2
Operation and Management

This chapter presents the panel's analysis of issues and problems that affect the Coop Network and Coop Program. The first section describes the management structure, the second technical and operational issues, the third management issues, and the fourth overall systems issues.

Network Management

Organizational Roles

Two organizations, NWS and the National Environmental Satellite, Data, and Information Service (NESDIS), manage the collection and most of the processing and dissemination of data. Both are major components of NOAA. The NWS is responsible for observations—that is, for station operations, instrumentation, and documentation, as well as for observer recruitment and training, initial data collection and quality control, and the transmission of data to NESDIS. Within NESDIS, the NCDC is responsible for long-term stewardship of the data, which involves data assimilation and archiving, quality control, and the generation and dissemination of products derived from the data. NWS uses the data primarily for operational meteorology and hydrology; NCDC uses the data primarily for climatological purposes. The volume of data processed by NCDC is quite large—currently more than 142,000 cooperative data forms per year, containing more than 25 million handwritten observations.

Management Structure

Figure 2-1 shows the management structure in the NWS and NESDIS with respect to the Coop Program and coop data. Data collection from the Coop Network is managed by the NWS Office of Systems Operations, acting through the NWS regional and forecast office personnel. Until recently, NWS Coop Program functions were managed at the field office level of the NWS by 51 full-time cooperative program managers (CPMs), with technical assistance from six regional CPMs. A national Coop Program manager oversaw the entire program and established national policy and procedures for the operation of the Coop Network.

In 1995, in conjunction with the modernization and restructuring of NWS offices, the CPM positions at WSFOs were abolished, CPM staff at regional offices were reduced, and management responsibilities for the Coop Program were transferred to data acquisition program managers (DAPMs), who are staff members of the 119 new NWS forecast offices. The DAPMs are assisted by hydrometeorological technicians (HMTs) and by meteorological interns. Both DAPMs and HMTs perform their Coop Program duties on a part-time, as-available basis.

NCDC's management role has not changed much in recent years. Processing and analysis of the coop data are the responsibility of NCDC's Data Operations Branch (DOB), which is also responsible for generating and disseminating products, such as monthly data summaries. A Database Management Branch is responsible for archiving the data. The Climate Services Division provides user services.

Technical and Operational Issues

Data Collection and Transmission

Instrumentation

The range of instruments that cooperative observers use to measure temperature and precipitation was described in Chapter 1. Instruments include liquid-in-glass maximum/ minimum thermometers (housed in cotton region shelters); MMTSs; standard nonrecording rain gauges; and Belfort Fischer-Porter and Belfort Universal recording rain gauges. As these instruments age, they are becoming increasingly difficult to maintain and calibrate (see Box 2-1).

The liquid-in-glass thermometers, for example, are difficult to read and expensive to replace. The cotton region shelters that house them (wooden structures) must be painted



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 12
--> 2 Operation and Management This chapter presents the panel's analysis of issues and problems that affect the Coop Network and Coop Program. The first section describes the management structure, the second technical and operational issues, the third management issues, and the fourth overall systems issues. Network Management Organizational Roles Two organizations, NWS and the National Environmental Satellite, Data, and Information Service (NESDIS), manage the collection and most of the processing and dissemination of data. Both are major components of NOAA. The NWS is responsible for observations—that is, for station operations, instrumentation, and documentation, as well as for observer recruitment and training, initial data collection and quality control, and the transmission of data to NESDIS. Within NESDIS, the NCDC is responsible for long-term stewardship of the data, which involves data assimilation and archiving, quality control, and the generation and dissemination of products derived from the data. NWS uses the data primarily for operational meteorology and hydrology; NCDC uses the data primarily for climatological purposes. The volume of data processed by NCDC is quite large—currently more than 142,000 cooperative data forms per year, containing more than 25 million handwritten observations. Management Structure Figure 2-1 shows the management structure in the NWS and NESDIS with respect to the Coop Program and coop data. Data collection from the Coop Network is managed by the NWS Office of Systems Operations, acting through the NWS regional and forecast office personnel. Until recently, NWS Coop Program functions were managed at the field office level of the NWS by 51 full-time cooperative program managers (CPMs), with technical assistance from six regional CPMs. A national Coop Program manager oversaw the entire program and established national policy and procedures for the operation of the Coop Network. In 1995, in conjunction with the modernization and restructuring of NWS offices, the CPM positions at WSFOs were abolished, CPM staff at regional offices were reduced, and management responsibilities for the Coop Program were transferred to data acquisition program managers (DAPMs), who are staff members of the 119 new NWS forecast offices. The DAPMs are assisted by hydrometeorological technicians (HMTs) and by meteorological interns. Both DAPMs and HMTs perform their Coop Program duties on a part-time, as-available basis. NCDC's management role has not changed much in recent years. Processing and analysis of the coop data are the responsibility of NCDC's Data Operations Branch (DOB), which is also responsible for generating and disseminating products, such as monthly data summaries. A Database Management Branch is responsible for archiving the data. The Climate Services Division provides user services. Technical and Operational Issues Data Collection and Transmission Instrumentation The range of instruments that cooperative observers use to measure temperature and precipitation was described in Chapter 1. Instruments include liquid-in-glass maximum/ minimum thermometers (housed in cotton region shelters); MMTSs; standard nonrecording rain gauges; and Belfort Fischer-Porter and Belfort Universal recording rain gauges. As these instruments age, they are becoming increasingly difficult to maintain and calibrate (see Box 2-1). The liquid-in-glass thermometers, for example, are difficult to read and expensive to replace. The cotton region shelters that house them (wooden structures) must be painted

