Emerging Weather Research and Transitional Needs
Several research and transitional needs have come to be recognized in the United States as increasingly important but are only in the early stages of understanding or implementation. The committee refers to these as emerging weather research and transitional needs—in contrast to the established needs discussed in Chapter 3. Three high-priority emerging needs were identified in the 2009 BASC Summer Study workshop and subsequent committee meetings. The three emerging needs discussed here include very high impact (VHI) weather, urban meteorology, and renewable energy production.
The reader may wonder why VHI weather is included here as an emerging need rather than in the preceding chapter as one of several established needs. The answer lies in the emphasis here on impact forecasting rather than the traditional focus on weather prediction per se. Urban meteorology was recognized in the United States in the 1960s as an important topic, and much seminal urban meteorological research was conducted until the early 1980s when it was abruptly deemphasized as a research priority. A reemphasis on the meteorology of the urban zone and its societal import began again in the 1990s and continues today. In contrast, Europe has focused steadily on urban issues for many decades, as has Japan. Lastly, the meteorological challenges associated with the special needs of the renewable energy industry have come into sharp focus over the past 5 or so years. In all three emerging areas, much remains to be done. As mentioned previously, virtually all research and R2O needs have both established and emerging aspects, and so many of the challenges and needs cited in Chapter 3 are relevant as well, and are closely coupled to those discussed here in Chapter 4.
VERY HIGH IMPACT WEATHER
Weather-related disasters result in loss of life and disruption of communities as well as billions of dollars in damages in the United States an-
nually. Between 1980 and 2009 there were 96 major disasters caused by VHI weather events that resulted in losses exceeding 1 billion dollars each (NCDC, 2010). It goes almost without saying that there is a great need for accurate forecasts and warnings of severe, hazardous, and disruptive weather conditions so that the resulting economic and societal impacts can be minimized.
VHI weather can be defined as weather that endangers public health and safety or causes significant economic impacts. VHI weather generally falls into two categories:
severe and disruptive weather hazards—including tropical storms and hurricane-induced extreme winds, rain, and storm surges; severe thunder-storms and tornadoes; lightning; flash floods; ice and snow storms; dense fog; and wildfires—which change rapidly on the timescale of minutes to hours or a few days; and
persistent weather hazards—including long-lasting heat/cold waves; drought; and flooding due to persistent rain events—which occur on longer timescales of days to weeks or even years (e.g., drought).
Advancing the understanding, monitoring, and prediction of VHI phenomena requires improving the accuracy and timeliness of observations, forecasts, and warnings in order to develop an efficient response system that helps minimize and mitigate the impacts of hazardous weather. An expansion in emphasis from weather prediction alone to the prediction of weather and related impacts is warranted. This would necessitate development of new modeling and observational tools, innovative forecast guidance products, and methods of information and warning dissemination to decision makers and stakeholders. Accordingly, new research and R2O priorities for VHI weather need to be established. Also required is the close collaboration of physical and social scientists in setting priorities and developing effective research and implementation programs. Social scientists, especially, will also play a critical role in developing O2R needs and priorities. To facilitate a rapid R2O transition, it is critical to train a new generation of researchers, forecasters, and decision makers in the need for, and use of, a fully integrated forecast and response system.
Current State of Affairs and New Opportunities
Weather research over the past several decades has led to many advances in monitoring, understanding, and predicting VHI weather, which
have contributed to major improvements in forecasts and warnings such as more accurate hurricane tracks and longer lead times for tornadoes and severe thunderstorms. Although the number of VHI weather phenomena is extensive, this section aims to illustrate the variety of phenomena that have major impacts on society and identify emerging needs and opportunities for weather research to better serve critical societal needs. A number of previous studies (Table 1.1) by the U.S. Weather Research Program (USWRP; e.g., Emanuel et al., 1995) and the National Research Council (e.g., NRC, 1998b) can serve as benchmarks for comprehensive reviews, because they identified many pressing needs and opportunities for atmospheric, hydrologic, and related research and development that still exist today.
A New Impacts Paradigm
The atmospheric community has for many years worked diligently to improve the accuracy and resolution in space and time of the raw quantities predicted by numerical models, such as temperature, humidity, wind, and precipitation. Statistical techniques have been used to predict additional quantities and to introduce probability of precipitation and other derived forecast parameters. With some exceptions, users have largely taken these weather predictions and used them in their own decision support and risk management processes. However, this approach has not always produced the desired or optimal outcome, especially when complex weather forecasts are difficult to understand and yet require public action in response to the forecast. For instance, probabilistic forecasts of a landfalling hurricane’s track and intensity, without specific impact information such as timing and location of storm surge, extent of flooding, extreme winds, and power outages, are insufficient for effective responses from emergency managers.
A new paradigm for the coming decades is for end users and scientists (both physical and social scientists) to work together toward also providing improved, explicit impact forecasts as well as advancing human comprehension of complex information. The paradigm shift from forecasting weather to forecasting weather and impacts will challenge the traditional weather forecasting approach and demand a full integration of the physical sciences with the socioeconomic sciences that is relevant to weather impacts and societal and environmental responses. Because of the implications for public safety and economic resilience, VHI weather phenomena are key targets for such integrative research and the transition of research results to operations.
As one example, consider the information in Figure 4.1, comparing traditional portrayals of weather forecasting, and the potential for impacts
forecasting. Traditionally, data from surface-based observing systems, reconnaissance aircraft, and satellites are used in numerical weather prediction (NWP) models to make predictions about a hurricane’s trajectory, intensity (wind speed), and precipitation. In the new, impacts forecasting paradigm, these models would be used in conjunction with decision support models to yield projections of possible impacts such as the extent of power outages and the time to power restoration for the region affected by the hurricane. In fact, there are private weather and risk management companies that are now working with electric utilities, insurance and reinsurance companies, and others to make industry-specific impact predictions for a variety of severe weather events. However, impact prediction remains to be implemented
more widely—not only by the private sector but also by the public sector with its responsibility to support and protect public health and safety.
One key component of, and a major challenge for, the prediction of impacts is to more fully exploit the capabilities of ensemble modeling of the atmosphere to produce probabilistic forecasts of atmospheric quantities, and for these to then be used to generate probabilistic forecasts of the impacts and risks of pending VHI weather situations, thereby enabling improved decision making. Rather than the meteorological community and the end-user communities working separately, teams of atmospheric scientists, social scientists, and professionals from user groups1 need to work together to define the needed observations and the desired predicted impact parameters. This approach has been recommended for hydrometeorological forecasting (Krzysztofowicz, 1998) in a seminal paper on the development and application of joint decision-making probabilities of river stage predictions and user risk tolerances.
Severe and Disruptive Weather Hazards
Hurricanes and Tropical Storms
Although there has been significant improvement in hurricane track forecasts, progress has been minimal with regard to storm intensity forecasts. The improvement in track forecasts is largely attributed to the advancement in satellite and dropwindsonde observations (Franklin et al., 2003) over the oceans and model improvement and data assimilation in global models over the past few decades. Limiting factors for hurricane intensity forecasts include the lack of understanding of rapid changes in storm structure and intensity, routinely (and continuously) available in situ observations, and high-resolution coupled air–sea–land models (Chen et al., 2007) in operational centers. Although some issues were identified by PDT–5 (Marks and Shay, 1998), many questions and problems remain unresolved. Other new issues have emerged since then.
Major landfalling hurricanes between 2004 and 2008, such as Katrina, Rita, and Ike, revealed many critical needs not only for improved weather forecasts but, more importantly, for forecasts of storm impact directly related to societal responses to these events. These are highlighted in four recent national reports calling for action to substantially improve hurricane forecasts (AGU, 2006; NSAB, 2006; NSB, 2007; OFCM, 2007); they particularly cite
the rapid intensity changes of hurricanes threatening the United States as a major challenge. Recent advances in science and technology, especially in high-resolution coupled modeling, ensemble model forecasting, high-performance computing, and social behavioral studies related to hazardous weather events, have presented a great opportunity to develop a strategy and action plan for an integrated forecast-and-response system that will support risk assessment, emergency management, and decision making.
