Observing Weather and Climate from the Ground Up: A Nationwide Network of Networks (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

6 How to Get from Here to There: Steps to Ensure Progress Several steps are required to evolve from the current circumstance of disparate networks to an integrated, coordinated network of networks (NoN). First, it is necessary to firmly establish a consensus among providers and users that a NoN will yield benefits in proportion to the effort required to establish it. This consensus-building step is essentially political, requiring agreement in principle at various levels of public and private participation, which leads to the collaborative development of an implementation plan. The new elements of a NoN are twofold: (1) the provision of services and facilities that enable individually owned and operated networks to function, more or less, as one virtual network, and (2) the provision of new observ- ing systems or facilities to enable the national observational goals. The former is largely separable from the latter, since considerable benefit may be achieved from improved functionality with existing observational assets. In this chapter we identify a minimum set of essential core services and facili- ties that must be established in order to realize the dream of a NoN. We also discuss the need for augmentation of infrastructure, which is critical to a systematic evolution of the NoN. Chapter 7 will address the broader organizational implications of a NoN if it is to become a fully integrated observing system that meets multiple national needs. PLANNING FOR THE FUTURE: CONVENING THE STAKEHOLDERS Recommendation: Stakeholders, including all levels of government, var- ious private-sector interests, and academia should collectively develop 138 

HOW TO GET FROM HERE TO THERE: STEPS TO ENSURE PROGRESS 139 and implement a plan for achieving and sustaining a mesoscale observ- ing system to meet multiple national needs. The plan should recognize and account for the complexity associated with the participants’ differing roles, responsibilities, capabilities, objec- tives, and applications, as well as lessons learned from past experiences. To launch the planning process • A mesoscale environment observing system summit should be con- vened to discuss and recommend the implementation of a NoN and to prescribe a process through which a plan will be developed. Participants from the private sector, federal executive branch, U.S. Congress, national organizations of governors and mayors, and key professional societies should attend. • Forums to further discuss and recommend implementations of the mesoscale observing system should be organized by professional societies and associations such as the American Meteorological Society, National Council of Industrial Meteorologists, American Geophysical Union, Com- mercial Weather Services Association, National Weather Association, American Institute for Chemical Engineering, American Society for Civil Engineering, and American Association of State Highway and Transporta- tion Officials. A leading role should be assumed by the Commission on the Weather and Climate Enterprise of the American Meteorological Society, the constitution of which is particularly well suited to this task. IMPROVING THE USE AND VALUE OF EXISTING ASSETS: ESSENTIAL CORE SERVICES Essential core services are defined as those services required to derive levels of function and benefit from a NoN that markedly exceed those cur- rently realized from the assemblage of relatively independent networks. Essential core services include but are not limited to • definition of standards for observations in all major applications, • definition of metadata requirements for all observations, • certification of data for all appropriate applications, • periodic “rolling review” of network requirements and user expectations, • definition and implementation of data communication pathways and protocols, • design and implementation of a data repository for secure real-time access and a limited period for post-time access, 

140 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP • generation of a limited set of products based upon the raw obser- vations, most notably graphical presentations of data fields and analyses thereof, • pointers to more sophisticated products generated externally, such as analyses produced from a short-term model prediction and multiple observation sources, • pointers back to data providers, where more products and services are available, • establishment of a link to National Oceanic and Atmospheric Administration’s (NOAA) National Climatic Data Center (NCDC) for archival of selected data, as deemed appropriate by NCDC, • development and provision of software tools and internet connec- tivity for data searches, information mining, and bulk data transmissions, • development and provision of a limited set of end-user applications software, which would enable selection of default network data configura- tions for major applications as well as tools for creation of custom network data configurations, and • provision of a data quality control service with objective, ­statistically based error-checking for all major categories of data, including manual intervention and feedback to providers. The premise for these services is to • have expert assistance in establishing and maintaining standards for the data provided, • know which additional data are available and suitable to one’s own application, • have compatibility with and ease of access to selected observations and analyses, • ensure the archival of selected data commensurate with their useful lifetimes, and • gain ease of access to the products and services of other providers. The Primacy of Metadata Metadata (information about the instruments themselves and how they are sited and used) should be required of every component in an integrated, multi-use observing system and should be kept up to date. Observational data have maximum value only if they are accompanied by comprehensive metadata. Examples follow: • contact information for person or organization responsible for providing data 