OCR for page 12
--> periodically; replacements must be built by skilled wood-workers and take about a year to procure. The replacement cost for the thermometers and shelters is about $1,000. Two types of gauges report precipitation on an hourly (or more frequent) basis and are used at about 2,500 sites in the Coop Network. Both of the devices that record precipitation on a paper tape for the Belfort Fischer-Porter rain gauge and the paper chart for the Belfort Universal gauge are obsolete and are expensive and time-consuming to maintain. The paper punch mechanism of the Fischer-Porter gauge has a history of frequent failures. The Fischer-Porter paper punch tapes and the analog charts of the Belfort Universal gauge are prone to numerous recording errors. The device that NCDC uses to read the Fischer-Porter tapes has become difficult to maintain, and replacement parts are not readily available. The NWS has proposed replacing the Fischer-Porter gauges with an automated device for the collection and transmission of data but has been unable to obtain funding. In some cases, evolving technology, along with increasing costs and decreasing availability of the old technology, has forced a transition to new instruments. This shift usually has both good and bad consequences. For example, the MMTS was introduced by the NWS in the 1980s as a replacement for the liquid-in-glass thermometer. In contrast to the cotton region shelters, the MMTS is housed in a relatively low-cost radiation shelter made of slow-weathering plastic with a life expectancy of 15 years. Because the display unit is mounted inside the observer's house, it is easy to read. Currently, about 60 percent of the A (climatological) stations have MMTSs. The MMTS has its share of problems, however. It is prone to failure with fluctuations in power lines and signal lines caused by lightning and other factors. The temperature sensor thermistor and readout equipment are rapidly becoming obsolete and do not permit storage of data. The backup battery supply is limited to four hours; after that, the unit ceases to record, and previously recorded data are lost. In other words, the data that are of most interest (i.e., during storms) are the data most likely to be lost. Burrowing animals sometimes cut the cables, and insects often nest in the shelters. Because MMTSs require cables, the location of the instrument can become an issue. (At one cooperative observer station in Alaska, for example, the new MMTS was installed six feet from the observer's house; the old cotton region shelter was 100 feet from the house. The potential for temperature errors caused by proximity to the house and the consequent discontinuity in measurements are obvious.) Access by NWS personnel and backup observers is difficult because the display is inside the observer's house; consequently, the display unit is often difficult to maintain and/or replace. Figure 2-1 Coop Program management structure. Source: National Weather Service

OCR for page 12
--> BOX 2-1 Exposure of Temperature Instruments The proper exposure of temperature instruments is usually the critical factor for obtaining a long-term record. Liquid-in-glass thermometers or thermistors are the most reliable temperature instruments, but they must be located properly. A temperature instrument in a standard wooden shelter is in a somewhat different environment than an instrument in a much smaller round plastic shelter that houses MMTSs. The shelters differ in thermal capacity and ventilation. The biggest differences, however, are in bright sunlight and calm wind conditions. Differences in exposure can be reflected in maximum/minimum temperatures as well as mean temperatures. About 60 percent of the A stations in the Coop Network now have MMTSs. If a coop station is moved, even for a distance of 20 or 30 meters, air drainage effects, especially at night, can cause differences in temperature. Factors that affect temperature measurements are land surface (grass, rock, etc.), trees near the shelter, and changes in the number of buildings in the vicinity. NOAA has issued guidelines to help manage exposure effects and maintain the continuity of measurements. Procedural Errors By Observers Instrumentation, including the use of paper tapes, represents a serious problem in terms of the accuracy, completeness, and timeliness of coop data. From the standpoint of the climate record, procedural errors by observers are also a serious problem. The high error rate in administrative entries on paper forms submitted to the NWS by cooperative observers represents a substantial problem for data processors at NCDC. But from the standpoint of the climate record, a far more serious problem is created by procedural errors by observers, which result in incomplete, incorrect, or misleading observations. Monthly average temperatures in the United States are calculated using only the daily maximum and minimum temperatures, preferably derived from the 24-hour period corresponding to a calendar day—midnight to midnight. When the ending time of the 24-hour-climatological day varies from station to station or over a period of years at a given station, bias is introduced into the calculated temperature. This so-called ''timeshifting'' is a major problem. Since the 1960s, hydrologic uses of coop data have prompted many observers to shift the time of daily observations from evening to morning. Thus, the daily high temperature reading was actually the previous day's high. When an observer arbitrarily changes the standard time of observation at a station, it causes an "apparent" climate change at that location (see Box 2-2). Nationwide, the change can be significant, as Figure 2-2 illustrates. At a time when climate change is an important scientific and policy question, this issue has serious implications. A related problem is "dateshifting," the practice of entering data for a different date than when the observation was BOX 2-2 The Problem of Timeshifting Monthly average temperatures in the United States are computed using daily maximum and minimum temperatures. For climatological purposes, the preferred measurement period is midnight-to-midnight. However, because readings by human observers at midnight are not feasible for a "volunteer" network, the vast majority of cooperative observers operate on a "climatological day" that does not correspond to the standard midnight-to-midnight calendar day. If the end of the 24-hour climatological day varies from station to station, or over time at a given station, a nonclimatic time-of-observation bias is introduced into the calculated mean temperatures. Thus, when a station changes its time of observation, an "apparent" climate change is introduced into the data set. Random changes have been made in the preferred observation time. For personal convenience, observers sometimes switch from a sunset-to-sunset (p.m.) climatological day to a sunrise-to-sunrise (a.m.) schedule or vice versa. Observations as close as possible to midnight are preferred. Although methodologies have been developed to adjust monthly mean temperatures for differences in observation times, this adjustment introduces uncertainties into the long-term database and adds time to the analysis phase.

OCR for page 12
--> Figure 2-2 Example of the effects of timeshifting. This figure shows the change in January mean monthly temperature resulting from  changing the time of observation from 5:00 p.m. local standard time to 7:00 a.m. Source: National Weather Service and  National Climatic Data Center made or combining daily totals. For example, some cooperative observers do not measure precipitation during the weekend. Instead, they include weekend precipitation in their Monday observation. This problem is difficult to identify because the monthly totals are consistent with those of nearby stations. However, dateshifting skews daily totals and influences statistics on extreme events—a valuable derivative of Coop Network data. Another problem is dropped or missing observations. Sometimes observers get sick, take vacations, leave home for a weekend or longer, use improperly trained substitutes, forget to take observations, record observations illegibly, or are too busy to take observations. Problems also occur frequently at institutional sites (such as radio stations, public parks, or water treatment plants). Although these sites may have around-the-clock staffing, their observers often have high turnover rates, and, because of inadequate training, motivation, or management, they may not be dedicated to taking consistent observations. Forms that contain incorrect observations create another set of problems. For example, temperature errors of 5° to 10°F occasionally are noted day after day, or impossible combinations of maximum and minimum temperatures are noted on an observer's form (i.e., one day's minimum exceeds the previous day's maximum). Many observers fail to record snowfall and/or snow on the ground accurately or at all. Some observers, when faced with one or more days of missing observations, enter representative measurements for several days or for a period of time when the temperature equipment was not reset or precipitation gauges were not emptied. Overall, about 10 to 20 percent of the cooperative data submitted to NCDC in any given month, or about 55,000 to 60,000 individual weather elements, are missing or inaccurate. These errors could be reduced through better recruitment, retention, training, and coordination by the NWS. The cooperative observer's role in transmitting observations also causes problems, especially documentation problems. The manual entry of observations produces a high error rate. Throughout the year, about one-half of all forms received at NCDC (or about 70,000) require some manual correction prior to keying. About 20 percent of the forms that need corrections have entry errors (usually data entered

