2
Observations Supporting the Fundamental Infrastructure for Mesoscale Monitoring and Prediction

This chapter focuses on an intermediate class of users whose primary functions are provision of current weather information, watches, warnings, and forecasts; generation of weather analyses and predictions by computer; and monitoring climate trends. Their products are publicly available on radio, television, the Internet, or by subscription. The largest user by far in this class is the National Weather Service, whose principal mission is the protection of life and property. Other large users are private-sector firms that provide free access to some of their products via the web and other media as well as specialized, fee-based services. We consider here the broad, observational needs of these intermediate users rather than the more specific needs of their customers.

Tax dollars pay for almost all the observations used by this intermediate class. In most cases, the U.S. government either operates and maintains the observing system or pays private corporations to do so, for example, the National Lightning Detection Network. It is fair to say that the observing systems supporting the intermediate class are the backbone of environmental service systems, ranging from simple displays of weather data to sophisticated products and decision-support tools. Without raw observations, their assimilation into atmospheric models, and computer-generated forecasts, the products tailored for the specific applications to be discussed in Chapter 3 would not be possible.

Given the heavy reliance of intermediate users on the raw observations and computer analyses and predictions, and the emphasis of the National Weather Service on the protection of life and property, this section focuses on observations required to support these functions. More specifically, it



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 23
2 Observations Supporting the Fundamental Infrastructure for Mesoscale Monitoring and Prediction This chapter focuses on an intermediate class of users whose primary functions are provision of current weather information, watches, warnings, and forecasts; generation of weather analyses and predictions by computer; and monitoring climate trends. Their products are publicly available on radio, television, the Internet, or by subscription. The largest user by far in this class is the National Weather Service, whose principal mission is the protection of life and property. Other large users are private-sector firms that provide free access to some of their products via the web and other media as well as specialized, fee-based services. We consider here the broad, observational needs of these intermediate users rather than the more specific needs of their customers. Tax dollars pay for almost all the observations used by this intermediate class. In most cases, the U.S. government either operates and maintains the observing system or pays private corporations to do so, for example, the National Lightning Detection Network. It is fair to say that the observing systems supporting the intermediate class are the backbone of environ- mental service systems, ranging from simple displays of weather data to sophisticated products and decision-support tools. Without raw observa- tions, their assimilation into atmospheric models, and computer-generated forecasts, the products tailored for the specific applications to be discussed in Chapter 3 would not be possible. Given the heavy reliance of intermediate users on the raw observations and computer analyses and predictions, and the emphasis of the National Weather Service on the protection of life and property, this section focuses on observations required to support these functions. More specifically, it 

OCR for page 23
 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP focuses on observations for accurate numerical weather analysis and predic- tion, timely watches and warnings in advance of hazardous weather, and the special requirements of climate monitoring. The charge to this committee was to (1) focus on time scales less than 48 hours, but keep longer time scales in mind; (2) focus on U.S. and adjacent coastal regions, but keep global observing system requirements in mind; (3) focus on ground-based in situ and remote sensing observations, but keep the utility of satellite observations in mind; (4) focus on the atmo- spheric boundary layer, but keep the deep troposphere in mind. Within this context, the hazardous weather events most important to detect, monitor, and predict are • flooding from a large-scale storm • Nor’easters • snowstorms and ice storms (For the above three items, precipitation type, intensity, and amount [in the case of snow and ice, liquid equivalent and accumulation on the ground] are all important.) • hurricanes and tropical storms air pollution1 • • thunderstorms, including mesoscale convective systems — lightning — flash floods — hail — straight-line damaging winds (resulting from squall lines or bow echoes) — tornadoes • windstorms without precipitation — downslope windstorms — pressure-gradient windstorms • fire weather • aviation hazards — in-cloud icing — downbursts — aircraft turbulence The order in the above list is roughly by size and longevity. The time and space scales associated with these phenomena are depicted in Figure 2.1. All 1This report covers the release of toxic substances, accidentally or deliberately. This topic is clos- est to “air pollution,” but, since it is not a natural phenomenon, it is not treated in Appendix A, nor is it mentioned in Table 2.1. The spatial and temporal scales for toxic releases (0.2 to 2.0 km and 15 min to hrs, respectively) are generally smaller than those for air pollution.