HOW TO GET FROM HERE TO THERE: STEPS TO ENSURE PROGRESS 141 • type of data provided and parameters measured • instrument type (e.g., liquid in glass thermometer, cup anemometer with vane, radar wind profiler, satellite radiometer; may need to include specifications such as beamwidth, operating wavelength or frequency, pulse repetition frequency, sampling time, etc.) • instrument manufacturer and part numbers • date of installation or most recent upgrade • manufacturer specifications for accuracy and precision • location of instruments (latitude, longitude, elevation); height above ground at which each parameter is measured • site description (e.g., open, grassy field; roof of school; under a tree) or observation platform (e.g., satellite, balloon, aircraft) • nearest obstacles preventing view of horizon; their distance and height (not applicable in all cases) • frequency of maintenance • time and frequency of measurement • information about any on-site data processing (e.g., averaging, smoothing, thinning) • frequency of data transmission • data format (units and order of magnitude information) • mode of transmission: land line, wireless communication, micro- wave, uplink to satellite • data latency (length of time between the raw measurement and receipt of the report by a collection center • documentation of any changes in instruments and their location or exposure since the site was established Given this information, each instrument at each site in the network of net- works can be continually monitored and evaluated regarding its utility in various categories of user application. Data access may then be streamlined to exhibit to the user a network configuration of greatest relevance to the particular application. As discussed in previous chapters, today’s metadata in support of mesoscale observations are incomplete at best and, in the case of surface observations, woefully inadequate for the great majority of them. The col- lection and maintenance of comprehensive metadata is necessary, tedious, time-consuming, and labor-intensive, but not nearly as costly as deploying new observing systems. Many mesonet data providers want to see their instruments perform up to specifications and serve the intended purposes. They are inclined to respond when end users or automated quality-checking software detect questionable data. The United States needs a program that informs the providers of mesoscale observations why metadata are so vital for quality-checking for use in multiple applications. 

142 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP Provision of metadata should be mandatory for membership in the NoN. Incentives and assistance should be offered to the operators of net- works who provide it. The contents of a metadata file should be carefully defined, and, once assembled, a national database of metadata should be frequently updated and accessible to all. If action is taken to improve meta- data and to fill gaps by supplying comprehensive information on undocu- mented systems, the value and impact of existing data will be improved far beyond the cost of gathering the metadata. Standards for Instrument Sites and Exposures In the case of surface observations, which are more numerous, diverse, and variable in quality than other observing systems, much useful guidance for traditional meteorological measurements has been compiled. • Siting criteria for the Oklahoma Mesonet stations are available in Shafer et al. (1993). Examples of metadata for the instruments at these sites appear in McPherson et al. (2007). • The World Meteorological Organization lists 20 principles in GCOS Climate Monitoring Principles. • Siting criteria for observations in urban areas are available in WMO (2006) and Oke (2007). • Siting standards for Road Weather Information Systems are described in a report by Mandredi et al. (2005), which is available at http://ops.fhwa.dot.gov/publications/ess05/ess05.pdf. Standards for data accuracy in specific applications vary widely. For example, the temperature at a climate monitoring site would normally be measured more accurately than the temperature in a school yard. In one case, climatologists would like to discern temperature trends of a fraction of a degree over a period of decades. In the other case, a teacher might decide whether it is safe for the children in her charge to go outside for recess. Quality-Checking of Observational Data Four methods of quality-checking are commonly employed: (1) Deter- mine whether the measured value is physically plausible (“engineering” check). (2) Compare an observed value with nearby neighboring values (“buddy” check), either directly or by means of a more sophisticated a ­ nalysis, under the assumption that conditions in the neighborhood are   http://www.wmo.ch/pages/prog/gcos/Publications/GCOS_Climate_Monitoring_Principles. pdf. 

HOW TO GET FROM HERE TO THERE: STEPS TO ENSURE PROGRESS 143 nearly uniform. This is often effective in uncovering the larger random errors. Usually, only like observations are compared, but there are excep- tions. (3) Compare the observed value with a predicted value (for example, a value extracted from a 1-hour model forecast). Often called a “back- ground” check, this comparison can uncover biases in an observing system. (4) Conduct periodic field checks of the sensors and their surroundings, and laboratory calibrations as necessary. See, for example, Shafer et al. (2000). Rolling Requirements Review Once every few years, the World Meteorological Organization (WMO) conducts a Rolling Requirements Review, first mentioned in Chapter 4, a survey of observing systems worldwide and their effectiveness in meet- ing needs in a number of application areas: global NWP, regional NWP, s ­ ynoptic meteorology, nowcasting and very short-range forecasting, ­seasonal to inter-annual forecasting, aeronautical meteorology, atmospheric chem- istry, agriculture, oceans and coastal regions, and hydrology. Each review concludes with recommendations for improving space-based, atmospheric in- situ, and surface-based components of the global observing system. The most recent WMO “Statements of Guidance” concerning observing systems in the above application areas may be found at http://www.wmo. int/pages/prog/sat/documents/SOG.pdf. Similar rolling requirements reviews, of which this National Academies report could be considered the first, would serve the United States well. The U.S. focus in this instance would be on mesoscale applications in the short time frame of 2 days or so, particularly events hazardous to health and safety and/or affecting the economic sectors discussed in Chapter 3. AUGMENTING EXISTING INFRASTRUCTURE The previous section discussed ways in which the use and value of cur- rent data can be enhanced in the absence of new or improved observing sys- tems. This section examines the augmentation of existing observational and computational infrastructure, including the establishment of new observing systems, diagnostic tools, and experimental facilities. The Role of Observational Testbeds Dabberdt et al. (2005b) describe a testbed in the following way: A testbed is a working relationship in a quasi-operational framework among measurement specialists, forecasters, researchers, the private ­ sector, and 