Tornadoes and Severe Thunderstorms
There has been considerable improvement in understanding, predicting, and warning for these hazardous phenomena as a consequence of successful research programs, deployment of a national network of Doppler radars, and other National Weather Service (NWS) modernization activities in the 1990s. Lead time for tornado warnings has increased from about 6 minutes in 1993 to about 13 minutes in 2008. However, questions still remain concerning why only a small fraction of supercell thunderstorms produce tornadoes while most do not. As a consequence, the false alarm rate for tornado warnings is high, at 75 percent in 2008, which is virtually unchanged from 1993 when it was 73 percent. Research programs such as VORTEX2 need to continue to address that question, reduce false alarm rates, better understand tornado genesis and dissipation processes, deduce how and why some tornadoes become strong and violent while others do not, and explore other unknowns about tornadoes and severe thunderstorms. Radars in the national operational NEXRAD (or WSR-88D) network are spaced too far apart to detect the low-level portions of most supercells, which is critical to detection of tornadoes.
Research needs to continue the development of low-cost, adaptive scanning radars as a means to fill these gaps (e.g., Brotzge et al., 2006) both to reduce false alarm rates and improve detection of tornadoes now missed. Recent studies by the NRC (e.g., 2002, 2008a) recommend additional upgrades to the national radar network. To dramatically improve tornado warning lead time will likely require a shift away from warnings based on detection to warnings based on forecasts. Much needed research is ongoing toward the development of a warn-on-forecast system (e.g., Stensrud et al., 2009), which will involve many of the improvements in numerical modeling and probabilistic forecasts recommended here and elsewhere in this report. There is also a continuing need for improved understanding of the four-dimensional structure of tornadoes, with practical applications such as improved building construction standards.
In many ways, flash floods are among the most difficult phenomena to predict (see the discussions in Chapter 3). Many issues related to quantitative precipitation forecasting and hydrologic and flood prediction have been discussed previously by USWRP PDT–8 (Fritsch et al., 1998) and USWRP PDT–9 (Droegemeier et al., 2000). There is little skill in predicting the exact location of an upcoming flash flood until the rain is well under way. Even then, the area of heaviest rainfall tends to be small and not always well observed. Improvements in precipitation estimation through use of multiparameter radars will help with this problem; the NEXRAD agency partners (Department of Commerce, Department of Defense [DOD], and the Department of Transportation [DOT]) have initiated a program to provide dual-polarization capability on all 166 WSR-88D radars with completion targeted for early 2013.2 Improvements in satellite-based precipitation estimation and in cloud-to-ground lightning detection (a useful surrogate for convective precipitation) can help in remote, mountainous areas not well sampled by radar.
Flash floods are an excellent example of VHI weather that can reap the benefits of a new impacts forecasting paradigm by utilizing numerical model forecasts and observations in a coupled hydrology–land–atmosphere model for flood impacts forecasting. The need for a coupled distributed hydrologic modeling framework is discussed at length in Chapter 3. A recent flash flood in the Atlanta, Georgia metro area illustrates the challenging nature of flash floods. On September 20–21, 2009, more than 15 inches of rain in less than 24 hours (with a maximum of more than 21 inches in 38 hours) fell in a narrow corridor generally less than a county wide. Although a broad area of the Southeast was under a flash flood watch owing to the presence of an unusually moist air mass, there was no obvious way to predict the exact location or magnitude of the event, even in hindsight. While a mesoscale boundary was oriented generally west–east through the area, a band of thunderstorms set up and moved roughly perpendicular to the boundary. However, all of this was unresolved by the operational regional models, and forcing for the location of formation of the initial storm was not obvious either from observed data or numerical model guidance. It remains to be determined whether a network of enhanced observations providing inputs to improved data assimilation techniques and cloud-resolving regional models can improve forecasts of heavy mesoscale precipitation. This issue
is further addressed in the Mesoscale Observational Needs and QPE/QPF sections of Chapter 3.
Wildfires are another opportunity to apply a new paradigm for impacts forecasting. Temperature, humidity, and dry lightning can play a role in wildfire initiation, development, and spread, while winds and terrain typically play key roles in spreading major wildfires. The wildfires themselves often develop their own weather, becoming firestorms. There is a need for continued improvements in satellite sensing, including fuel availability and the detection and monitoring of fires and their intensities. Fire models can be coupled with atmospheric and land-use models to generate impact forecasts of threatened areas (e.g., Clark et al., 2004). Numerical model improvements are crucial, including terrain and urban effects. A prediction system of this type has been proposed (e.g., Bradley et al., 1999).
Surface and Air Transportation
According to a recent NRC (2004) study, weather significantly affects the safety and capacity of the nation’s roadways. Adverse weather is associated with over 1.5 million vehicular accidents each year, accounting for approximately 800,000 injuries and 7,400 fatalities (FHA, 2009). Poor road or visibility conditions often cause drivers to slow down, thereby substantially reducing roadway capacity, increasing travel times, and in some cases contributing to chain-reaction accidents. It is estimated that drivers endure over 500 million hours of delay annually on the nation’s highways and principal arterial roads because of fog, snow, and ice. This conservative estimate does not account for considerable delay due to rain and wet pavement. An improved strategy for addressing the impacts of weather on surface transportation has the potential to help mitigate roadway congestion and save lives. High-quality weather observations and forecasts specific to the roadway environment could help users make better decisions, thereby increasing travel efficiency and safety during adverse weather conditions. Improved road weather information could also help those who construct, operate, and maintain the roadways to better respond to weather problems.
Weather also is a major factor in causing delays (about 65 to 70 percent of the total occurrence) and economic losses to commercial aviation. According to the Joint Economic Committee of Congress (U.S. Congress, 2008), air traffic delays in 2007 cost airlines $19 billion in increased operat-
ing costs and the United States economy $41 billion. As the volume of air transportation increases, the demand for even greater efficiency will require improved quality and use of weather information. NextGen3 is a multi-agency (DOT, DOD, Federal Aviation Administration, National Aeronautics and Space Administration [NASA], and White House Office of Science and Technology Policy) initiative to dramatically improve the management of air transportation by 2025. Weather information plays an important role in NextGen, enabling the identification of where and when aircraft can and cannot fly. In building NextGen, weather information is being designed to integrate with, and support, decision-oriented automation capabilities and human decision-making processes. Weather information supports trajectory-based planning and decision making. The NextGen weather and automation capabilities will support cataloging and analyzing flight plans and provide recommended routes to pilots and dispatchers. Weather information in the form of meteorological variables that are observed or forecasted (e.g., storm intensity, echo tops) need to be translated into information that is directly relevant to NextGen users and service providers, such as the likelihood of a flight deviation, airspace permeability, and capacity. Uncertainty in meteorological phenomena that have significant impact on air system capacity is being managed through the use of probabilistic forecasts that will include the three-dimensional location, timing, intensity, and the probability of all possible outcomes.
Persistent Hazardous Conditions
These VHI weather phenomena generally occur as a result of prolonged anomalous weather conditions lasting days, weeks, or even years. Impacts can result from either excess or deficient precipitation, anomalously warm or cold temperatures, and often in combination with anomalies of wind and sunshine.
Although sometimes generated by a single rainfall event, the worst floods are often a combination of prolonged rain and snowmelt, and occur over a longer timescale than flash floods, such as the Midwest floods of 1993, which are thought to be the largest and most significant recorded
flood in the United States. Flooding occurred across nine states, resulting in 48 deaths and approximately $21 billion in damages (NCDC, 2010).
Improvements in the ability to predict anomalously wet patterns could have an impact on the ability to anticipate river floods and allow emergency and water managers to plan ahead.