OCR for page 12
--> Figure 2-3 Sample form for manual entry of weather observations. Source: National Weather Service

OCR for page 12
--> in the wrong column). About 60 percent of the forms requiring corrections have errors in the station data, including administrative information, such as month or time of data observation. About 70 percent contain errors in the recording of meteorological data (not the data values), such as misplaced decimals in precipitation amounts or inconsistencies between the comments on the paper forms and the actual data values. (These percentages exceed 100 percent because a single form may have more than one type of error.) A major problem is created by the data forms themselves. The blocks for entering data on the forms are small, the headings are difficult to read, and rows of blocks and data are difficult to track visually. Figure 2-3 shows Form B-91 used by observers to record daily data at both climate stations and hydrologic stations. Simple changes in format could considerably improve the user-friendliness of these forms. For example, pre-printing the station administrative information would eliminate one type of problem. Declining Number of Sites One continuing problem is "observer drain," the slow but steady decline in the number of cooperative stations. Demographic shifts, such as the demise of small farms and the shift of populations toward the coasts and cities, along with more mobile and faster-paced lifestyles, have made it increasingly difficult to recruit and retain volunteer observers who will record and send reliable daily measurements, year after year and decade after decade. Since its peak in 1972, the number of Coop Network stations has declined by roughly 15 percent (see Figure 2-4). The number of published stations (i.e., those producing high-quality, reliable FIGURE 2-4 Decline in the number of coop stations since 1970. Source: National Weather Service Figure 2-4 Decline in the number of coop stations since 1970. Source: National Weather Service data used in NCDC publications) has also declined significantly. Even the Historical Climate Network (HCN), the most stable group of stations, is losing 1 percent of its sites annually. Unfortunately, meteorological variables, especially precipitation, are not homogeneous throughout a given area, and neighboring stations may have markedly different precipitation totals from the same storm. Observers move, age, and die; properties change hands. If a new owner does not want the responsibility of being a cooperative observer, either the station must be closed or a new observer must be found in a nearby, but different, location. Only a handful of stations around the country have remained in place for a century; about half last 15 years or less at a given location—not nearly long enough to provide the data required for climate studies (see Figure 2-5). Stability of the Coop Network requires that the best observers be retained and that replacements be recruited promptly. Figure 2-5 Site stability of coop stations. Source: Climatic Data Center

OCR for page 12
--> Figure 2-6 Cooperative data flow. Source: National Climatic Data Center Data Transmission and Local Quality Control As Figure 2-6 illustrates, observational data flow along two different paths. Data transmitted daily from Coop Network stations go to NWS WFOs and RFCs. These data are used to prepare forecasts and warnings and are disseminated on the NWS communications network. Data sent via ROSA and ATDTDCS are distributed via transmissions in a standard hydrometeorologic format to other NWS offices and RCCs, and thence to many other users, including most state climatologists. The data are then entered into a precursor to the local data acquisition and dissemination (LDAD) system for widespread daily dissemination. (LDAD is a component of AWIPS, which is scheduled to be installed at all modernized and restructured NWS forecast offices by June 1999.) Since 1988, all monthly transmissions of data from both A and B stations have been sent first to NWS personnel who are responsible for Coop Network observations. After preliminary quality control, the data are forwarded to NCDC for further processing, storage, and dissemination. The current procedures have created bottlenecks in getting coop data into the system and to those responsible for summarizing and disseminating them. The manual data collection and communication processes are both time consuming and labor intensive. For example, NCDC's summary of the day may not be available until 60 or even 90 days after the end of the data month; the hourly precipitation data take about 65 days. The sooner these data and products can be made available, the more useful they are. Automated transmission would greatly accelerate the availability of near-real-time data (see Box 2-3). One problem with ROSA, which is old technology but currently the closest approach to automated transmission, is that the observer has to encode, rather than simply enter, the data, and encoding is an obvious source of many errors. The accuracy of the information varies tremendously among coop observers. In the past, NWS CPMs were aware of the lower-quality stations and compensated for the problems with more stringent quality control. Because quality control specialists at NCDC are not as familiar with individual sites, their handling of these problem sites is less efficient. NCDC estimates that a "bad" station requires 30 times as much time for quality control as a "good" one. Automated