OCR for page 23
 OBSERVATIONS SUPPORTING THE FUNDAMENTAL INFRASTRUCTURE these phenomena are “high impact” in that they affect life, property, and the economy. All pose the same problems for forecasters: time of initiation, intensity and intensity variations, and end time. And, although some of the phenomena near the top of the list are large and persist for days, mesoscale features embedded within them, especially convective elements, cause most of the havoc. Observations useful in the context of this study are equally useful for monitoring phenomena that lie outside the time-space envelope considered FIGURE 2.1 Time and space scales associated with the “high-impact” weather 2-1.eps phenomena that are discussed in Appendix A and summarized here. NOTES: The bitmap image scale is logarithmic in both directions. Common units of time are noted on the vertical coordinate. Common notions of size are listed on the horizontal axis. The sizes and lifetimes associated with each phenomenon are typical but not necessarily definitive. Not portrayed is the size of mesoscale features that may be embedded within the larger events.

OCR for page 23
 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP here, for example, a shift in the large-scale hemispheric circulation pat- tern, heat waves, drought, and climate change. Reanalyses of atmospheric observations collected over many decades with sophisticated assimilation systems, in effect, synthesize all we can know about atmospheric behavior since the dawn of the meteorological satellite era. A PHENOMENOLOGICAL APPROACH TO OBSERVATIONAL REQUIREMENTS For each phenomenon listed above, we ask • Why is the phenomenon important? • What variables (e.g., temperature, moisture, wind) are sufficient to characterize the phenomenon? • What spatial density of observations and what frequency of mea- surement are required not only to detect and monitor the phenomenon but also to describe its internal workings and predict its onset and future behavior? Appendix A addresses these questions in detail and gives a rationale for choosing a spatial and temporal resolution appropriate for observing each phenomenon. As far as we know, such an analysis has not appeared else- where. A summary of the discussion in Appendix A follows in Table 2.1. The phenomena listed in Table 2.1 are familiar to everyone. Another atmospheric entity, gravity waves, are virtually unknown to the general public, yet they affect many of the listed phenomena. Gravity waves are ubiquitous in the atmosphere. They are wave disturbances in which buoy- ancy acts as the restoring force on parcels displaced from their equilibrium position. Gravity waves can spawn thunderstorms, generate severe turbu- lence in the vicinity of mountains, create persistent chinook (warm and dry downslope wind) conditions, increase winds and the snowfall rate in winter storms, and drive trace amounts of gas from the soil through “pres- sure pumping.” A characteristic wavelength for gravity waves ranges from a few kilometers to several hundred. A characteristic propagation speed is 10-20 m s–1. In order to capture gravity waves, the observational net has to be fine enough to resolve the layer of static stability in which the waves move, most often at the tropopause or in the lower troposphere, and often within an inversion. This implies, roughly, a horizontal resolution of 5-10 km and a vertical resolution of 100 m, in temperature, moisture, and wind measurements. The main conclusions to be drawn from Table 2.1 and the brief men- tion of gravity waves are

OCR for page 23
TABLE 2.1 Emphasizing observation requirements not currently met in the vicinity of listed phenomena that would improve definition of mesoscale structure and predictions out to 48 hours Resolution Parameters to Phenomenon Size Duration Observe ∆x ∆t ∆z Flooding from 300-2000 km 0.5-5.0 days Temperature 50 km 3h 200 m (up to 5 km MSL) large-scale storms Moisture Wind Precipitation Nor’easter 500-2000 km 0.5-4.0 days SST 10 km 12 h Temperature Moisture 50 km 3h 100 m (up to 12 km) Wind Snowstorms swaths 2 h-2 days Temperature 30 km 2h 100 m (up to 5 km MSL) Ice storms 200 km wide Moisture 1000 km long Wind Precipitation Hurricanes and 100-2000 km 1-7 days Temperature Track forecasts tropical storms Moisture 100 km 6h 500 m (up to 16 km) Wind Intensity changes SST 10 km 3h 200 m (up to 16 km) Precipitation Air pollution and 20-1000 km 6 h-5 days Temperature Metro areas toxic releases Moisture 5 km 15 min 50 m | up to top of mixed Wind Rural areas | layer, generally Sources/sinks 20-30 km 30 min 50 m | below 4 km AGL) Concentration  continued