Drought has huge implications for agriculture, water supply, recreational industries, and various commercial enterprises. Shifts in population to low-precipitation areas where water is normally in short supply—and to urban areas where supplies may be limited—have already brought water supply problems, water restrictions, and disputes to many places in the United States. A recent workshop has summarized some of the issues and research needs facing prediction of drought on seasonal to decadal timescales (Schubert et al., 2007). Research needs range from a greater understanding of the forcing factors (oceanic, El Niño–Southern Oscillation [ENSO], aerosol feedback, vegetation, and others), especially with respect to various climate change scenarios, to improved numerical modeling of the coupled land–ocean–atmosphere system.
Heat and Cold Waves
Each of these, typically lasting up to a few days, can have serious implications for human health, agriculture, and other industries. Hundreds to thousands of fatalities (CDC, 2006) result annually from heat waves in the United States alone, including about 700 excess deaths during the 1995 heat wave in Chicago alone. There is also a strong correlation between heat waves and pollutant levels, and the synergistic effects of heat and poor air quality lead to elevated morbidity levels. Because heat waves with the largest impacts last for many days, there is a pressing need to improve medium- to extended-range forecasting. Temperature anomalies have huge economic implications for energy use and for commercial utility providers. For example, urban electricity usage increases 3 to 5 percent for each 1°C increase in ambient air temperature above about 22°C (Sailor and Dietsch, 2007).
Air quality and its impacts on health and the economy involve far more than just the pollutants emitted into the atmosphere. Weather factors such
as inversions, wind speed and stability, precipitation, and other factors are often controlling influences. Air stagnation episodes often involve a pattern of strong, low-level inversion and light winds that persists for a few days. One study (Schwartz and Dockery, 1992) indicated that 60,000 people die in the United States each year because of poor air quality.
Improvements in Impacts Forecasting
The impacts of VHI weather episodes tend to be maximized in urbanized areas where large numbers of citizens and infrastructure are concentrated. Even for the smallest of the VHI weather phenomena (e.g., tornadoes and flash floods) significant portions of an urbanized area can be seriously impacted. Further issues specific to the urban environment are addressed in following sections.
Research that leads to improved understanding, monitoring, prediction, and communication of VHI weather phenomena will result in fewer injuries and fatalities and reduce the impacts of the tens of major natural disasters that annually impact the United States. In addition to the economic impacts of VHI weather already described, there are countless applications in which improved weather information can result in enhanced cost-effectiveness and savings for public entities, business, and industry. Social scientists and economists are needed to help quantify the benefits of weather data and forecasts to these applications and to define key impact parameters (as distinct from traditional weather parameters).
Impacts of Climate Change on Very High Impact Weather
A major challenge is to understand the effects of climate change on VHI weather and its potential long-term socioeconomic implications. Potential changes in storm tracks and intensity, and the frequency of severe drought and flooding events are of great value in risk assessment, adaptation, and mitigation. Although recent studies have suggested that the intensity and perhaps even the number of extreme weather events may increase (e.g., Anthes et al., 2006; Knutson et al., 2010; Trapp et al., 2007; Trenberth et al., 2003), it is difficult to evaluate and validate the results because of the lack of observations and limitations in climate models to represent “weather.” The relatively low resolution and insufficient physical representation in the current climate models have led to great uncertainty in the assessment of the impact of climate change on VHI weather. A well-designed systematic approach is urgently needed to improve climate model physical param-
eterizations (in consequence of increases in model grid resolution), model prediction of VHI weather statistics, downscaling of local impacts of changing climate using high-resolution cloud-resolving and impact models, and rigorous model verification with observations;4 this would require
fundamental research to improve understanding of VHI weather phenomena, especially how underpinning dynamical and physical processes are impacted by larger scale forcings;
better understanding of predictability and predictive skills of VHI weather on shorter timescales as well as weather statistics on longer timescales (e.g., the persistent large-scale flow patterns that produce drought and flooding events); and
development of a seamless, integrated weather–impact prediction system from global to regional and local scales, which can be used for risk and benefit assessment and decision making.
To improve predictive skills, research is needed to better understand and identify the “sources” of predictability for various VHI weather phenomena. For instance, improving severe weather forecasts requires very high resolution cloud-resolving models with improved data assimilation techniques on short timescales of hours or less. On the other hand, improving forecasts for heat waves and prolonged cold outbreaks as well as flooding from persistent rain events requires better global model prediction on extended timescales beyond a few weeks. For the latter, monitoring and data assimilation of soil moisture may be especially critical. Prediction of drought extends from seasonal to interannual and decadal timescales (Schubert et al., 2007). The use of ensemble model prediction for probabilistic forecasting requires new systematic verification methods with quantitative uncertainty estimates. These priority research needs for improved weather prediction on all scales are common to those presented in Chapter 3. Beyond that, research is needed to develop, test, and verify impact predictions for multiple applications.
Recommendation: The federal agencies and their state and local government partners, along with private-sector partners, should place high priority on providing not only improved weather forecasts but also explicit impact forecasts. An effective integrated weather–impacts prediction system should utilize high-quality and high-resolution meteorological
Refer also to the discussion in the modeling and observations sections in Chapter 3.
analysis and forecast information as part of coupled prediction systems for VHI weather situations.
This will require
fundamental research in both the physical and social sciences to improve understanding and prediction of VHI weather phenomena, and the provision of warnings and risk assessments in support of decision making;
development of impact parameters and representations for multiple applications (e.g., morbidity, electric grid vulnerability, surge and flood inundation areas);
research to determine and obtain critical and timely observations;
end-to-end participation by multiple sectors and disciplines (including modelers, observationalists, forecasters, social scientists, and end users) to jointly design and implement impacts-forecasting systems; and
multidisciplinary undergraduate and graduate programs that can address the emerging field of VHI weather–impacts prediction, risk assessment and management, and communication through fully integrated research, education, and training for the new generation of scientists, forecasters, emergency managers, and decision makers.
A key aspect of the VHI weather impact recommendation is to encourage diverse government agencies, academia, and the private sector to work together in defining and addressing problems in which meteorological information—in current or future improved fashion—can be used as part of impacts forecasting. In many cases, this will involve use of weather information in coupled models or as data for input to specific impact models. Examples include the prediction of wind, rain, storm surge, and inland inundation from hurricanes, coupled with detailed topographic, land-use, and population mappings, to delineate in a probabilistic manner which locations will be most impacted. This will make it possible for emergency managers and the public to make more effective decisions regarding hurricane evacuations, and for utility companies and disaster recovery organizations to better anticipate the scope of the relief and recovery efforts likely to be necessary.
Moving forward, there needs to be recognition of the extensive efforts of the private sector in predicting weather-related impacts and assisting in
decision making and risk management for their clients. Obvious examples include preparation for storms, crop forecasts, energy management and trading, airline operations, ship ocean routing and port operations, and recreational enterprises, to name but a few. Some years ago, NWS abandoned impacts forecasting because it could not serve some industries directly while ignoring others, while needing to focus on its primary goal of promoting public health and safety. The current NWS strategy,5 stated simply (perhaps overly so), is to obtain the observations and provide the forecasts necessary for the protection of life and property and then make that information readily available to the public and the private sector to serve their special needs, including prediction of impacts. However, impact forecasts are needed as much in the domain of the NWS and other public agencies as they are needed in the domain served by the private sector. Some examples of impact forecasts that fall within the purview of the public sector include heat stress (in contrast with temperature and humidity), respiratory stress (from the synergies of elevated air pollution levels and temperatures), wind chill (temperature, humidity, and wind speed), and so forth.
To achieve these goals, a mechanism is needed to encourage communication between the meteorological community and those involved in decision making and other types of impact modeling. This would involve socioeconomic scientists and end users as full participants and partners. It may involve creation of an integrated weather–impact modeling testbed or other such mechanism. A great deal of cross-cultural education is needed for each community (meteorologists and impact specialists) to become familiar with the terminology, capabilities, and needs of the partner group(s).