OCR for page 12
--> BOX 2-3 Precipitation Data for the National Centers for Environmental Prediction For daily analyses and forecasts, the National Centers for Environmental Prediction (NCEP) require in-situ observations of precipitation and snow, as well as radar data. The receipt of these data in real-time has been gradually improving. In March 1998, it was about as follows: Hourly precipitation data from gauges were received from about 2,800 sites. The main sources were from the U.S. Army Corps of Engineers, the National Forest Service, and the U.S. Geological Survey. The data from another 2,500 stations that use paper tape were not available in real-time. 850 ASOS stations reported hourly precipitation, but NCEP only received data from about 300. On any given day, NCEP received daily data from 5,500 to 6,000 stations. About 3,500 were estimated to be coop stations. Approximately 10 percent (600) provided only daily summaries of the hourly data. quality control (one of several tiers performed at NCDC) checks overall patterns (e.g., flags instrument "drift" and "scroll") on a monthly, as well as a dally basis; but subtle problems that occur on a multimonth or yearly basis may go undetected. In general, the closer quality control is to the site of the observation, the better the result. Daily observations from many stations (mainly B stations) are transmitted to NWS forecast offices either orally or electronically. Precipitation and river stage data from B stations are fed into operational hydrologic models that predict future river conditions, including floods. The data are also shared with the RCCs. But most data are not disseminated directly to interested parties, such as the media, utilities, and agricultural concerns. Under current procedures, all cooperative observers mail their handwritten observation forms at the end of each month to the NWS.1 There, the DAPM, assisted by HMTs, visually scans the data entry forms, inventories them, and conducts preliminary quality control (mainly identifying missing forms and identifying missing station identification or "metadata," missing observations, and implausible entries). Because of the severe constraints on NWS staff time, quality control at this stage is cursory; NCDC provides more stringent quality control. The NWS rewinds Fischer-Porter tapes, assembles them in batches, and checks for indications of maintenance problems. Understandably, most of the NWS staff time is focused on the coop data from a meteorological standpoint. State climatologists generally receive some data, both hydrologic and temperature data, directly from the WFOs and RFCs; but this is an ad hoc, informal arrangement between individual state climatologists and individual WFOs. State climatologists report that, in the past, they were able to obtain relevant data from the CPMs; but now, with more than twice as many DAPMs as CPMs, the data for each state are often fragmented among several WFOs and are more difficult to obtain. The data are transferred to NCDC via the U.S. Post Office, which also delivers data from cooperative observers to the WFOs. The slowness of this mall-based system is a growing problem because for many users early data is becoming increasingly important, even if they are not complete or fully checked for errors. Metadata In the past, management and oversight of the Coop Network "as a system" has been inefficient, partly because important site-descriptive information (generally referred to as "metadata" or station history information) had to be laboriously entered on complicated federal forms known as B-44s. These forms were filed away in various NWS offices and at NCDC and were not readily accessible. Critical information, such as the latitude and longitude of each site, was often estimated by local NWS officials. No doubt, this led to many siting errors. Furthermore, although site photos had been taken, they were not available to outsiders for oversight purposes. Consequently, some sites were located inaccurately, and instruments were placed at some locations that violated siting standards. Fortunately, technological advances in the last decade, such as desktop computers, file servers with on-line memory, the Internet, hand-held global positioning system (GPS) receivers, and digital cameras, are now available to improve the management and oversight of the Coop Network. It would be relatively easy to equip each NWS office with these tools, require DAPMs to locate coop sites with a GPS receiver, and provide panoramic site photos for on-line computer files. 1    Neither ROSA nor ATDTDCS is configured to retain data or to deliver climatic quality data. Therefore, the manual entry of each observation onto the observer's form and the monthly mailing of that form are essential.

OCR for page 12
--> Manual vs. Automated Observations Manual observations and data entry clearly result in errors and contribute to a slow, inefficient process. The obvious alternative is an automated process, but automation often brings its own problems—the incompatibility of instruments and unstable power supplies, for example. Trade-offs must be made in cost, accuracy, resolution, stability, and maintainability. Table 2-1 shows the advantages and disadvantages of automated and manual cooperative observations. Greater automation, however, appears to be inevitable. In addition to more accurate observations and more efficient reporting, an automated system would provide more frequent observations, including observations at night and observations during periods of severe weather, which require the immediate transmission of data to the NWS. Volunteer human observers are not always available to meet these needs. Today, some cooperative stations do report data in near-real-time through ROSA and ATDTDCS. These data are used for forecasts and warnings and as data for public service programs. The need for more near-real-time data is expected to increase as the NWS moves into an era of improved mesoscale analysis and forecasting. Because neither ROSA nor ATDTDCS is configured to retain data, the manual entry of observations on hard copy and the mailing of forms are still required. With automated hourly temperature and precipitation observations, 24-hour summaries could be derived for any time period. If a clock and a small amount of memory had been designed into the MMTSs, the time of temperature observations at all cooperative stations could have been standardized, and occasional absences of the observer would not be a problem. NCDC officials estimate that, if data input were electronic and quality control were more automated, most cooperative data products could be generated 10 to 15 days after the end of a month (15 to 20 days for precipitation data), compared with the current 60 to 90 days, and preliminary data could be made available immediately. Complete automation need not (and probably should not) be attempted. Partial automation, a simple "interactive data terminal/modem" concept, with backup to diskette, for example, could be a transition phase between manual and automated observations and electronic transmission. A transition phase is discussed in Chapter 3 in the broader context of approaches to automation. During any transition to new observing instruments, every effort should be made to maintain the temporal continuity in the database. Many climate researchers are concerned that this continuity may have been damaged or lost with the introduction of the MMTS. Figure 2-7 shows the bias introduced by this change. For analyses of lengthy time series, NCDC analysts adjust the cotton region shelters/liquid-in-glass temperature data to agree with the newer MMTS data; however, the question of which data set is actually more accurate has not been answered. The automation of cooperative stations should not be done at the expense of data integrity (either of values or consistency). A prudent approach would be to automate selected elements gradually, automating temperature observations first and precipitation data when the technology improves. TABLE 2-1 Comparison of Automated Coop Observations with Manual Observations   Automated Observations Manual Observations Advantages Observations can be gathered quickly. Daily observations can be made at several precise times. Other low-cost instruments can be added. Observations of quantitative precipitation are good. Observations of snow conditions are good. Observations of other events (hail, lightning, wind damage, etc.) can be made. Disadvantages Quantitative precipitation measurements are often not accurate enough for climate applications. Snow observations are not available. Hardware or electricity may fail. No information on general weather conditions is available. Costs (initial and often continuing) are higher. Additional maintenance and training are necessary. Observer may be absent for one or more days. People make mistakes. Time of observation is less precise. Notes: Backups for failures of equipment must be planned. Some manual observations will still be needed. Rapid data collection can still be planned.