OCR for page 23
TABLE 2.1 Continued  Resolution Parameters to Phenomenon Size Duration Observe ∆x ∆t ∆z Fog and 10-500 km 1 h-1 day Temperature 25 km 15 min 30 m (up to 3 km AGL) low clouds Moisture Wind Lightning 1-20 km 5 min-1 h Temperature Prediction of thunderstorm initiation Moisture 2 km 15 min 100 m (to top of PBL) Wind Soil moisture Spherics Flash floods 2-20 km 5 min-1 h Temperature Assess instability Moisture 50 km 1h 200 m (up to 12 km) Wind Characterize sub-cloud layer Soil moisture 20 km 15 min 100 m (up to 2 km AGL) Precipitation Capture low-level jet 30 km 2h 200 m (up to 3 km AGL) Hail 0.5-10.0 km 2-30 min Temperature Same as for flash floods Moisture Wind Straight-line 5-10 km wide 10 min-2 h Temperature 1 km 5 min 100 m (up to 12 km) damaging winds 50-300 km long Moisture Wind Hydrometeor mixing ratios

OCR for page 23
Tornadoes 20 m-2 km 1 min-1 h Temperature Pre-storm environment Moisture 50 km 1 h 200 m (up to 6 km) Wind Non-supercell tornadoes (sub-cloud layer) 0.5 km 5 min 100 m (up to 3 km AGL) Downslope 20 km 2-5 H Temperature Pre-storm environment windstorms Along wind Moisture 100 km 3 h 200 m (up to 15 km) 100 km Wind Local variability Across wind 1 km 15 min 100 m (up to 15 km) Pressure- gradient 100-300 km 2-12 H Temperature 100 km 6h 500 m windstorms Moisture Wind Pressure Fire Weather 10-100 km 2 h to 5 days Temperature 1 km 15 min 100 m (up to 5 km) Moisture Wind Insolation In-cloud icing 10-300 km 30 min to 12 h Temperature 5 km 1h 100 m (within any layer where Moisture temperature lies between 0°C and Wind –20°C) Hydrometeor mixing ratios Downburst 100-3000 m 1-10 min Temperature 1 km 1 min 200 m (up to 8 km) Moisture Wind Hydrometeor mixing ratios  continued

OCR for page 23
TABLE 2.1 Continued 0 Resolution Parameters to Phenomenon Size Duration Observe ∆x ∆t ∆z Aircraft turbulence 10-100 m 1-30 s Temperature 1 km 1 min 50 m (where vertical shear is (clear air) Moisture strong) Wind NOTES*: km / m kilometers / meters h / min / s hours / minutes / seconds ∆x / ∆t / ∆z horizontal resolution / temporal frequency / vertical resolution MSL / AGL above mean sea level / above ground level SST sea-surface temperature PBL planetary boundary layer *Sizes and durations listed in the table give a typical range but may not cover extremes. The recommended spacing and frequency of observations should be considered rough estimates, not hard numbers.

OCR for page 23
 OBSERVATIONS SUPPORTING THE FUNDAMENTAL INFRASTRUCTURE • Temperature, moisture, and wind are universally required parameters. (Of the three, moisture by far is the most poorly measured.) For a few critical applications, the concentration of atmospheric aerosols and selected gaseous constituents, and hydrometeor mixing ratios are essential. • Most unmet requirements and the most demanding requirements for observations lie below 5 km altitude. There are several reasons for this: (1) The planetary boundary layer, that part of the atmosphere most respon- sive to surface conditions and the diurnal cycle, is where many mesoscale phenomena have their roots. Its depth seldom exceeds 5 km, except over deserts. (2) Major exchanges of heat, moisture, trace gases, and momen- tum occur near the Earth’s surface. (3) Atmospheric gradients tend to be stronger in the lower troposphere and are heavily influenced by topography. (4) Current observing systems do not sample atmospheric conditions just above the ground as well as they sample surface or upper tropospheric con- ditions. For example, infrared sensors aboard satellites cannot see through clouds, and the horizontal density of in-situ observations is much less in the lower troposphere than at the surface. Importantly, rawinsonde sites are hundreds of kilometers apart, and soundings are taken only once every 12 hours (0000 and 1200 UTC, in most longitudes, neither at the peak nor at the minimum of the boundary layer development). • For smaller and shorter-lived mesoscale phenomena, the recom- mended spatial density and temporal frequency of observations is high. As discussed below, however, the blending of information from observations and models by means of data assimilation allows for some relaxation of requirements. DATA ASSIMILATION: SYNERGY BETWEEN OBSERVATIONS AND PREDICTION MODELS The purpose of data assimilation is to combine in an optimal fashion information gleaned from observations and models (Daley, 1991; Kalnay, 2003; Rabier et al., 2000; Wu et al., 2002). When observations are used to correct a short-term model forecast, the dynamical consistency of models and the temporal continuity between successive model states almost always result in a better subsequent forecast than would have been made in the absence of assimilation. The synergy between observations and the under- standing of atmospheric behavior incorporated in the model equations leads to a more accurate representation of the atmospheric state, especially when the assimilation is repeated frequently (at least every 6 hours), as is the case in operational centers. Does this “assimilation cycle” permit relaxation of some of the requirements for observations? Undoubtedly, but the extent to which this is true has not been carefully investigated. Data assimilation is

