation. Wherever feasible, cooperative stations should be provided with personal computers or data loggers for automated ingest of data from one or more electronic sensors. These personal computers or data loggers must be able to operate on battery backup power for at least 10 days and should have user-friendly interfaces for the manual ingest of data. Computers or loggers should also have an on-site error feedback mechanism and quality control during the manual input of data.

Stations with both automated ingest of data and automated communication will provide a wealth of near-real-time information, which will be critical to improving weather and hydrological warnings and forecasts and will be extremely useful for emergency managers, transportation officials, and the general public. More detailed statistics and more frequent observations would also improve on-site monitoring of sensor performance and quality control, which would also improve scheduling of field maintenance.

Once a cooperative station has been automated, additional sensors can be installed to meet the needs of federal or state agencies or other users of Coop Network data. Perhaps the easiest one to add would be a sensor to measure relative humidity, which is often measured with a capacitance transducer packaged with a temperature sensor in a single instrument. Another easily added sensor is a pyranometer to measure total incoming solar radiation. Fairly accurate pyranometers are available and require relatively low maintenance. Other sensors would entail a noticeable increase in cost and/or maintenance. High-quality anemometers, for example, are moderately expensive, require good exposure for representative measurements, and, in most cases, must be mounted on a tower. Accurate pressure measurements also require moderately expensive instruments. Soil temperature and moisture sensors are inexpensive and could provide very useful data. However, the soil parameters can vary considerably, so a representative, but still accessible, measurement location must be carefully selected.

Because there are 11,800 cooperative observer stations in the United States, automation will necessarily be a gradual process. Some observers may be offended by changes in their role, and training them will take time. Any increase in the complexity of the cooperative observer's role is likely to lead to problems. Also, the continuity of data is a vital feature of the Coop Network, and rapid changes in equipment would inevitably disrupt that continuity. For these reasons, the panel anticipates that automation will be phased in nationwide and that there will be a mix of systems across stations and over time. The panel agrees with the basic NWS proposed strategy for modernizing the Coop Network and does not recommend the full automation of sensors. Substantial participation by volunteer manual observers will be necessary for the foreseeable future to provide backup and maintenance and accurate precipitation measurements for weather and climate applications, to keep costs low, and to report weather phenomena that are not detectable or measurable by automated systems.

Conclusion. The Coop Network should not be completely automated but should continue to have a large manual

component. It would be prudent to build on what works well and repair or augment what does not work very well, rather than to start over "from scratch." Maintaining the continuity of data is another factor that favors maintaining as much of the existing network as is feasible.

Conclusion. The panel wishes to specify capabilities rather than specific hardware. The architecture/planning recommended earlier would be invaluable for upgrading coop stations in the next five to ten years.

The transition to electronic sensors must be made carefully so as not to interrupt the long-term climate record. Long-term continuity is a unique and absolutely indispensable feature of the Coop Network and the data it provides, and it must be protected. As new instruments are introduced, studies to determine adjustment factors to account for differences between old and new sensors, gauges, and shelters should be made. The goal is to preserve the temporal continuity of station databases and make the change of equipment as seamless as possible in terms of the official climate record.

Conclusion. Every effort should be made to ensure that the transition to new instruments does not cause a significant discontinuity in the climate record.

Recommendation. New sensors should be introduced gradually across the Cooperative Observer Network. Changes in instrumentation should be tested at selected sites by thorough comparisons with the old instruments for at least a year.

The comparisons must be done under a wide variety of climatic conditions to account for regional and seasonal differences. Studies of previous instrument changes in the Cooperative Program and at first-order NWS stations have demonstrated the necessity and value of these comparisons. In one study (Quayle et al., 1991), a comparison of electronic MMTS observations with readings taken using liquid-in-glass thermometers situated in cotton region shelters showed that the MMTS daily maximum readings were lower and the minimums higher. Regional differences in this relationship were observed, probably related to snow cover, the intensity of solar radiation, and wind speed, among other variables. Studies of the continuity of climate data conducted for the NWS when the new ASOS was introduced showed that changes in instrument location can lead to temperature differences of as much as 1°F (Guttman and Baker, 1996; McKee et al., 1996; and Schrumph and McKee, 1996).