OCR for page 12
--> Figure 2-7 Estimated bias introduced by new temperature sensors. Source: National Climatic Data Center Currently, no commercially automated system for measuring all types of precipitation is available that works at the necessary resolution, especially for climate monitoring purposes. Data Assimilation And Quality Control The NCDC has the primary responsibility for processing and interpreting coop data and for disseminating it to users in useful forms. NCDC typically receives the data forms and paper tapes from NWS forecast offices two to four weeks after the end of the calendar month. Because these forms are sent by mail from 119 NWS offices, the NCDC does not receive them all at the same time. After varying amounts of review and preliminary quality control, NCDC prepares the forms and paper tapes for entry into a preliminary electronic database. More intensive quality control of the coop data is performed after keying. Once all data have undergone full quality control processing, the NCDC prepares various data products. At the data assimilation stage, NCDC's DOB catalogs the forms and maintains inventory control. Administrative data (e.g., station name, number, and other heading information) must be verified when the material is received to minimize errors. The data are double-keyed and processed through an interactive computer-edit system. Because the handwritten hard copy is used for initial logging and distribution, the data must be entered manually, which is a labor-intensive, slow, expensive process that is prone to errors (see Box 2-4). Because observers enter the data by hand, a significant number of entries are incorrect or illegible (as distinct from errors in the observations themselves). BOX 2-4 The Cost of an Error One precipitation observation that was wrongly keyed during the summer of 1988 almost cost a farmer his drought insurance claim of $70,000. A rainfall of 0.07 inches was keyed as 0.17 inches, putting the seasonal total above the threshold for collecting on the policy. Only when the records were rechecked was the error noticed. Source: Robinson, 1990.

OCR for page 12
--> After the data have been digitized, validation and quality control are performed. Special software is used to flag suspect data, and errors in temperature, precipitation, snowfall, and snow depth are identified and corrected. All data are then prepared for archiving, publication, and dissemination. At present, nine people in the DOB are assigned to the quality control and preparation of coop data (this figure does not include key-entry personnel or computer specialists). The time involved in manual processing is considerable. Some specific examples are listed below: Paper tapes from Fischer-Porter rain gauges require seven minutes of processing per tape; NCDC processes 2,848 of these tapes per month, for a total of 332 staff hours. Universal rain gauges require 200 hours of processing time per month (one hour for each of 200 sites). Quality control for all manual data forms requires approximately 160 hours per month. Cooperative observers are only one source of the weather data NCDC receives. Data are also received from ships at sea, satellites, aircraft, NEXRAD, ASOS, and wind profilers. Data also arrives from international sources, including world data centers, and as country-to-country exchanges. The data arrive in many different forms, including diskettes, microfilm, film negatives, magnetic tapes, electronic mail, optical disks, video cassettes, and publications. Although cooperative data are only one of many data streams processed at NCDC, they require a disproportionate amount of time to process because manual entry of data from handwritten forms is a slow, error-prone process. Data Dissemination Data from all sources are used in preparing various NCDC products. However, three important products are prepared solely from cooperative data: The summary of the day compiles temperature and precipitation data from all Coop Network sites around the nation on a given day. The hourly precipitation data report provides rainfall on an hour-by-hour basis for a select number of sites. Fifteen-minute interval rainfall is reported for Fischer-Porter sites. The publications, Climatological Data and Hourly Precipitation Data, provide hard copy of daily data for each station on a state-by-state basis. All three are produced monthly. (An annual edition of Climatological Data is also prepared.) The summary of the day and hourly precipitation data are available in hard copy, on diskette, on CD-ROM, and on the worldwide web. Orders for these products, along with copies of the original cooperative data forms, represent about 36 percent of the orders for NCDC data sets (see Figure 2-8). Figure 2-8 Major datasets purchased by NCDC customers. Source: National Climatic Data Center The range of customers for NCDC's cooperative data products has expanded greatly in recent years. The highest percentage of requests now comes from the legal, insurance, and business communities. (NCDC performs an average of 50 certifications of data per day for attorneys.) This increase is attributable to three factors: Orders from legal and insurance customers have increased because of increased litigation. NWS WFOs now refer more customers to NCDC.2 The ASOS installations at airport NWS forecast offices do not measure snowfall, leaving NCDC's Coop Network records as the sole source of data on snowfall near ASOS locations not staffed by NWS personnel. Given the use of paper data forms and hard copy publications, NCDC is heavily burdened with paper. In a typical year, more than one million copies of data bulletins are printed. Some 700,000 copies are sent to 33,000 subscribers. About 1.5 million original data forms are archived. In fact, NCDC has more than 320 million paper records in its archives. Each day, about 1,000 paper copies of original records are sent to users. The cost of handling this much paper is more than $500,000 per year in contractor payments, plus postage. Distributing the summary of the day alone requires one full-time employee and costs, on average, about $50 per coop station per year. The media used to disseminate NCDC products have expanded in recent years. For example, in 1997 paper copies of forms accounted for 54 percent of orders, down from about 70 percent in 1982. The demand for electronic data, especially on CD-ROM, diskette, and web access is growing rapidly. Electronic mail service for filling orders was established at NCDC in 1992. The summary of the day and hourly precipitation databases became available on the NCDC Web Page in 1998. With 1.2 million on-line users per year, the 2    Restructuring of the former weather service offices and WSFOs into fewer weather forecast offices has put pressure on NWS offices to respond to phone and dial-up data demands, which they now tend to pass on to NCDC, RCCs, or state climatologists.

OCR for page 12
--> Figure 2-9 NCDC orders by major media type. Source: National Climatic Data Center NCDC Web Page is now the single biggest source of orders (although they tend to be small orders from individuals, especially students). Although the shift toward the electronic dissemination of products is necessary and inevitable, a growing problem for NCDC is that orders are becoming smaller and thus less profitable. Figure 2-9 shows orders by medium. In response to NCDC's rapidly growing customer base, as well as more and smaller orders, the emphasis is shifting to electronic dissemination rather than hard copy paper products. Management Issues National Weather Service's Management Commitment The priority of the NWS has always been short-range forecasts and warnings. Modernization has, if anything, intensified this focus. Activities related to climate are given a lower priority, even though in recent years climate has become a major driver of policy and funding decisions of government and even businesses. As a result, the priorities of the users of network data and their requirements for how those data are made available have changed while the priority of the NWS management has not. NWS's decision not to focus on climate and the Coop Network was not an easy one. In part, the NWS is driven by forces beyond its control. For example, the RCCs were transferred to the NWS in the late 1980s, but since 1990 NOAA has attempted to drop support for the RCCs from its budget, and Congress has restored the annual funding. The RCCs have recently been placed under NESDIS. The NWS has sustained budget cutbacks and reductions in staff at some locations; in fact, since 1990 its total spending power has been eroded by about $80 million. Implementing expensive new technologies like NEXRAD, ASOS, and AWIPS has attracted most of NWS's funding and attention. In contrast, upgrades to the Coop Network in recent years (an example is PC ROSA) have not been part of a structured modernization plan. New technology has been introduced on an ad-hoc basis. For the most part, staff of weather forecast offices do not consider non-real-time cooperative data as operational data. The panel observed that in offices where more of these data are made available in near-real-time, forecasters depend more heavily on cooperative data—for example, in preparing "nowcasts" and zone forecasts. Ironically, just when the NWS is completing its modernization to improve small-scale forecasts, the program that could best provide the data for the preparation and verification of small-scale predictions is in decline. The NWS's current plan for future operations incorporates a vision of "end-to-end integrated forecasts" (see Figure 2-10). The NWS has continued to focus on short-term, small-scale forecasts, even though its goal is to move toward Figure 2-10 NWS's vision of end-to-end-integrated forecasts. Source: National Weather Service