144 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP government agencies aimed at solving operational and practical ­ regional problems with a strong connection to the end users. Outcomes from a test­ bed are more effective observing systems, better use of data in forecasts, improved services, products, and economic /public safety benefits. Testbeds accelerate the translation of R&D findings into better operations, services, and decision-making. A successful testbed requires physical assets as well as substantial commitments and partnerships. The main purpose of an observational testbed is to demonstrate that a particular collection of new observations improves regional weather predic- tions and all of the decisionsupport systems that affect life, property, and economic well-being. Testbeds enable the acquisition of knowledge on how best to sample atmospheric properties and phenomena, oftentimes resulting in improved knowledge of the observed phenomena and their statistical properties. Effective testbeds have the following hallmarks: • They require considerable advanced planning, resources, personnel, and time (often more than 1 year); hence, they are used sparingly and only when multiple purposes are served. • Testbed planners enlist the support of stakeholders and involve them in planning. • The program plan states expected outcomes and defines the mea- sures of success. • The program is flexible; it can adapt to changing conditions. • By definition, the testbed is limited in scope, but it is clear how to generalize results to larger regions or to the solution of other problems. • The testbed is end-to-end, starting with observations and ending with decisionmaking by stakeholders. • Ideally, the testbed operates in real time in a quasi-operational setting. The essence of a testbed is the test and refinement loop illustrated in Figure 6.1. If the experimental observations or derived products stand up to rig- orous tests of utility, accuracy, reliability, computational efficiency, cost- effectiveness, and repeated scrutiny by users, they can make the transition to operations. Otherwise, user feedback leads to modifications and another round of testing or to the elimination of the proposed observing system. The testbed is but one step on the pathway from research and develop- ment to operations (routine, real-time application). The sequence of steps is • Develop a concept for a new observing system. 

HOW TO GET FROM HERE TO THERE: STEPS TO ENSURE PROGRESS 145 Introduce new Experiment and observing systems to demonstrate. improve forecasts and decision-making. Input Test and refinement Assess impacts. Revise and iterate. loop New observing systems become operational. Output FIGURE 6.1  Conceptual schematic of a testbed refinement loop (Dabberdt et al., 2005). SOURCE: Reprinted with permission from the American Meteorological 6-1.eps S ­ ociety. © Copyright 2005 American Meteorological Society (AMS). • Build a prototype. • Calibrate the instrument in the laboratory; compare its measure- ments with those of similar instruments. • Field test the observing system for performance, ruggedness, and reliability. • Incorporate the system in a testbed; integrate it with other observ- ing systems. • Verify that the new observing system yields positive benefit for the intended application and collateral benefits for other applications when used with existing systems. • Deploy the new observing system in an operational network. An example of a working testbed is the Hydrometeorological Testbed (see http://www.esrl.noaa.gov/psd/programs/2008/hmt/), which deploys special observational assets in a quasi-operational environment to improve the prediction of winter storms that cause flooding along the West Coast. That testbeds already exist is reason to include them in this section on “Augmenting Existing Infrastructure.” Yet testbeds will evolve to serve broader needs. A spectrum of additional testbed applications related to 