OCR for page 23
 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP not a cure all for limitations in data coverage or accuracy. Quality control will always remain an important issue. In certain rapidly changing and dynamically complex situations (e.g., severe convection), models are of limited capability, with or without good observational data as input. Even so, a reduction in spatial density and temporal frequency listed in Table 3.1, each by a factor of two, may be possible and would definitely lower the cost of observations. Here we provide two examples of the synergy between observations and models. In-situ observations of soil moisture are sparse and unevenly distributed over the United States Satellites provide broader coverage, but estimates apply only to surface wetness, and these are degraded in the presence of dense vegetation. In response to these shortcomings, land data assimilation systems (LDASs) have been developed to generate physically realistic soil-moisture profiles for use in initializing computer prediction models (Mitchell et al., 2004). Spinning up LDASs requires frequent input of radiation data from satellites; “Stage-IV” precipitation estimates from the National Oceanic and Atmospheric Administration’s (NOAA’s) Office of Hydrology based on radar reflectivity measurements, rain gauge data, and sometimes satellite information; and model integration. Spin-up time is measured in months, yet the use of LDASs has led to a more complete characterization of soil moisture than could be obtained by the direct measurements alone and, collaterally, to somewhat improved forecasts of convective precipitation in summer. Current data assimilation systems are very sophisticated. They invariably require comprehensive information on observation and model errors and their spatial correlation. Two kinds of error characterize an observation: 1. Measurement error—the error intrinsic to the operation of the instrument. Every instrument samples a particular volume of the medium being measured, whether it is a thermometer mounted in a shelter, a mois- ture sensor rising through the atmosphere on a weather balloon, or a satel- lite measuring radiation in a particular wavelength interval upwelling from the atmosphere. Also associated with each measurement is a time interval over which the instrument responds to the medium being sampled. The difference between the number associated with the measurement and the true value (never known precisely) integrated over the sample volume and the sampling time is the measurement error. Measurement errors are usu- ally estimated by calibrating the field instrument against a more accurate laboratory standard. 2. Representativeness error—not really an error, but a measure of the discrepancy between the space-time dimensions of the measurement and the space-time dimensions that can be captured by the model (related to the gridpoint spacing and the time step). For example, a summer afternoon

OCR for page 23
 OBSERVATIONS SUPPORTING THE FUNDAMENTAL INFRASTRUCTURE thunderstorm may occupy just one-tenth of the area of a model grid square. A surface temperature representative of the entire grid square (what the model is supposed to compute) may disagree by 10°C with an accurate tem- perature measured directly underneath the thunderstorm. If the model had sufficient resolution to predict the thunderstorm explicitly, this discrepancy would be far less. Estimating representativeness error is still more of an art than a science. Thus, information about observation errors can be as important as the value of the measurement itself, because effective data assimilation assigns weights to information from observations and models, taking into account the size of their respective errors. Good metadata—detailed information about the instrument, its expo- sure, calibration, and exact location—is vital for estimating both measure- ment and representativeness error. Metadata receives more emphasis in Chapter 6 of this report. SPECIAL REQUIREMENTS FOR CLIMATE MONITORING To reiterate a point made in the introduction, climate, like weather, has mesoscale variability occasioned by topography and land/ocean surface conditions. That is why climate monitoring cannot be ignored in this dis- cussion; it is one of multiple national applications supported by mesoscale observations. Climate monitoring imposes demanding requirements for absolute accu- racy and long-term stability of measurements. As examples, consider the following trends. Global surface temperature has increased about 0.76°C in the past 100 years; global average sea level has risen 1.8 mm/year from 1961 to 2003 (IPCC, 2007). These changes are small but significant, in that they have already affected many regional ecosystems. Only long-term, stable measurements and considerable averaging enable the detection of such trends. The National Oceanic and Atmospheric Administration is slowly con- structing a U.S. Climate Reference Network (USCRN), with a 2008 target of 114 stations nationwide, whose purpose is to provide future long-term homogeneous observations of temperature and precipitation that can be coupled to past long-term observations for the detection and attribution of climate change (see http://www.ncdc.noaa.gov/oa/climate/uscrn/index. html). A minimum of five parameters is measured at each site: air tempera- ture, precipitation, wind (speed only), ground surface temperature (with an infrared sensor), and hemispheric solar radiation (with a pyranometer). The lack of a wind direction measurement decreases the value of USCRN observations for weather-related applications.