OCR for page 12
--> the longer time-scale and eventually to produce very long-term forecasts based on an understanding of long-term phenomena, such as ''storm climatologies.'' Cooperative data will play an increasingly important role in this progression by providing measurements of initial conditions upon which forecasts for 10 to 14 days can be based. These data will also be used for verification of the entire range of forecasts, from mesoscale warnings to decadal forecasts and predictions. With further automation of the collection and transmission of cooperative data, data from the Coop Network could also contribute operationally to short-term, small-scale forecasts. In other words, the Coop Network could directly facilitate the NWS' progress toward its long-term goals while, at the same time, becoming a more integral part of NWS's current operations. The panel noted a disturbing trend, even during the course of this study. Other government agencies, such as the U.S. Geological Survey and U.S. Department of the Interior, are reducing or eliminating their support for portions of the Coop Network. Rising per-station support costs, a lack of control over data collection, delays in making data available, and the availability of cheaper (but less accurate and reliable) automatic instrument packages were cited as reasons. The NWS will have to reassess and strengthen its policies related to the Coop Program to retain interagency support for the Coop Network. Effects of NWS Restructuring and Budget Reductions From the perspective of NWS field managers, the Coop Program workload has increased while the resources have decreased. Table 2-2 summarizes the situation for CPMs of the Coop Program up to 1995, when the responsibilities were shifted to DAPMs and HMTs, and compares it with the situation today. Some of these changes are associated with the NWS modernization and restructuring. Others, however, are directly related to budget restrictions that have affected travel, training, and hiring throughout the NWS. Field Management Prior to the modernization, 51 CPMs across the nation were dedicated to the Coop Program. Now, one DAPM assisted by a staff of part-time HMTs manages coop activities for their WFO's area of responsibility. Although the total staff hours devoted (on paper) to the Coop Program has not changed, and some of the former CPMs are now DAPMs, the people responsible for the program all have additional duties. Because Coop Program duties are part-time, DAPMs and HMTs require time and training in conducting their coop TABLE 2-2 Changes in Coop Program Operations (based on information provided by NWS field managers) Prior to 1995 Today One hydrometeorological staff member (the CPM) fully dedicated to the Coop Program in each given area (roughly each state) Several staff members involved in the program part-time Flexibility in scheduling site visits, administrative duties, inspection, recruiting, repair, etc. Limited flexibility in scheduling visits, inspections, recruiting, etc. due to part-time, shared responsibilities of the staff Funding adequate to perform the mission Inconsistent use of time and resources Good support at the national and regional levels Good support at the regional level but not at the national level because of the retirement of the national program manager; no one person available full-time at the national level to act as a liaison with regional or forecast offices from September 1996 to October 1997 Annual regional training conference for CPMs Follow-up training restricted because of funding cutbacks Availability of experienced personnel to assist new CPMs Many more inexperienced field personnel Transportation readily available Transportation for field work shared in some offices; some of the available vehicles inappropriate Time available for troubleshooting equipment and testing new procedures for regional and national headquarters Inadequate time or funding available for troubleshooting or testing new procedures Time available to perform initial quality control of Fischer-Porter tapes before forwarding them to the NCDC Quality control time reduced or nonexistent because of operational requirements in the forecast offices, which are considered more important

OCR for page 12
--> Figure 2-11 Cooperative observing sites per WFO. Source: National Weather Service duties, but most of them have had little or no prior training or experience. The situation is now much more open to conflicts in priorities. The management change from 51 WSFOs to 119 WFOs has also created great variations in workload. On average, each DAPM is responsible for about 100 coop stations (compared to 200 stations for the former CPMs). However, the number of stations per WFO now ranges from as few as 17 to more than 200 (Figure 2-11). The area of responsibility for each WFO is shown in Figure 2-12. Obviously the same staffing model cannot be applied at all of the WFOs because of differences in topography, weather regimes, and forecast areas. For all of these reasons, the Coop Program is operating less efficiently than it was before the staffing was restructured. Most WFOs are struggling to maintain the visitation rate by DAPMs or HMTs. The WFO-based DAPM/ HMT model will require a number of changes in travel policy, hiring, staffing, training, and incentives. Site Visits DAPMs are explicitly required to "ensure the conduct of field visits as required, for the purpose of assuring and/or certifying the establishment, quality, availability, and adequacy of the cooperative and second-order observational programs in the WFO service area" (NWS, 1993). According to former CPMs and current DAPMs, maintaining personal contact with cooperative observers is essential to keeping them motivated, especially if they are located in isolated areas. Cooperative observers either donate their time (the great majority) or provide it for very little pay ($10 to $30 per month, mainly to cover out-of-pocket expenses). In return, they expect recognition of their efforts. Budget reductions over the past several years and especially the severe budget cuts in fiscal year (FY) 1997 that led to an NWS-wide restriction on travel have also affected the site visitation program. In some NWS regions, WFO staff are not permitted to stay overnight even though many sites are a considerable distance from the home office. In addition, many cooperative observers are away from home during the day, so some cooperative observer sites must be visited in the early morning or evening. If DAPMs or HMTs were allowed to stay overnight en route to a coop site, instead of having to return to the WFO, they could visit many more sites in the same number of days and at a lower overall cost. Hiring Because of the hiring freeze, there was a long delay in filling the position of national cooperative program manager after the previous incumbent retired in 1996. Consequently, the Coop Program did not have adequate support from NWS headquarters. The new national program manager was nominated in October 1997, when he took up his duties as acting program manager. The nomination was confirmed in January 1998. Staffing Staffing for the Coop Program is supposed to be changing to "5+1" (five HMTs plus one DAPM) per NWS office. In reality, it appears to be changing to "3+1+2" (three HMTs plus one DAPM plus two interns). If this trend continues, the participation of HMTs will effectively be eliminated. Several NWS field managers told the panel that interns often perceive the Coop Network as a low-tech, part-time duty and that they are more interested in forecasting and modeling severe weather than in visiting farmers and laying MMTS cable. HMT staffing at every WFO is necessary to manage the Coop Program effectively. Training Because Coop Program duties for DAPMs are part-time, adequate training is essential for them. The NWS Training Center offers a Cooperative Network course five or six times per year, with an enrollment of 16 DAPMs per class (a few