A predetermined number of cooperative observers should be recruited to participate in comparative studies, which should last for at least a year (ideally two years, to account for inter-seasonal variations). Once the new equipment has been validated for the prescribed interval, the old equipment should be removed to avoid confusion over which instrument is being used to make official measurements at the site. In the event that a full suite of climatic conditions has not been observed, it would be useful to maintain comparative observations for a longer period of time at a subset of stations. This would also test whether the relationships between old and new equipment change with time.

In addition to comparisons of instruments, studies should be done to determine the impact of changes in observation times on the climate record at individual stations, and adjustment factors to account for these differences should be developed. These studies should be patterned after previous studies (Karl et al., 1986) in which adjustments were determined for observations taken in the evening or the morning to a standard midnight observation time.

In this section, the panel suggests some modifications of the management structure and practices to ensure that improvements in the Coop Network lead to real improvements in services.

The preeminent management issue for the future of the Coop Network is the question of ownership and stewardship—operation, management, and policy direction. Various other federal agencies that rely on cooperative data have been frustrated with the low priority, slowness, and inconsistent quality of cooperative data under NWS management. Partly for this reason, the Tennessee Valley Authority, the U.S. Army Corps of Engineers, and the Bureau of Reclamation no longer rely as heavily on Coop Network data as they once did. The USDA even made a tentative offer before the panel to take over operation of the network. However, all of the user agencies would prefer to have the NWS continue ownership, if improvements can be made. NWS officials also expressed a desire to continue operating the Coop Network.

Conclusion. The NWS has the infrastructure and experience to continue operating the Cooperative Observer Network successfully if the changes recommended in this report are made.

Recommendation. The National Weather Service should improve its management of the Cooperative Observer Network.

Given the substantial, long-standing interest of many federal agencies in the health of the Coop Network and in the

use of its data, and considering the difficulties that the NWS has had providing adequate operational and funding support for the network, it is reasonable for other agencies to participate in the policy direction and support of the network. There are several applicable models. One is the National Atmospheric Deposition program (NADP), which was established in 1977 to address the problem of atmospheric deposition (e.g., acid rain) on crops, forests, surface waters, and other natural resources. The NADP was later merged with a federal acid-rain monitoring and research program and assumed responsibility for a 200-site network of monitoring stations. Seven federal agencies (NOAA, USDA, the Bureau of Land Management, the Environmental Protection Agency, the U.S. Geological Survey, the U.S. Forest Service, and the National Park Service) support this program and participate in policy and technical guidance through various management committees. Support is provided by these agencies, other federal and state agencies, universities, public utilities, and industry. The amount of support from each entity is determined on the basis of need. This arrangement for joint management and support appears to work well.

A similar interagency mechanism more relevant to the Coop Network is the Federal Committee for Meteorological Services and Supporting Research (FCMSSR), which is directed through the Office of the Federal Coordinator for Meteorology (OFCM). The FCMSSR was established in 1964 "to promote coordination and cooperation among the federal agencies having weather-related activities so that the most effective and best possible weather information and user services are provided for the funds made available by the government." Fourteen federal agencies participate, including the U.S. Department of Commerce, the USDA, the U.S. Department of Defense, the U.S. Department of the Interior, the U.S. Department of Transportation, the Federal Emergency Management Agency, and the Environmental Protection Agency.

The OFCM carries out a number of coordinating functions, including the ones listed below:

• the documention of agency programs and activities in coordinated national plans
• the provision of structure and programs to promote the development and coordination of interagency plans and procedures for meteorological services and research
• the preparation of analyses and evaluations for use in the appropriations process
• the review of federal weather programs and requirements for meteorological research, with suggestions for revisions of current programs

At present, the OFCM does not play a substantial role in any climate observing network. However, a number of program councils and other groups are relevant to climate observing networks, including the following:

• Working Group for Climate Services
• Panel for Observing Systems
• Working Group for Meteorological Information Management
• Program Council on Automated Weather Information Systems

These groups have functioned very effectively with respect to meteorological activities in relation to aviation and in a few other focus areas that require strong interagency coordination and where the participating agencies have agreed to let the OFCM play a strong role. For example, the NEXRAD Program Council has done an excellent job coordinating the development and implementation of the radar network to meet multi-agency needs. The structure is in place within the FCMSSR for a similar focus on the Coop Network; indeed, the federal coordinator of FCMSSR has expressed a willingness to use that structure to coordinate interagency participation and support for the Coop Network.