OCR for page 23
 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP The USCRN follows the climate monitoring principles developed by the World Meteorological Organization (WMO) over the past decade for the Global Climate Observing System (GCOS). The WMO promulgated an original set of 10 principles, which apply mainly to surface observations, in 1999. It endorsed 10 additional principles, pertaining to climate moni- toring by satellites, in 2003. (All principles are listed at http://www.wmo. ch/pages/prog/gcos/index.php?name=monitoringprinciples.) The WMO has designated selected sounding sites to be part of a GCOS Upper Air Network (GUAN) to provide (1) long-term, high-quality climate records, (2) anchor points to constrain and calibrate data from more spa- tially dense global networks, including satellites, and (3) where possible, a larger suite of co-related variables such as cloud properties, infrared radia- tion, and trace gas concentrations that have import for climate monitoring. Twelve GUAN sites are in the United States, including three in Alaska and one in Hawaii. (Design principles, accuracy requirements, and a list of best practices are available at http://www.gosic.org/gcos/GUAN-spec.htm.) Because observed climate changes have very likely (Intergovernmental Panel on Climate Change wording) been caused by an increase in green- house gases, it becomes ever more important to expand measurements of the chemical constituents of atmosphere and ocean, especially greenhouse gases and aerosols, not only their concentrations but also their sources and sinks. Of course, such measurements would also support day-to-day air quality monitoring and prediction. By design, climate reference measurements are intended to be especially accurate and stable for the long term. Situated in the midst of a denser network of mesoscale observations from diverse sources, the reference mea- surements serve as control and calibration points. Conversely, the mesoscale observations indicate how larger-scale climate trends are experienced on the regional scale and modulated by characteristics of the lower boundary. The effects of climate change may well have high spatial variability. To be convinced of this, examine Figure 3 on page 4 of Hotter and Drier: The West’s Changed Climate (Saunders et al., 2008). The figure compares the average surface temperature for the lower 48 states, from 2000 to 2007, by climate zone, with the 20th century average. There is considerable spatial variability. MESOSCALE OBSERVATIONS FOR RESEARCH The research community has a multi-faceted relationship to mesoscale observations that serves multiple national needs. It can both draw from and contribute to the broader enterprise, but often in ways that are unlike those of other prospective partners. Appropriately funded operational observa- tions are stable, reliable, and dependable, often at the cost of flexibility and

OCR for page 23
 OBSERVATIONS SUPPORTING THE FUNDAMENTAL INFRASTRUCTURE adaptability and sometimes sensitivity or precision. Research observations are often episodic, ephemeral, and of limited areal extent, and tend to focus on process-level questions in considerable detail. Consequently, research contributions may fail to contribute reliably or consistently to an ongoing operational enterprise and therefore could be viewed as untrustworthy, disruptive, or even parasitic. However, the research community has a long track record of pointing the way to what eventually becomes routine in operational observing sys- tems. What we now view as core components of an operational weather monitoring network typically had their origin in the academic community and/or national research laboratories. Examples include but are not limited to Doppler radar, polarimetric radar, radio-acoustic sounding systems, wind profilers, eye-safe aerosol backscatter lidar, portable automated mesonet systems, use of geostationary satellites as primary data collection platforms, a data transfer system for dropwindsondes, and solid state sensors and digi- tal electronic systems that enabled markedly improved performance of sur- face meteorological stations at the dawn of the digital electronics era. One could view the broader research community as a somewhat autonomous development arm of the atmospheric observation enterprise and a major contributor to its infrastructure, while also recognizing that there are highly directed components, for example, in the NOAA labs. Another role of the research community is that of a user with special needs. Some research needs are so specialized and ephemeral they would constitute an excessive and costly burden on a national network if imple- mented. Examples might include extensive use of mass spectrometry to detect and quantify thousands of constituent compounds associated with reactive chemistry in the atmosphere; large arrays of sonic anemometers at each surface station to fully characterize turbulence and related land- atmosphere fluxes; or a battery of acoustic, optical, and radio frequency profilers at each surface meteorological station. In the foreseeable future, such demands, as useful as they might be, likely would not serve the broader interests well. Nearly all tropospheric and related research programs have use for a mesoscale network to provide a high-quality and minimally aliased sam- pling of the environment. The requirements of the research communities are broadly consistent with those of other users, namely reliability, calibration, documentation, ability to retain at least minimal flexibility and adaptivity, and the general ability to serve broad functions to an acceptably high standard. A core national network, by reason of its permanency, reliability, long periods of record, and geographic extensiveness, offers high value to the research community. If anticipated prior to network implementation, a flexible short list of auxiliary research-motivated sensors could be considered for deploy-