OCR for page 12
--> Figure 2-12 Each of the 118 NWS WFOs is responsible for one of the areas outlined above. (A recently established WFO in northern Indiana is not shown.) HMTs also take the course). The eight-day course (CPM01) covers the following topics: Coop Program networks; observer recruitment and training; Coop Program administration; requirements for, and maintenance of, equipment; interagency activities; quality control, forms administration. By the end of 1999, the NWS chief of science and training expects that 400 to 500 DAPMs/HMTs will have completed this course. Training observers during site visits is also important. Many of the procedural errors could be eliminated with adequate observer training. DAPMs are required to "certify and train weather observers" (NWS, 1993); however, because of limited time and the lack of priority on-site visits, DAPMs have had little opportunity to train observers. The NWS Office of Systems Operations does provide written guidelines to coop observers, but this is not enough (NWS, 1989). Incentives Because of the low priority assigned to Coop Program management in most WFOs, DAPMs and HMTs have little incentive to perform their duties fully and conscientiously. A morale problem is making the situation worse. The panel was told that most HMTs do not believe their positions will exist in 10 years. (They expect to be pushed out by the shift toward hiring more science-oriented staff meteorologists.) The absence, until recently, of a national cooperative program manager to ensure that the program received the necessary high-level management resources, attention, and planning also contributed to program deficiencies. Signals from NOAA and the NWS that the Coop Program has a low priority have seriously weakened the management structure. Cooperative observers also have some morale problems. Limited contact with NWS managers has left many of them feeling isolated and unimportant. The threat of automation (and the fear of changes in technology) has also affected their morale. (The panel was told that, when ROSA was introduced, about 20 percent of the affected cooperative observers quit because ROSA increased their workload.) As more changes are introduced, morale problems may become more serious. Program Funding The Coop Program is funded through the operational budget of the NWS but is not treated as a separate program. The funds for communications are usually provided by a mix of

OCR for page 12
--> TABLE 2-3 Coop Program Costs and Reimbursements for FY 1996 Total NOAA costs (NCDC and NWS) $9,256,000 NCDC costs $800,000 NCDC costs recovered $251,642 NWS costs $8,456,600 NWS cost recovered from:   U.S. Army Corps of Engineers $611,370 Bureau of Reclamation $77,578 U.S. Department of the Interior $34,000 Net NCDC costs $548,358 Net NWS costs $7,733,652 Net NOAA costs $8,282,010   Source: NWS national, regional, and local offices. The personnel monies are in the general staffing budget for each office. National, regional, and local offices often share the costs of new equipment. The travel and per diem costs are borne by the local offices. Observers are usually paid from regional offices. Table 2-3 shows Coop Program costs and reimbursements for FY 1996. The total annual cost of each coop station to American taxpayers is estimated to be about $700 (including the annual operating cost and the cost of NCDC operations). The annual cost of $8.2 million for the Coop Program (Table 2-3) includes the cost of publishing and disseminating data products, which is approximately equal to the annual cost of upper-air expendables (mainly weather balloons and their instrumentation) and considerably less than the total annual operating cost of the 850 fully automated ASOS installations nationwide (about $12.5 million per year or $15,300 for each). Many NCDC products are disseminated free of charge to other agencies and to the public. About 30 percent of NCDC's operating budget is underwritten by interagency transfers from government customers and reimbursables from other customers. But income from both is declining. When prices for a number of products were increased, orders from the public went down. In FY 96, cooperative data represented 11 percent of NCDC's reimbursable income, or a little more than $250,000. Thus, the NCDC portion of the Coop Program does not even cover its modest costs. Indeed, the gap between costs and income is widening. System Issues The effective operation and management of the Coop Network requires understanding of the broad system of weather observations and other networks (both national and international). The Coop Network must continue to satisfy its traditional purposes with respect to longer-term climate monitoring and prediction and, at the same time, play a more important role in meeting the national need for near-real-time meteorological data. If stations in the existing Coop Network could report in nearreal-time, the need for parallel networks would be reduced. Some applications that require real-time or near-real-time observations are listed below: flood forecasting (data from thousands of stations in conjunction with radar data) numerical weather forecasts, forecast verification, and improved models calibration of radar data monitoring of crops for agriculture operational weather forecasting by NWS offices highway conditions and road crew work fire weather All of these tasks require a large number of observations that can be gathered quickly. The requirements for the accuracy and the long-term continuity of measurements are less stringent in these applications than when the data are used to determine long-term climate trends and statistics. Mesonets And Regional Networks The panel was asked to identify approaches for improving the effectiveness and efficiency of the network through new technology or a new organizational structure associated with the NWS modernization. Automated observing networks, such as state and local mesonets, are reviewed in that context. Local or regional networks are customized to meet their users' needs and are usually automated to provide near-real-time access to data. Mesonets are operated by federal, state, and local governments, as well as by private-sector organizations. Some state transportation departments have installed automated data stations along the roadsides. Some power utilities and large cities have small networks to provide meteorological data at key locations. Data from many of these networks are also used by the NWS. For example, the NWS collaborated with the University of Georgia's mesonet to provide weather coverage for the 1996 Olympic Games in Atlanta (see Box 2-5). Examples of collaborative local and regional networks are listed below: The Regional Observation Cooperative of the Forecast Systems Laboratory is organizing various existing reporting sites into a mesonet to collect, process, analyze, and disseminate surface observations from Colorado and adjoining states. A large number of networks of various sizes have been established in agricultural areas around the country. An Oklahoma Mesonet of 114 stations, jointly supported by Oklahoma State University and the University of Oklahoma, is operated by the Oklahoma Climatological Survey (see Box 2-6). Data are collected and