Although the path from research to operations will likely differ somewhat from one VHI weather phenomenon to another, a general methodology to use in implementing the new impacts-forecasting paradigm includes
Interactions among data/forecast providers, users, and social scientists to identify the user data needs in order to define the nature of the end product impact forecasts.
Research and development by the data provider and user communities to complete their components of the impacts forecasting system, and by social scientists to help tailor the output of the impacts forecasting system for human end users. When the end users are the general public, social scientists also need to be engaged to help design educational materials and programs to prepare the public to understand and use the new impact forecasts.
The NWS Strategic Plan for 2005–2010 is available at http://www.weather.gov/sp/; the updated strategic plan for 2010–2015 is expected to be available in the third quarter of 2010.
Testing of the impacts forecasting system and training personnel in its use. Depending upon the nature of the VHI weather phenomenon, this type of activity could be performed in one of the existing testbeds, such as the Hazardous Weather Testbed or the Hurricane Testbed or the urban testbed recommended later in this report.
When the public is the ultimate end user of the output of the impacts forecasting system, an education and training campaign could be launched to help people understand and appropriately respond to the impact forecasts.
VHI weather phenomena have a significant impact on health and safety and economic vitality in the United States and worldwide. Priority needs to be given to improving understanding and prediction of these phenomena, particularly toward developing and implementing impact-prediction systems in order to meet critical societal needs.
Since the end of the Second World War, urbanization of the world’s population has given rise to more than 400 cities around the world with populations in excess of 1 million and more than 25 so-called megacities with populations of over 10 million (e.g., Figure 4.2; Brinkhoff, 2010; Pearce, 2006). Some metropolitan regions now contain between 20 million and 30 million people, including Tokyo, Japan (34.0 M); Mumbai, India (22.8 M); Seoul-Incheon, Republic of Korea (24.2 M); and Mexico City, Mexico (23.4 M. Even greater Los Angeles (17.9 M), which now stretches from Goleta in the north nearly to Ensenada (Mexico) in the Baja, is nearly a virtual “super-megacity.” Recent reports published by the United Nations estimate that in 2007, 50 percent of the world’s population, and more than 75 percent of the population in developed countries, lived in cities (UN, 2008). All indications are that urbanization will continue through much of the 21st century, though the rate of growth of urban populations may slow somewhat from what was seen in the last decades of the 20th century (Brockerhoff, 1999, 2000).
Urban meteorology is the study of the physics, dynamics, and chemistry of the interactions of the Earth’s atmosphere and the urban built environment, and the provision of meteorological services to the populations and institutions of metropolitan areas (usually divided into Metropolitan Statistical Areas [MSAs] and Micropolitan Statistical Areas [μSAs], see Box 4.1
for a brief discussion on what is considered an “urban area,” at least in the United States). Because one of the goals of applied meteorology is to provide services to society, urban regions where people are highly concentrated merit special attention. Urban populations can benefit greatly from a wide range of weather and climate services tailored to their urban environments. Although the details of such services are dependent on the location, geomorphology, and synoptic climatology of a particular city, there are common themes, such as enhancing quality of life and responding to emergencies, which are relevant to many cities. The urban landscape, with its distinct patterns of surface roughness and fluxes of heat, moisture, and pollutants, presents significant challenges to researchers and operational meteorologists. Urban meteorology can benefit from implementation of many of the other recommendations in this report, such as the development of high-resolution mesoscale networks and weather prediction models, hydrologic prediction models, impacts predictions and socioeconomic analyses (see Chapters 2, 3, and 4), which could then be further refined and tailored to the urban landscape.
Identifying appropriate boundaries for a metropolitan area is difficult. In the United States, the federal government has formally defined Metropolitan Statistical Areas (MSAs; see figure below). These regions are composed of counties or equivalents. MSAs are delineated on the basis of a central urbanized area, which is a contiguous area of relatively high population density with a population greater than 50,000. The counties containing the urbanized core are known as the central counties of the MSA; surrounding or outlying counties are included in the MSA if they have strong social and economic ties to the central counties as measured by commuting and employment. As of 1 July 2009, using data developed by the U.S. Census Bureau (2009), the federal Office of Management and Budget recognized 366 MSAs in the United States. Some of the largest MSAs are subdivided into Metropolitan Divisions. These contained about 233 million people in the 2000 census.
Since 2003, the U.S. government has also identified Micropolitan Statistical Areas (μSAs); these are areas centered on a small city with a population in the range 10,000 to 49,999 (Figure 4.3). The area is again based on counties. While individual micropolitan areas do not have the economic or political impact of MSAs, collectively they contribute significantly to the national statistics for population (~30M) and economic activity because of their large number (560 based on 2000 census data). Frequently μSAs have relatively low labor and land costs, and so several have developed surrounding regions of urban sprawl. Because the designation as a μSA is based on the core city, in some cases μSAs are actually more populous than some MSAs. (Note that based on the 2000 census, only ~45 million out of an estimated 308.5 million individuals in the United States did not live in either an MSA or a μSA.)
The United States has a population of more than 308,500,000.6 It is largely an urban population, with about 81 percent residing in MSAs or μSAs as of mid-2005; the equivalent worldwide urban rate was 49 percent (UN, 2008). Cities and suburbs are home to the large majority of the U.S. population, yet they only occupy between 2 and 3 percent of the land area of the United States.
These urban dwellers and their supporting institutions and infrastructure have needs for tailored weather information and services that differ from those living in rural areas. The most notable shortcoming of current urban meteorological services is the lack of spatial resolution. Current observing and prediction systems are structured to provide more or less uniform services across the whole United States. Observations are more or less uniformly distributed, particularly where terrain is not an obstacle, on scales of tens to a few hundred kilometers. Further, even with the recent shift by the NWS to a digital gridded forecast, the current resolution is still coarse, with only a few points per county. The current observing and prediction systems simply do not provide the level of resolution and surface specificity necessary for the production of meteorological products and services on the urban scales, which range from a few meters to a kilometer or so at most. An example is winds at street level—observing and predicting these can be both a quality-of-life service for those walking in the city and a safety and security issue for emergency responders dealing with dispersion of chemical, biological, or radioactive agents.
The urban environment also merits special products and services dealing with air quality (including pollutants related to respiratory stress, heat and cold stress on humans and infrastructure, and monitoring of conditions on the transportation networks—light rail as well as roads—that tie the urban complex together.
Further, because of the complexity of the urban environment, meteorological support needs to be part of an integrated or multihazard warning system that considers the full range of environmental challenges and provides a unified response from municipal leaders. The World Meteorological Organization (WMO) has responded to this need for a comprehensive response to high-impact weather and weather-related events with its Multi-Hazard Early Warning System (MHEWS)7 initiative, as exemplified by pilot projects under way in France and Shanghai, China. Examples of urban weather that such a system would consider include winter hazards, such as snow and ice; flash flooding; sand and dust storms; extended periods of extreme temperatures;
adverse air quality; tropical cyclones with attendant high winds, storm surge, heavy rain, and flooding; and drought with attendant water shortages.
The Challenge: Enhance Meteorological Services to Metropolitan Areas
Modern urban meteorology in the United States likely had its beginnings in the provision of support to ice and snow removal efforts. It expanded in the 1960s with the mandated monitoring of atmospheric conditions to meet air quality standards. As cities have grown larger, local and regional governments, as well as local industry, have sought more environmental input to aid their decision making. Today, urban meteorology is recognized as a type of regional-scale meteorology where the region is highly populated and a large portion of the surface is covered with built infrastructure.