OCR for page 23
 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP ment nationwide, or in regions and seasons of greatest relevance. These systems could be funded by various research interests at a fraction of the cost otherwise incurred if fielded independently. These research-motivated observations are collectively referred to as a national research backbone, which enables other research-based observations to be placed in a properly documented environmental context. In many instances, research-motivated observations point the way forward to future operational network capa- bilities, thereby contributing to the developmental aspect of research com- munity participation. Recommendation: The national network architecture should be suffi- ciently flexible and open to accommodate auxiliary research-motivated observations and educational needs, often for limited periods in limited regions. If history is a proper judge, many of the research-motivated sensors and observations will become operational, serving existing societal needs better and future societal needs well. The impact is likely to be felt at or near the surface and be relevant to both managed and natural terrestrial and marine ecosystems and the heavily built environment. The National Ecosystem Observing Network (NEON), sponsored by the National Science Founda- tion (NSF), is one promising example of beneficial research engagement. A more seamless blending of formal university education with observations, operational forecasting, and research will promote the capacity building required to satisfy personnel needs of the future. The developmental role carried out jointly by the scientific and research engineering communities is pivotal to successful implementation of a national mesoscale network. This is the transition from a research dem- onstration of concept to reliable and hardened operational performance. Often this phase of research and development (R&D) is assumed by a mix of national laboratories and industry. It includes robust design and quasi- operational demonstrations with prototype limited-area networks. Data may be assimilated into limited-area models to evaluate and verify forecast impact. Applied R&D should include but not be limited to transitional activi- ties, including the operation and evaluation of prototype networks and their forecasted impacts; development of tools to facilitate data access for real-time assimilation; development of tools to serve and educate the general public; and exploration of advanced and innovative technologies to serve multiple national needs better, cheaper, and sooner than otherwise might be possible. Testbeds are an appropriate vehicle for these activities (see Chapter 6). Also discussed in later chapters are the concepts for the design of an integrated, nationwide network of networks (NoN) and the

OCR for page 23
 OBSERVATIONS SUPPORTING THE FUNDAMENTAL INFRASTRUCTURE services needed to enable such a network. Here we highlight a few research examples to support a recommendation for the accommodation of auxiliary research observations in a NoN: Research Example : Initiation of Convection in Numerical Prediction Models Weather radars, notably WSR-88D radars, often show quasi-linear boundaries in the clear-air reflectivity display, and Geostationary Operational Environmental Satellites (GOES) visible satellite images show lines of grow- ing cumulus clouds (e.g., Wilson and Schreiber, 1986; Purdom, 1976). Both features indicate boundary-layer convergence, and thus they often mark favored locations for the initiation of deep convection. Short-term field programs have shown that quite small variations in the vertical struc- ture of temperature, moisture, and wind in the boundary layer determine whether a thunderstorm will form (Weckworth, 2000; Sun and Crook, 2001). The crux of the matter is whether low-level air parcels can be brought to their level of free convection. This is true for both stationary and moving boundaries (e.g., gravity currents, bores, and other trapped gravity waves). For more than a decade, committees affiliated with the North Ameri- can Observing System (NAOS) program and the U.S. Weather Research Program (USWRP) have advocated for dense observations in the bound- ary layer. The Atmospheric Radiation Measurement/Cloud and Radiation Testbed (ARM/CART) site comes closest to meeting the requirement for horizontally dense measurements of boundary-layer structure, but even its observations are not dense enough. Radiometric measurements from space do not have the vertical resolution for this application, and infrared mea- surements cannot penetrate clouds in any case. Ground-based remote sensing, perhaps a combination of a profiling Doppler radar, water vapor lidar, radar refractivity from 88D and other radars, and Radio Acoustic Sounding Systems (RASS), seems to have the best prospect for measuring boundary-layer wind, temperature, and mois- ture at sub-kilometer horizontal resolution and 50-m vertical resolution. Such resolution might be feasible in a small research network. Model calculations of heat and moisture fluxes at the surface are not very accurate. Because these fluxes influence the evolution of the boundary layer and ultimately the initiation of convection, their calculation must be improved. That will be difficult without more complete observations of soil moisture and temperature and vegetation fraction (Advanced Very- High-Resolution Radiometer [AVHRR] and MODerate-resolution Imag- ing Spectroradiometer [MODIS] satellite data are relevant for the latter).