OCR for page 12
--> BOX 2-5 Olympic Weather Watch The XXVIth Olympiad was held during the hot, humid, thunderstorm-prone Atlanta summer. To ensure the success of the games, the NWS provided weather information to athletes, spectators, and the media. Weather conditions were monitored using satellites, Doppler radar, and a network of surface monitoring stations that comprise the University of Georgia's Automated Environmental Monitoring Network (AEMN). The AEMN was supplemented by additional stations installed by NWS in data-void areas to support high-resolution numerical models. Two NWS Olympic weather support offices in Georgia received and analyzed the data and provided weather forecasts for each venue of the games. The forecasts were transmitted to officials, coaches, and athletes through the Atlanta Olympic Committee's information system and were broadcast to numerous hotels and made available to the media. Source: Hoogenboom and Garza, 1996. transmitted automatically every 15 minutes and are available to users about five minutes later. Figure 2-13 shows a typical station in the Oklahoma Mesonet. The Educational Network has 2,000 sites mounted on school roofs; the Four Winds Network is a small network located at schools in the Washington, D.C., metropolitan area. The purpose of both networks is to educate students and communities about the importance of environmental data. Generally speaking, although mesonets have some important capabilities that the Coop Network does not have, they lack many features that have made cooperative data so valuable. For example, mesonet data are mostly tailored to meet the needs of particular users and are focused on weather data rather than climate; in some cases, there is less quality control of the data, they are less available to the public than the coop data, and their period of record is much shorter than the 100-year record of the Coop Network. In many cases, the type of measurements is limited (e.g., no observations of snow, hail, thunder, etc.). Some mesonets do not archive or summarize their data, and some lack routine maintenance and calibration or do not meet basic exposure standards. Some mesonets resemble a low-cost ASOS, but with fewer expensive instruments and added low-cost solar sensors. Mesonets can collect data quickly at fixed times, but precipitation measurements usually come from tipping buckets, which are not as reliable or accurate as those from the Coop Network's standard 8-inch precipitation gauges, and no snowfall or snow depth measurements are included. If mesonets could be standardized, they could possibly be incorporated into the Coop Network. A "network of networks," partly government funded and partly privately funded, could provide increased coverage. However, a BOX 2-6 The Oklahoma Mesonet The Oklahoma Mesonet, perhaps the most extensive and successful of all mesonets to date, is a joint project of the University of Oklahoma and Oklahoma State University. Built in 1993-94, it has 114 environmental monitoring stations statewide—more than one per county. Each automated station (see Figure 2-13) measures temperature, humidity, wind speed and direction, barometric pressure, solar radiation, soil temperature, soil moisture, and leaf wetness. Data are collected every 15 minutes (3 sets of 5-minute observations) and are available via the statewide law enforcement communications system. The operating budget for the mesonet is about $1 million per year, with a maintenance budget of about $900,000. The system is partly supported by user fees. As of late 1997, there were 500 authorized users of the data, with the largest categories being K-12 schools and teachers (who use the mesonet data as a classroom teaching tool), university researchers, and agricultural agents. Access is restricted via password and user software. Data are disseminated by means of an electronic bulletin board and the World Wide Web. Source: Oklahoma Climatological Survey.

OCR for page 12
--> Figure 2-13 Oklahoma Mesonet station configuration. Side view (to the north) of a typical Mesonet station. Source: Oklahoma Climatological Survey serious disadvantage of relying on mesonets is the uncertainty of their long-term sustainability and reliability. Because many local networks are beyond government control, relying on them to support federal programs would be risky. (Cooperative observing networks in other countries are described in Appendix C.) Noaa's National Weather System To understand the changes necessary for improving the Coop Network, the network must be seen in the context of the overall system of NOAA weather and climate data services. The place of the Coop Network in the overall system—the relative scope and scale of Cooperative Program activities—is an essential guide to determining its future. Figure 2-14 shows a high-level view of weather-related activities under NOAA, reflecting the completely modernized NWS. The figure illustrates the overall architecture for the NOAA National Weather System that the NWS plans to use to guide the development of its current and future weather modernization technologies and climate services. Two of the stated objectives for the architecture are to "provide for efficient and timely delivery of data from national observing systems to the NCDC" and to "support the integrity of the long-term climate record from weather systems" (NOAA, 1997). Cooperative observers (upper left in the figure) are only one of several sources of environmental data; Coop Network data are shown as being transmitted automatically into the LDAD/AWIPS at WFOs. Automated sources in the system (ASOS, other surface observation networks, NEXRAD, satellites, etc.) will soon produce—indeed, are already producing—an ocean of data, and coop data will represent a mere "drop in the bucket" in terms of volume. The NWS Office of Meteorology is currently conducting a study of the status of all federal, state, county, local, and private surface observational networks to determine the feasibility of integrating them into a real-time national network.

OCR for page 12
--> Figure 2-14 NOAA National Weather Service high-level component system view (cooperative observers are on the top left portion of the figure.). Source: National Weather Service

OCR for page 12
--> An integrated surface observation network would maximize the utility and cost effectiveness of the national investment in surface observations. Although NOAA has developed a concept of the modernized National Weather System architecture, there is no comprehensive observing system architecture for surface observations comparable to the planning architecture for atmospheric observations under the auspices of the North American Atmospheric Observing System (NAOS) Program. One of the primary features of NAOS is a scientific evaluation program to assess the value of various combinations of upper-air observing systems to numerical predictions and operational weather forecasts. The NWS Office of Meteorology appears to recognize the need for, and has initiated a study of, an integrated surface observation network to leverage the information and resources of the many independent sources of surface observations throughout the United States. A comprehensive observing system architecture for integrated surface observations would provide a clear blueprint for the future management and operation of the Coop Program as part of the National Weather System.