Recommendation. The National Oceanic and Atmospheric Administration should work with other agencies to establish an interagency management council to guide and provide support (including funding) for the Cooperative Observer Network. The Office of the Federal Coordinator for Meteorology could administer the operation of this council.

The panel observed that differences in operational priorities and ineffective coordination between high-level NWS and NCDC managers in addressing budget and data deficiencies have exacerbated operational and fiscal support problems for the Coop Network and the Coop Program. Under the present organizational structure, policy guidance, long-term planning, and budgetary advocacy are inadequate. The panel attributes this sense of "rudderlessness" to the absence of appropriate management representation for the Coop Network and Coop Program on NOAA's staff.

Conclusion. The National Oceanic and Atmospheric Administration should have a management oversight function for the Cooperative Observer Network and Coop Program. The lack of integrated management of the Coop Program suggests that the program should be managed above the level of the NWS.

Recommendation. The National Oceanic and Atmospheric Administration should improve the overall management of the Cooperative Observer Program. One approach would be to establish a climate observations management office to oversee the activities of the of Cooperative Observer Program of the National Weather Service and National Climatic Data Center. This office would ensure that the Cooperative Observer Program is given a high priority by the National

Oceanic and Atmospheric Administration, the National Weather Service, and the National Environmental Satellite, Data, and Information Service. Operational management of the Cooperative Observer Network would continue to be the responsibility of the National Weather Service.

The NOAA climate observations office should work jointly with the NWS Cooperative Observer Network and NCDC management teams to perform the following functions:

• shape the current and future directions of the Coop Program, enlisting interagency support in planning, policy-making, and funding
• provide effective advocacy for the Coop Program in budgetary planning by NOAA, the NWS, and NESDIS
• work with other federal agencies and states that have cooperative observer programs and/or mesonets to develop a coordinated approach to the management and maintenance of the Coop Network and ancillary networks (including setting standards for sensor performance, maintenance, and calibration)
• collaborate with regional climate centers and state climatologists to ensure that high-quality climate data and derived products are available to users on a timely basis and that they are properly archived

To improve the management of the Coop Network, the NWS must improve the image of the network and give it a higher priority. The recent appointment of a full-time NWS Cooperative Program manager was an important step toward increasing the effectiveness of NWS's management.

Recommendation. To ensure the effectiveness of the Cooperative Observer Program manager, he or she should have direct access to National Weather Service top management and should be well connected with other federal agencies that use both real-time and historic climate data.

The Cooperative Observer Network manager should perform the following functions:

• provide the network with strong, credible leadership
• maintain and improve network site stability and data standards, with due consideration of human factors and methods to ensure that data are representative
• promote volunteerism and public appreciation of the Coop Network
• work with the NOAA climate observations office to oversee the program and provide advocacy
• ensure that specific, consistent, reliable information about the Coop Network and its value to society are readily available to policy makers and the public, as well as to NWS and NCDC managers

In general, the current WFO staffing for the Coop Program is marginal because Coop Program duties are part-time and vary in priority among WFOs. The staffing model needs to be adjusted for each WFO based on the number of cooperative stations assigned, the distances involved in managing the stations, and other operational factors. Maintaining consistent personal contact with volunteer observers is critical to the Coop Network.

Conclusion. Each WFO should have an individual on staff who is the primary point of contact for cooperative observers in the WFO's area of responsibility.

Recommendation. At each weather forecast office, staffing for the Cooperative Observer Program should focus on the data acquisition program manager as local manager, supported by an adequate number of hydrometeorological technicians to carry out its responsibilities in the forecast area. One of these responsibilities is to maintain consistent personal contact with the volunteer observers.