All metropolitan areas face a number of serious environmental challenges. Air quality management is one; another is effective response to many types of emergencies, such as heat waves, large fires (including wildfires), and toxic chemical spills. Some MSAs are located in regions prone to particular hazardous atmospheric phenomena, such as Miami, which is threatened by tropical cyclones, and Chicago, which deals with severe thunderstorms, wintertime cold and snow, and summertime heat waves. A major concern in recent years is that cities have become targets for terrorist attacks with possible releases of dangerous airborne agents into the urban environment. As a consequence, a number of homeland security issues are strongly tied to urban meteorology (NRC, 2003c). Furthermore, all MSAs rely on surrounding regions to produce continuous supplies of food, water, raw materials, and energy. As a consequence, metropolitan areas are sensitive to meteorological events at a distance, such as wildfires or strong winds damaging electrical services or a decline in mountain snowpack reducing the available supply of water.
The Need: A National Initiative to Enhance Urban Meteorological Services
A national initiative to enhance meteorological services tailored to and provided in MSAs is a high-priority need for a wide variety of stakeholders, including the general public, commerce and industry, and all levels of government. Enhancements would improve the quality of life and the public safety of those living and working in urban areas, increase the efficiency and competitive positions of urban-based industry and commerce, and aid emergency, medical, and law enforcement services in addressing threats to
life, property, services, and urban infrastructure. Some of the activities that need to be included in such an initiative include
Conducting basic research and development in boundary-layer meteorology; observations and network design; meso- and microscale meteorology; data assimilation and prediction systems; road, rail, and aviation weather; air quality and atmospheric chemistry; and hydrometeorology.
Prototyping and other R2O activities by the NWS to enable very short- and short-range predictions that employ advanced nowcasting techniques in the 0- to 3-hour range together with short-range mesoscale models; to improve analyses of initial and boundary conditions and short-range predictions for mesoscale model forecasts in the 12- to 48-hour range; to enable the merging of probabilistic guidance associated with nowcasting and dynamical predictions in the 3- to 12-hour range; to provide a basis for object-oriented verification of probabilistic forecasts resulting from ensemble techniques; and to facilitate technique development for advanced applications of mesoscale observations to locally disruptive weather such as fog, surface icing, thunderstorm initiation and motion, assimilation of precipitation measurements, conditions near hurricane landfall, other hazardous lake and coastal ocean conditions, hazardous urban conditions, fire weather predictions, and hydrologic predictions and warnings.
Supporting and improving productivity and efficiency in commercial and industrial sectors8 such as minimizing energy consumption in manufacturing and transportation; optimizing electric energy generation (see the following section on Renewable Energy); efficient building management; wholesale and retail sales across a broad range of projects; the commercial air, rail, maritime, and road transportation industries; the broadcast media; commodities exchanges; and insurance industries.
Urban planning for long-term sustainability that includes minimizing releases of polluting materials; minimizing the urban carbon footprint by reducing or eliminating greenhouse gas emissions from mobile and stationary sources; development of comfortable urban microclimates to enhance the quality of life; minimizing the impacts of potentially hazardous weather phenomena; and informing the inhabitants of current and expected future weather conditions. Fostering local solar and wind power, green roofs, and urban gardening in green spaces are other activities.
In the United States, the expertise needed to carry out the activities described above is distributed. The NWS has a national observing system and produces a wide range of forecast products each day. However, it is often private-sector meteorologists (less frequently, meteorologists employed by state or local governments) who tailor the general government forecasts to provide urban-focused weather products and services. These private-sector meteorologists usually target and are employed by niche markets, such as local departments of transportation, which need very specific information to plan for clearing ice and snow. So, although expertise and experience exist to address the above topics, they are not currently organized in any coherent way. The federal government, in particular the NWS, is unlikely to obtain the resources necessary for it to provide the full range of urban services described here. However, the NWS is in a position to provide leadership within the professional community and seed development and demonstration projects in such a way that the full spectrum of needed products and services becomes available, some from the government and some from the private sector.
Progress in the Past Decade
The importance of urban meteorology has been recognized in the past. For instance, in 2000, PDT–10 (Dabberdt et al., 2000) stated,
The urban weather problem is multidimensional in scope. Weather has special and significant impacts on a large fraction of the U.S. population who live in large urban areas. Conversely, large urban areas can impact the local weather and hydrologic processes in various ways. It is important to recognize that urban users have different needs for weather information than do their rural counterparts.
Further, recognizing the critical role of accurate information about the atmospheric conditions for predictions of transport, diffusion, and removal of atmospheric pollutants, the PDT–11 (Dabberdt et al., 2004) stated, “Improving atmospheric forecasts to provide improved air quality forecasts suitable for decision makers and the public is a major challenge.” As the key measurement problem for urban areas, they identified “the specification of vertical wind, turbulence, and temperature profiles, both below and above the urban canopy.”
Although some progress has been made to develop the special services that target the MSAs as recommended by PDT–10 and –11, recent studies in both the United States and elsewhere agree that there continues to be a critical need to significantly improve weather and environmental moni-
toring and prediction services for metropolitan areas and their supporting regional infrastructure. A fundamental issue is that most observing system and NWP model output, and hence the resulting products and services, are too coarse in resolution and lack representative surface conditions (i.e., aerodynamic roughness and displacement height, surface heat capacity and conductivity, and so forth) to provide the degree of detail necessary for true urban forecasts. In urban products, the meteorology needs to be done on scales representative both of where people live and work and of the relevant atmospheric processes.
Recognizing the need for increased attention to urban issues, the WMO devoted a session of its Third World Climate Conference to a discussion of climate and more sustainable cities.9 The session resulted in a statement of three priorities in urban weather and climate: climate research for hot cities, urban climate modeling, and education, training, and knowledge transfer in urban climatology (WMO, 2009). Relevant to the present discussion, Grimmond et al. (2009) identified and prioritized where improvements in observations, data, understanding, modeling, tools, and education are needed to ensure that in the next decade urban meteorology supports the development of more sustainable cities (Appendix A).
Around 2001 and running through 2006, the IBM Corporation demonstrated Deep Thunder, an experimental weather prediction service using very high performance computing facilities. The demonstration system provided local, high-resolution short-range weather predictions customized to weather-sensitive business operations. A goal was to provide weather forecasts, more or less on demand, at a very high level of precision. A prototype operational system provided 24-hour-, 1-km-resolution forecasts, updated twice daily, for the New York City area. It also produced forecasts at 4-km resolution covering the greater tri-state region and beyond, and then 16-km resolution for the rest of the northeastern United States. That capability was extended to include other parts of the United States. Although efforts were focused on refining the forecasts for the New York area, IBM also extended the system to provide forecasts for six additional metropolitan areas in other parts of the United States, including the Miami/Fort Lauderdale area. Deep Thunder demonstrated that with adequate computational resources it is possible to produce useful prediction products for the urban environment.10
Since the publication of PDT–10 and –11, new challenges and opportunities have arisen. For example, some urban and regional planners, public health officials, and meteorologists have developed an appreciation
of the connections between weather, public health, and concerns regarding long-term sustainability of large cities (Patz et al., 2000; Penney and Dickinson, 2009; WHO, 2009). In addition, both observations and critical weather information are being provided to large numbers of people using cell phones, personal digital assistants, and other such personal communication devices.
Observing systems, including traditional meteorological observing systems (providing measurements of temperature, pressure, moisture, wind, and precipitation), air and water quality monitoring systems, road weather sensors, and especially video cameras have proliferated in all urban areas. However, these systems are operated more or less independently by a wide range of governmental and private organizations and generally are utilized for narrow purposes; few of these data are shared with others. Options for developing partnerships to share data and services have been discussed at a recent USWRP workshop (Dabberdt et al., 2005) and in a recent BASC study report (NRC, 2009b).
Opportunities for Continuing Progress
Meeting the challenges of urban meteorology—monitoring and forecasting weather, hydrology, and air quality on the meso- and microscales in and around the complex urban built environment—requires that phenomena unique to the urban environment be better understood. Examples of urban phenomena warranting study include
Street-level weather: Winds, temperature, precipitation, and surface water (in all its phases) all impact quality of life, business operations, and transportation. Careful building design, urbanwide planning based on local climatology, and operational decision support systems are examples of actions that can be taken to minimize negative effects and make urban areas more effective and pleasanter places to live and work.