146 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP network design is envisioned, including the application of observing system simulation experiments (OSSEs), the development of assimilation systems for chemical weather, and the addition of key sensors and their siting to facilitate the merging of in-situ and satellite observations to provide finer- resolution spatial distributions. Recommendation: Federal agencies and partners should employ “test- beds” for applied research and development to evaluate and integrate national mesoscale observing systems, networks thereof, and attendant data assimilation systems. Among other issues, testbeds should address the unique requirements of urbanized areas, mountainous terrain, and coastal zones, which currently present especially formidable deficiencies and challenges. Diagnostic Studies Closely akin to the notion of testbeds are various forms of diagnostic studies, both observationally and numerically based. In Numerical Weather Prediction (NWP), one can gauge the worth of specific observing systems in two ways. First, for existing systems, one can include and then withhold specific observations from the initial analysis (e.g., eliminate all aircraft observations from the initial analysis) and see what happens to the forecast. One can also thin existing observations (e.g., cut the number of rawinsonde soundings in half) and examine the degradation in forecast accuracy. Alternatively, one may temporarily enhance the operational network with research-based sys- tems to examine forecast accuracy improvement. These are called observing system experiments (OSEs). A few numerical weather prediction centers conduct OSEs annually. Such experiments help to determine which observ- ing systems most affect forecast accuracy. On the basis of OSEs, one might conclude that fewer observations of a certain type might not harm the forecast but more observations of a different type would improve it. Second, with regard to proposed or experimental observations, one can conduct an OSSE. An OSSE estimates the effect on forecasts of adding a new observation source (e.g., a Doppler wind lidar aboard a satellite). The major difference between OSEs and OSSEs is that all observations in the latter must be simulated: the new system that is being evaluated and all systems currently used in operational data assimilation, namely those systems that the new system will be competing against. A credible OSSE requires “calibration” from known observing systems, extensive compu- tational resources similar to those used in operational numerical weather prediction, and careful execution. 

HOW TO GET FROM HERE TO THERE: STEPS TO ENSURE PROGRESS 147 Communication Many observing systems produce voluminous information. For most applications, at least near real-time communication is essential. Ground- based remote sensors such as radars and lidars have intrinsically high data rates, necessitating an adequate communications bandwidth, which has increasingly become available and affordable. Rapid and flexible access to stored data is often a point of failure, requiring efficient data structures and applications software that are well matched to a wide range of user needs. A communications architecture that permits selective access to full resolution or general access to lower resolution data and analyses should be devised. Larger market forces governing the evolution of the data communications and data storage industries over the coming decade should easily accom- modate these requirements. User Interface In order to screen mesoscale observations for specific applications, a sophisticated user interface is required. Behind this interface is a rela- tional database that contains comprehensive metadata from each observ- ing source, a pointer to the repository of each source, and high-bandwidth communication to each repository. These attributes will make it possible to retrieve information based upon highly selective criteria. In the future, given sufficiently detailed metadata for each observing source and specific enough criteria for the intended application, it should be possible to extract from geographically distributed repositories just the information that directly serves the application, no more and no less. Specific examples will more effectively make this point than generalizations: • Search by application: “Show me highway pavement, temperature, and visibility conditions in north central Illinois.” • Search by application: “Show me regional chemical weather fields east of 85ºW.” • Get information in the vertical: “Get me the best estimate of the current vertical profiles of temperature, moisture, and wind over Chicago’s O’Hare Airport.” • Search for information that is sensor-, time-, and location-specific: “Show me precipitation data from tipping bucket rain gauges in Missouri between 1800 UTC 6 Aug and 0600 UTC 07 Aug 2007.” • Search for information from a specific network: “Show me all tem- perature data from the AWS (WeatherBug) surface network at 1200 UTC 06 Apr 2007.” 

148 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP • Search for observations of a specific variable, imposing criteria for instrument exposure: “Show me all winds measured in the last hour between 8 and 20 m above the surface where no obstacles higher than 20 m exist within 50 m of the anemometer.” • Search a historical archive: “Show me where no precipitation has fallen in the past 30 days.” • Search according to a designated threshold: “Show me where the wind speed currently exceeds 30 knots.” Clearly, the user interface capable of fulfilling these requests must be versatile, and the database must be quickly accessible. No interface with this sophistication currently exists. IDENTIFYING A CENTRALIZED AUTHORITY Recommendation: To ensure progress, a centralized authority should be identified to provide or to enable essential core services for the net- work of networks. Initially, the focus of such activities should be on markedly improved use and value of data from existing observing systems. As new observa- tional, computational, and communications infrastructure is added, the focus should shift to the prompt and seamless accommodation of these new elements and their related objectives. The provision of core services is essential for adequate access to and the utility of mesoscale observations as applied to multiple national needs. What the Centralized Authority is Not The recommendation for a modest degree of centralization is tightly focused on essential core services. It specifically excludes centralization for the purpose of acquisition and operation of observing systems, which are owned and operated by agencies, corporations, and other organization to serve their specific missions. The centralized authority is an enabling element of the broader enterprise that comes into play only insofar as it is necessary to derive added utility and functionality from the network of networks. It does not speak to the ownership, operation, upgrading, or maintenance of the individual networks themselves. It follows that the centralized authority is envisioned as a relatively small but vital fraction of the entire NoN enterprise.