OCR for page 23
 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP Such information will improve land-surface parameterization and the flux calculations that depend upon it. The objective of this research is to successfully and consistently simulate convective initiation with a numerical model, having incorporated these detailed observations in the initial conditions. This accomplishment will demonstrate at least the minimal observational requirements for successful prediction of this ubiquitous, disruptive, and often dangerous phenomenon. Research Example : A Soil-Moisture Network to Support Short-Range Climate Variability Modeling Soil-moisture changes are the terrestrial equivalent of sea-surface tem- perature changes in providing memory to the climate system. Koster et al. (2004) have shown that predictability can be enhanced in selected regions of the globe by better understanding of land-atmosphere interactions that are principally governed by soil moisture. The central United States is one of these regions. Improved measurements of soil moisture would improve climate-forecast model initialization and would allow for more accurate simulation of a significant and slowly changing component of the hydro- logical cycle. Better simulation of the soil-moisture reservoir in the growing season will improve the realism of the surface energy budget, which in turn affects gradients in boundary-layer stability and the location and timing of convective processes. Improved simulation of water recycling through evaporation and precipitation would improve the prediction of water leap- frogging across the region during extended periods of weak synoptic forcing and strong convection. Recently refined surface-atmosphere interaction models are capable of more realistically simulating the surface energy, water, and trace gas cycles, but imprecise knowledge of the soil-moisture reservoir is a severe limitation. The Oklahoma mesonet delivers a soil-moisture measurement every 30 minutes in every county of the state. These measurements calibrate satellite-based estimates of surface wetness and verify model estimates of water and energy exchange between the surface and the atmosphere. Other, less dense soil-moisture measurements are taken by the Illinois State Water Survey and the University of Nebraska, but no systematic nationwide or even region wide effort exists to collect and distribute these data. Research Example : Surface Heterogeneity and Its Impact on Boundary- Layer Structure and Convective Precipitation Although weather prediction has improved, the prediction of warm season convective precipitation has lagged behind. One of the suspected

OCR for page 23
 OBSERVATIONS SUPPORTING THE FUNDAMENTAL INFRASTRUCTURE reasons for this is the influence of vegetation, soil-moisture, and topogra- phy on sensible heating and moistening of the atmospheric boundary layer (ABL). Horizontal variability of surface properties results in horizontal variability in buoyancy fluxes. Buoyancy fluxes and their horizontal vari- ability influence the growth of the ABL through development of thermally direct mesoscale circulations. Long-term surface, subsurface, and boundary layer observations are needed because of the seasonal variability of thermally direct circulations. For example, in Kansas, C3 winter wheat is green in the months of April and May, after which it senesces, and it is harvested by mid-June. On the other hand, the mixture of native grasses starts greening up in May and becomes lush and green by June. This means that the crops and grasses reverse roles, with the largest sensible heat flux over the grasses in spring and the largest sensible heat fluxes over the harvested wheat fields in the summer. If elevated areas are associated with grasses (as they are in Kansas), the “elevated heat source” effect of ridges reinforces the extra heating in the spring, thus increasing the likelihood of topographically influenced circulations. What kind of observations would enable better prediction of thermally direct circulations? A network of soil temperature and moisture probes, deployed long-term at the mesoscale. Add to that radar wind profilers and lidar water vapor profiles (see Chapter 4) to track the evolution of the planetary boundary layer and to compute vertical air motions, the local level of free convection, and convective available potential energy. These observations permit the documentation of boundary-layer depth and, with knowledge of ambient synoptic conditions, the likelihood and strength of deep moist convection. A good estimate of the horizontal variation of precipitation would be obtainable from network Doppler and polarimetric radars when assimilated with network precipitation gauge data. The unique contributions of a national mesoscale network would be long-term regional scale observations of soil moisture and temperature, profiles of winds and divergence fields, water vapor distribution, and gauge/ radar coverage of precipitation. In combination with research observing sys- tems such as instrumented aircraft and flux towers, considerable advances in the prediction of warm season convective precipitation would be real- ized. A complementary strategy is to site research flux towers so that differ- ent types of land cover, soil types, and topographical settings are sampled at several points for relatively long periods. Long-term observations, of the type well-suited to a national mesoscale network, are needed in order to obtain a representative period of record, with and without thermally driven mesoscale circulations.