Wind in the urban canyon: Helical street-canyon circulations formed as air flows through the artificial canyons created by multistory buildings are a special element of microscale weather in the urban area. These wind circulations are complex and critical in the local transport and diffusion of pollutants, particulates (snow, sand, and dust), and toxic materials.
Evolution of the urban boundary layer: Urban surface properties (roughness, albedo, emissivity, heat capacity and conductivity, and permeability) and their changes with location in the metropolitan area contribute to the evolution of the urban boundary layer across the city. Account also
needs to be taken of local bodies of water and regional topography that can produce land and sea breeze effects and various types of slope-related winds.
Urban heat island: Ambient temperatures are moderated locally by surface heating, producing an urban “island” of warm air as first quantified by Luke Howard in 1833. This increase in temperature occurs because paved areas and buildings absorb (lower albedo) and retain (greater heat capacity) more solar energy than the surrounding natural landscape and so become relatively warm. The release of anthropogenic waste heat also contributes to the urban heat island. That stored energy is released slowly and so is retained well into the night. Consequently, cities tend to stay warm even as the surrounding natural landscape cools at night. The heat island effect contributes to and amplifies the number of casualties in urban areas during heat waves.
Urban flooding: All urban areas contain many roads, parking lots, and other areas with hard packed surfaces nearly impenetrable to water. In many urban areas, water courses have been converted into concrete-lined channels that may be either covered or open. Little rainwater is retained in the urban area, and so urban water courses sometimes quickly become raging torrents even for otherwise modest rain events. In this sense, urban flooding can be considered a special form of flash flooding. Not only do the rapidly rising waters threaten the lives of those in them at the time or who subsequently drive into the water courses as the water rises rapidly, the water can become quickly contaminated from the debris that accumulates in the channels between infrequent flushings of these urban waterways. In some cases, stormwater runoff can enter the normal sewage system and result in the release of untreated sewage. The result is a threat to life, the spread of disease, losses of high-cost property and infrastructure, and contamination of the water supply.
Wildfire at the wildland–urban interface: At the periphery of many urban areas, the city blends slowly into the surrounding forests and grasslands. This interface region can extend for many tens of kilometers and contain residences and small business, as well as other infrastructure. These areas can be highly susceptible to wildfire, posing a direct threat to life and property in the urban periphery but also in the urban core (due to atmospheric dispersion of the smoke, resulting in respiratory stress and also the disruption of vehicular and rail transportation routes due to reduced visibility).
Air quality: Poor air quality contributes to serious health issues, especially in the respiratory system. Although there have been decades of
research on air quality in urban areas, important research questions remain regarding pollutant photochemistry, transport and dispersion, as well as the interactions between pollution and humans that cause illness and disease.
Weather forecasting in support of urban emergency response: This is a particularly challenging task. Such support requires models that quickly capture essential features of current urban temperature, moisture, and wind fields; generate the necessary information; and provide predictions rapidly to users. Such information then can be blended with other information, such as traffic flow, locations of emergency response assets, critical assets or dangerous materials at the site of the emergency, to provide city leaders with the full picture of what is needed for a proper response. The NRC recently concluded that there exist critical gaps in the ability of operational emergency-response models (NRC, 2003c), especially in observing and prescribing the local dispersion environment and identifying and quantifying the source term.
Urban Weather Testbeds
Urban testbeds are needed in cities with widely different annual climates and settings (topography; water bodies) to conduct the research necessary to understand the phenomena mentioned above, to test observing and modeling techniques, and to develop and test products and services under a wide range of conditions. However, as pointed out by M. Ralph (in Dabberdt et al., 2005), such testbeds are much more than just a physical facility or an assembly of hardware and software:
A testbed is a working relationship in a quasi-operational framework among measurement specialists, forecasters, researchers, private-sector, and government agencies aimed at solving operational and practical regional [insert phenomenon or forecast challenge] problems with a strong connection to the end users. Outcomes from a testbed 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 long-term commitments and partnerships.
Urban testbeds would develop, test, and bring together the meteorological and socioeconomic aspects (see Chapter 2) of enhancing services to metropolitan areas. These testbeds would be charged with fostering the necessary basic and applied meteorological and socioeconomic research and with providing for the smooth transition of research findings to operations. Cultural issues, such as how to communicate effectively to the often
very diverse population found in urban areas, also require consideration. The advantages of employing a testbed strategy include efficiencies, formal opportunities to leverage earlier results from other researchers and other research activities, and a well-focused R2O mission.
Because most urban areas consist of many different government units and businesses, urban testbeds would need to engage and interact with a wide array of stakeholders. A unique aspect of the urban testbed is the strong emphasis on identifying, understanding, and communicating the impacts of urban weather on people, businesses, infrastructure, wildlife, and the natural environment. Effective communication of this information is an important component of this activity (see Chapter 2).
Operational attributes of an effective urban testbed include the following:
Location: Ideally it would be collocated with an NWS forecast office, and with easy access to a university with researchers in meteorology, engineering, urban and regional planning, etc.
Archive: As mentioned in Chapter 3, the testbed needs to archive both the full set of observations and record copies of products and services produced. Such data are essential for case studies, verification of forecasts, and trend analysis. The archive would also be useful for a variety of urban and building design activities.
Staffing: The testbed would be staffed with an interdisciplinary team from the public sector, academia, and industry that includes observationalists, modelers, forecasters, and socioeconomic researchers and practitioners; and should have close ties to the full spectrum of users. For example, social scientists are needed to evaluate the impact of focused weather products and forecasts on decisions by those living in the urban area, from the city management, to leading industrial and commercial entities, to individuals on the street, and to those at home. Staff should be familiar with and draw up unique local environmental resources, such as nearby sites of the U.S. Long Term Ecological Research Network.11
An extended, multiyear period of continuous effort, punctuated with intensive observing and forecasting periods, is envisioned for each testbed. These urban weather testbeds have two major areas of focus: (1) on the measurement and prediction of meteorological fields and phenomena and (2) on assessing, predicting, and mitigating the resulting socioeconomic impacts.
High-Impact Weather in the Urban Context
Forecasters and researchers need to work with local government agencies, private weather providers, and end users to identify and prioritize the weather services and products needed to address situations with high-value/critical impacts (see prior discussion of VHI weather). These priorities will then help determine research and development priorities in urban weather model development. The testbed would be an active partner in the city’s emergency operations center (EOC), promoting the extension of the EOC to cover more weather-related phenomena. An example is provided by the Shanghai MHEWS, which is a demonstration project of the WMO (Tang, 2006).
Observations and Measurements12
A key component of the urban testbed is its meteorological and air quality measurement network—the urban mesonet—which provides observations at high spatial and temporal resolutions from the urban core to the surrounding hinterland. Current modeling and observational expertise does not permit a priori specification of an optimal measurement network design—that is, the specification of the density and sampling attributes of an integrated suite of in situ and remote sensing systems. That, in fact, is one of the principal goals of the urban testbed. The commonly accepted approach is to oversample in the testbed and use data denial modeling techniques to identify an optimal network design (or multiple, optimum designs). Data from the urban mesonet then provide the basis for a number of important activities in the urban testbed: development and testing of data assimilation and prediction models; model-verification metrics; and applications where the observations themselves support various applications (e.g., cloud-to-ground lightning measurements are used to ascertain the hazard to airport ramp workers). Lastly, the urban testbed is a fertile environment for government agencies and industry to test new measurement devices and sampling techniques (such as the value of alternative adaptive and targeted sampling strategies).