The meteorologist-in-charge (MIC) sets priorities at each WFO. Ultimately, work assignments, as well as attitudes about the Coop Program, derive from the attitude of the MIC. The panel is aware of some WFOs where the Coop Program operates extremely well because of the innovations, interest, and enthusiasm of the MIC.

Conclusion. The support and involvement of MICs is essential to the success of the Coop Program at WFOs.

Recommendation. The National Weather Service should hold forecast office managers accountable for the health of the portion of the Cooperative Observer Network under their purview. Performance evaluation criteria should be developed to encourage accountability.

The current Coop Network cannot be sustained at present funding levels. Reimbursables are declining (in part because of free Internet access to selected data and the distribution of coop data by state climatologists and RCCs), and support from other agencies is in jeopardy. Modernization of the Coop Network will require substantial new funds, not only for the acquisition of equipment, but also for ongoing operations and maintenance. Appropriate funding levels can be determined by the proposed interagency council after the recommended comprehensive review of requirements for surface observations.

Conclusion. Funding above the current level will be needed to finance the modernization of the Coop Network as outlined

in this report. Even with new appropriations, a mechanism for obtaining funds from other sources—including user agencies, the public, and industry—will be necessary for upgrading the current system.

The constituency for cooperative data is extremely diverse (essentially encompassing the entire U.S. population). Therefore, the support for the network that produces those data should also be diverse. However, the various elements of the national climate services structure that are in close contact with this broad constituency have not been effective in gaining public and political support for funding new or upgraded Coop Network capabilities.

The Coop Network has served the nation well for more than a century. The network and the data it provides have become a crucial resource:

• For the NWS, coop data are the indispensable foundation of the nation's surface weather observing systems, the link between past, present, and future weather.
• For NCDC, state climatologists, and RCCs, coop data are vital for climate research and are essential for meeting the climatic needs of a myriad of customers.
• For the USDA, the Coop Network has been a vital monitoring and warning system for the agricultural community and, with proper upgrading and support, it can continue to provide this much-needed service.
• For the U.S. Army Corps of Engineers, the Coop Network provides the core data for its flood control network.
• For the U.S. Department of the Interior, the network has been a part of its automated observing systems and, in the future, can provide an even larger portion of the information its bureaus and services need to manage the nation's lands and parks.
• For the Federal Emergency Management Agency, the Coop Network offers a way to gauge the relative severity of weather-related disasters and respond appropriately.
• For the Environmental Protection Agency, cooperative data can be an important element in the calculation and prediction of environmental problems and in determining how to mitigate them.
• For the U.S. Department of Transportation, cooperative data already play a role in the design and construction of new and upgraded highways and in determining when to deploy emergency services. In the future, these data will be a necessary component in the management of intelligent highway systems.
• For state governments, cooperative data are vital to the design of facilities, the enforcement of regulations, and the design of highway, water, and agricultural systems.
• For the public and large portions of the private sector, cooperative data are important for literally hundreds of applications in every area of human activity.

The Coop Network should be substantially refurbished and modernized to make it an integral component of the national weather system of the next century. Thus, a new perspective on its mission is in order.

The panel envisions a Cooperative Observer Network that is structured, managed, and equipped to provide high-quality weather and climate data rapidly and cost-effectively to the full spectrum of users who require spatially detailed information for a multitude of purposes, including not only the description and understanding of climate and climate change, but also operational meteorology, hydrology, agriculture, environmental protection, and a myriad of business, legal, economic, and personal decisions. Observations will encompass a very broad range of measurements that are technically and economically feasible, and instruments will be standardized across the network. Automated and manual observations will be combined to maintain the continuity of data for a broad spectrum of weather and climate information. The communication of data to analysts will be rapid and robust. A broad range of useful data products will be available in a variety of forms. Flexibility, compatibility, reliability, quality, and rapid accessibility will be the watchwords of the modernized Cooperative Observer Network. Coordination, cooperation, and integration will be the hallmarks of the revitalized Cooperative Observer Program.

With a modernized Cooperative Observer Program and Network that is part of a national (and perhaps international) observing system, the United States will be in a position to monitor climate with scientific precision, not only as a metric to assist in planning virtually all human enterprises, but also as a gauge for detecting trends and assisting in the construction and verification of predictions of climate changes.