OCR for page 23
0 OBSERVING WEATHER AND CLIMATE FROM THE GROUND UP Research Example : Improving Chemical Weather Predictions Predicting chemical weather is of growing importance to society. Chemical transport models have become essential tools for providing science-based input into best alternatives for reducing urban pollution levels, designing cost-effective emission control strategies, siting of facili- ties, interpreting observational data, and assessing how we have altered the chemistry of the global environment. The forecasting of chemical weather has become a new application area, providing important information to the public, decision makers, and researchers. National weather services throughout the world are broadening their traditional role of mesoscale weather prediction to also include prediction of other environmental phenomena (e.g., plumes from biomass burning, volcanic eruptions, dust storms, and urban air pollution) that could potentially affect the health and welfare of the public. Currently hundreds of cities worldwide are providing real-time air quality forecasts. While chemical weather prediction and weather prediction are closely aligned, there are important differences between the two. One important difference is that weather prediction is typically focused on severe, adverse weather conditions (e.g., storms), while the meteorology of adverse air quality conditions frequently is associated with benign weather. Boundary-layer structure and wind direction are perhaps the two most poorly determined meteorological variables for chemical weather prediction. Meteorological observations are critical to effectively predict- ing air quality, yet meteorological observing systems are typically designed to support prediction of severe weather and not the subtleties of adverse air quality. Research needs associated with the meteorological elements of air quality prediction have recently been assessed (Dabberdt et al., 2004a). The additional processes associated with emissions, chemical transforma- tions, and removal also differentiate chemical weather prediction from weather forecasting. Because many important pollutants (e.g., ozone and fine particulate sulfate) are secondary in nature (i.e., formed via chemical reactions in the atmosphere), chemical weather models must include a rich description of the photochemical oxidant cycle. It is also important to note that the chemical and removal processes are highly coupled to meteo- rological variables (e.g., temperature and water vapor), as are many of the emission terms. In the case of windblown soils, emission rates correlate with surface winds, and evaporative emissions correlate with temperature. In the case of emissions associated with heating and air conditioning, the demand responds to ambient temperature. The chemical observation networks were designed to support compli- ance and regulatory functions and not prediction. Chemical weather predic- tion places greater emphasis on real-time access to data, with broader spatial

OCR for page 23
 OBSERVATIONS SUPPORTING THE FUNDAMENTAL INFRASTRUCTURE coverage and vertical extent for initializing prediction models. Advances in our predictive capabilities will require a better matching of the observa- tional capabilities with chemical weather prediction needs (Carmichael et al., 2008). The mesoscale observing system outlined in this report will provide the backbone of data that will enable and accelerate the field of chemical weather prediction. One important research activity will be the use of chemical data assimi- lation systems to help design the observing systems needed to produce better forecasts. We need to rigorously quantify the value added to a forecast by adding observations of additional species and above the surface, extending surface coverage; and adding and enhancing the utility of observations from satellites for chemical weather applications. Research Example : Assimilation of Mesoscale Observations into Prediction Models Only a small percentage of satellite observations is assimilated into prediction models. Debate continues about how to treat measurements of upwelling radiation. Preliminary studies at the UK Met Office have shown that soundings derived from the Atmospheric Infrared Sounder and the Infrared Atmospheric Sounding Interferometer, when inserted into com- puter prediction models, have had a greater impact on numerical predic- tions than the direct assimilation of radiance from those sensors, contrary to prevailing opinion. The reason may be that the radiance data are thinned both spatially and spectrally, whereas the derived soundings use all of the spectral information. Only further experimentation will resolve this issue. The use in prediction models of cloud and hydrometeor information from satellites, surface-based ceiling observations, aircraft observations of clouds and icing, radar reflectivity, and lightning data is still primitive. More sophisticated assimilation techniques are sorely needed. Correct specification of the statistical structure of model forecast errors is required for optimal performance of three- or four-dimensional varia- tional data assimilation, especially the spatial covariance of model errors. The direct approach to this problem relies on an extensive network of dense observations for the direct calculation of differences between fore- cast and observed values and their means, standard deviations, and spatial covariances. Another approach is to estimate situation-dependent model errors by means of ensemble forecasts. Further study is needed on how uncertainty in the initial state (primar- ily due to the sparsity of observations with respect to the grid resolution of today’s operational prediction models) translates into uncertainty in the model forecast.