The urban mesonet would utilize existent urban infrastructure wherever possible (e.g., communication towers and streetlight and traffic-light towers). Observations from tall buildings, communication towers, and remote sensing profilers provide essential information on vertical atmospheric structure,
The special observing needs of the urban environment are a subset of the issues discussed in detail in the section on Mesoscale Observations in Chapter 3.
such as lapse rate, wind shear, and mixing depth. The network might also include a mobile component with instrumentation on selected vehicles: city utility, police, fire, and emergency response vehicles, and delivery trucks (e.g., FedEx, UPS). New and existing sensors will need to be developed and adapted, respectively, for the special case of the urban street-canyon environment. Extensive and frequently updated metadata will be needed for each observation site and measurement system.
The urban mesonet would not only measure conditions in the urban core—a major challenge given the complex urban topography—it would also measure the atmospheric state over a domain sufficiently extensive to specify boundary and initial conditions appropriate to the temporal scale of the prediction models.
In principle, data from the extended urban mesonet would be assimilated in a high-resolution mesoscale numerical prediction model (Chapter 3) and then used to initialize an urban numerical prediction model. The urban landscape is an extremely rough and complex surface, one that makes it challenging to represent in a numerical prediction system. However, it is essential that such data be incorporated, and some schemes have been developed for this “urban canopy”; see, for example, the work by Chin et al. (2005) to develop and evaluate an urban canopy parameterization in a three-dimensional mesoscale model to assess the impact of urban roughness on surface boundary layer over the city.
Modeling the microscale airflows in the street canyons between tall buildings is a particularly challenging problem; however, predictions of such winds have numerous applications, ranging from transport of materials from a toxic release, to operation of large-building HVAC systems, to safety and quality-of-life issues for pedestrians on the street. Strictly speaking, such modeling could be addressed via computational fluid mechanics. However, the degree of complexity needed in models for such situations is a function of both the quantities to be determined and the relevant time frames in which they are to be made available. For example, European researchers have demonstrated that much useful knowledge about street-level conditions can be inferred through the use of relatively simple models, which . Operational examples are being used by the United Kingdom Meteorological Office and the research studies by the Environmental Prediction in Canadian
Cities (EPiCC)13 program; such efforts have demonstrated the feasibility of modeling at the street scale (Best et al., 2006). However, recent studies evaluating multiple urban land surface schemes, as reported by Grimmond et al. (2010), have documented very different performances, indicating the need for more research and development.
In addition to data from the atmospheric measurement network, model input will need to include anthropogenic fluxes of heat, moisture, and chemical compounds. A companion geographic information system (GIS) database is necessary to maintain information on fine-scale land-cover and building-height data, roadways and parking surfaces, buried infrastructure, soil and vegetation distribution, the urban forest canopy, and location data on critical facilities such as hospitals and residential areas. Many of these quantities evolve seasonally and change on a timescale of weeks to months. A GIS-based land-surface database could include processed, gridded data including important input parameters used in atmospheric and hydrologic models, such as roughness length, displacement height, plan area density, slope, soil type and vegetative cover. Initial efforts have been made to generate an urban database for various U.S. cities with the emerging National Urban Database with Access Portal Tool14 project, and the further development of such a database would be a high priority.
Urban predictions could be developed using high-resolution, nested, NWP models with advanced data assimilation schemes. Adapting existing models and upgrading them with high-resolution, improved physics might be a practical first step. Research needs arise particularly in testing and improving the boundary-layer and surface flux parameterizations used in these models because they were not designed for the required model resolutions and urban complexities. Representations for the urban canopy need to be developed and tested against observations on the appropriate scales. Development and continuous improvement of such models will be the primary thrust of the urban testbed. These improved models could also account for the surface roughness and modifications in the radiation balance. Evaluation and comparisons of the work being done in Canada (EPiCC) and Europe15 for street-canyon scale predictions could also be productive.
Uniformity of Service
As noted previously, U.S. urban dwellers and institutions have needs for weather information and services that differ significantly from those living in rural areas. It is thus worth revisiting a recommendation made some years ago by Carbone (2000), who suggested that the NWS reconsider its policy of “uniformity of service.” Service is currently distributed without consideration of population density. Even with the recent shift by the NWS to a digital gridded forecast, the current resolution is still coarse, with only a few points per county. Also, observing systems are more or less uniformly distributed, particularly where terrain is not an obstacle. As reported in PDT–10, Carbone suggests a different view; he argues for per capita uniformity of service, rather than geographic uniformity of service. The potential for weather-related societal and economic disruption in urban and rural forecast zones would be considered and uniformity could then be based on a comparable minimization of disruptions (Dabberdt et al., 2000). A new approach needs to be considered—one that focuses an increased level of resources and services on where the majority of the people and infrastructure are to be found, the urban areas.
Recommendation: The federal government, led by the National Oceanic and Atmospheric Administration, in concert with multiple public and private partners, should identify the resources needed to provide meteorological services that focus on where people live, beginning with a high-priority urban meteorology initiative to create infrastructure, products, and services tailored to the special needs of cities.
Although NOAA should be the lead agency in such an initiative, its success will require effective partnerships with other federal, state, and local government agencies, academia, and the private sector, as well as with all sectors of the user community, both public and private. Under the leadership of NOAA, a consortium of national and local partners should establish a small number of urban testbeds for the purpose of determining urban user needs for tailored meteorological information and then developing, testing, and evaluating various observing, modeling, and communication strategies for providing those end users with an effective suite of societally relevant and cost-effective products and services to meet those needs. The goal of such testbeds would be to conduct or foster the necessary basic and applied research and then transition the research findings together with the practical lessons learned into operations, and to extend these capabilities, appropriately scaled, to cities across the nation.
WEATHER INFORMATION TO SUPPORT RENEWABLE ENERGY SITING AND PRODUCTION
The production of energy by so-called renewable sources—principally water, biomass, municipal waste, geothermal, wind, and solar—is an integral part of the challenge to reduce reliance on fossil fuels, achieve a meaningful measure of energy independence, and mitigate global warming by anthropogenic carbon dioxide emissions. The U.S. Energy Information Agency (EIA, 2008a) estimates that the maximum electric power consumed domestically today is about 2.1 TW while globally it is 12.5 TW; EIA estimates that by 2030, peak domestic and global power consumption will rise to 2.8 TW and 16.9 TW, respectively, if the mix of power sources does not change. However, if the world switched to renewable sources only, then according to Jacobson and Delucchi (2009), the peak power consumption in 2030 would actually drop to 1.8 TW in the United States and 11.5 TW globally due to the increased efficiencies of direct electric power production and usage.
According to estimates of the Annual Energy Outlook by EIA (2008a, b), U.S. electricity generation in 2010 from all renewables will be a small but important fraction—about 10.7 percent—of total domestic electricity generation. Hydropower represents the largest share (60 percent) of electricity produced domestically from renewable sources, followed by wind (18 percent), biomass (12.4 percent), municipal waste (4.7 percent), and geothermal (4.0 percent). Solar energy presently contributes less than 1 percent of the electricity generated by all renewable resources. EIA estimates that despite the large contributions from these sources today, there will be little growth in electricity generation from hydropower,16 geothermal, and municipal waste over the next two decades (the extent of the EIA projections). In contrast to other renewable energy sources, electricity production in the United States by both wind and solar is growing rapidly. From 1990 to 2006, wind power production increased at a 14 percent compound growth rate (23 percent from 1997 to 2006) and solar production from on- and off-the-grid sources (e.g., residences and commercial buildings) grew at an estimated 30 percent compound rate from 2000 to 2008. The estimated annual growth rate from 2007 to 2030 is 6.2 percent for wind and 13.3 percent for solar. EIA projects that in 2030, electricity generation from all renewable resources
will be 14 percent of the total electricity generation from all sources, and that renewable resources will have accounted for 28 percent of the growth of 1 billion kilowatt-hours in projected electricity generation since 2010. And in the same way, wind and solar will contribute more than 24 percent of the growth in generation by all renewable resources.
Compared to fossil fuels, most renewable resources differ in several ways: they cannot be transported to where the power they generate is needed; their energy density is low (and so they must occupy large tracts of land); for many, the fuel is free; and most are weather dependent (directly or indirectly). By the same token, improvements in weather prediction will enable efficiencies in generation or grid operations. Improvements in hydrologic forecasts will lead to some improvements in hydropower production and management, and biomass yields may be increased with more accurate weather predictions. However, in the case of wind and solar generating systems, there are weather dependencies and uncertainties that pose significant challenges to their integration in a production and distribution system that must provide stable, predictable, and reliable electric power. These challenges, in turn, define an emerging set of priorities for weather research and the transition of findings and results into operations, and are the basis for the following discussions.
Wind energy today provides less than 1 percent of the domestic supply of electricity. At the end of 2007, the United States was ranked second in the world with an installed capacity of 16.8 GW (~18 percent of global total), but wind power generation is growing rapidly (Figure 4.3) in the United States which led the world in new capacity installed in 2007 with 5.2 GW. The Department of Energy (EIA, 2008a) is exploring the feasibility of increasing to 20 percent (300 GW) the total amount of domestic power produced from wind in 2030. Apart from the practical challenges of increasing wind-generating capacity more than 20-fold in 20 years, there is a host of weather-related research and operational challenges that will need to be met in order to ultimately provide power that is reliable and predictable. These challenges pertain to wind turbine design and operation, wind energy exploration and wind plant siting and design, and the long-term challenges of wind resource changes in a changing climate. Like so many of the other weather research and R2O priorities identified throughout this study, those pertaining to wind energy also involve both observations and modeling.
The newest generation of wind turbines have blades that sweep out an area of up to 126 m in diameter and produce up to 6 MW of power (GWEC, 2009). Most wind turbines typically operate at wind speeds in the range of about 4 to 25 m s–1, although the newer models (Vestas Corp., 2009) are projected to have a threshold speed of about 3 m s–1. Offshore turbines tend to be larger and generate more power (up to 5 to 6 MW) owing to the higher installation and maintenance costs, and the subsequent goal of generating equivalent power with fewer systems. Land-based wind turbines have stabilized in the range of 1.5 to 3 MW, which enables economies of scale in production costs. Modern turbines are able to control their power output by changing the angle of their blades to the wind (pitch control), by turning (yawing) in response to wind direction changes, and operating at variable speed (enabling it to synchronize with the operation of the electric grid). For obvious reasons, wind park revenues are very sensitive to wind speed but the actual sensitivity is enlightening: a difference of 1 m s–1 in the annual average wind resource can result in an annual revenue differential of about $1 million for a 200-MW wind park.
The weather-related research and R2O needs of the wind energy industry have recently been discussed and documented by a January 2008 community workshop organized by DOE (2008) and an August 2009 community meeting organized by the American Meteorological Society (AMS, 2010). Each event brought together more than 120 atmospheric scientists and wind energy engineers from the public, private, and academic sectors. This report draws heavily on the recommendations from both events. While the generation and delivery of electric power is the role of the private sector, the challenge of providing secure and sustainable energy is a national priority that involves federal, state, municipal, and regional agencies and the intramural and extramural research efforts that they support. Expanding America’s electricity generation capacity using renewable resources is closely linked to our nation’s emerging climate change mitigation policy17 as electricity generation from wind, solar, geothermal, and hydro resources leaves virtually no carbon footprint (notwithstanding the carbon emissions that result from the manufacture of the wind turbines and other renewable energy systems).
For example, the Kerry-Lieberman bill calls for a 17 percent cut in emissions below 2005 levels, by 2020, with the emission limits applying in different ways to power plants, petroleum refiners, and trade-sensitive manufacturers. See http://kerry.senate.gov/imo/media/doc/APAbill3.pdf.
Assessing the wind energy resource requires quantifying the distribution of wind speed (and vertical shear) as a function of location, time, and height. At present, the number of surface weather stations in the continental United States (CONUS) that monitor wind is about 30,000, which yields an average of about one station per 100 km2. Unfortunately, their spatial distribution is heavily biased against the areas that have the greatest wind energy potential. Further, there is very little information on vertical shear—there are only about 125 boundary-layer wind profilers in CONUS (and these are not sited for wind energy assessment) and about 100 radiosonde stations that make twice-daily soundings of winds and state variables (again, these are not well sited for wind exploration). Although not measured widely, wind shear is an important factor in wind turbine design and operation.
Wind Turbine Design
Wind turbines must be designed to operate in a very turbulent wind environment for at least 20 years according to the standards of the International Electrotechnical Commission (IEC). Wind turbines must operate with loads that are the result of their own inertial effects, as well as from the spatial and temporal changes in wind speed, direction, shear, and vorticity. These loads have increased as wind turbines have become larger and because they were being installed at many locations that are intrinsically very turbulent. As discussed by Veers and Butterfield (2001), wind turbine failures have been caused by inaccurate estimation of design loads, which mandated analysis and prediction techniques that yield more detail in the inflow—both in the undisturbed ambient flow and in the wake of upwind turbines.
Detailed structural dynamic models have subsequently been developed that include turbulence models (that simulate the stochastic inflow fields), aerodynamic models (that predict aerodynamic loads from the turbulent inflow), and turbine control algorithms (that control turbine pitch, yaw, and braking actions). These load analyses and predictions are especially challenging as they need to capture not only the “normal” wind and turbulence characteristics of the flow field, but also those resulting from external meteorological conditions (apart from classic surface-layer theory), and terrain effects at a wide range of sites. Especially important are the rare events that can impose extreme loads on the system.
In this way, load predictions need to consider a wide range of atmospheric conditions. Because of their large length, wind turbine blades are
exposed to variable stresses throughout a significant fraction of the lowest region of the atmospheric planetary boundary layer (PBL) where wind shear and turbulence are typically most severe. At night when the PBL can be very shallow (50 to 100 m), the lower part of the blades can be exposed to stable conditions with little mechanical turbulence but strong vertical wind shear in both speed and direction. At the same time, the upper reaches of the blades can rotate through (and above) the capping PBL inversion where shear and turbulence can both be significant. The nonhomogeneous nature of the environment in which the turbine operates can place significant mechanical stresses on the turbine blades and gearbox. In addition to the natural stresses from ambient conditions, the wind turbines are exposed to wake turbulence created by neighboring turbines within the wind park. These conditions place demands on the design of the wind turbine that, in turn, requires detailed knowledge of the mean and turbulent flow conditions throughout the wind park that can only be achieved through a combination of representative local observations and numerical modeling.
Obtaining reliable summary data on the role of weather (wind, turbulence, icing, temperature, and lightning) on the failure of wind turbines in general and specific subassemblies is difficult. Often the cause of the failure is not known or is not released for proprietary reasons. But there are some studies that summarize the failure rates (without causal attribution) for wind turbine systems and their various subassemblies. Tavner et al. (2008) provide a valuable overview of the reliability of wind turbine systems based on an analysis of more than 6,000 wind turbines in Denmark and Germany that have been in operation for 11 years. They found failure rates for various subassemblies of 1 MW turbines ranged from about 0.05 to 0.5 failures per turbine subassembly per year. The higher rates occurred with the electrical systems and the rotors (blades and hub) while gearboxes had the lowest failure rate. They also found failure rates increased with turbine size, but that failure rates had decreased over time (Figure 4.4).
Wind Plant Architecture
It has long been known that the longitudinal and lateral spacing of individual wind turbines strongly influences the power generating efficiency of the wind farm. If the wind turbines are placed too close together, there will be a reduction in the output of all downwind turbines due to the wake velocity deficits of the upwind turbines. Placing the turbines too far apart reduces the number of turbines in the wind farm and thus reduces the amount of energy that can be generated per unit area of the wind farm.
Both of these effects negatively influence the cost and efficiency of the wind farm and the power it can generate. Recent studies (e.g., Werle, 2008) have demonstrated that even with the relatively large longitudinal and lateral separation distances of seven turbine-blade diameters, downwind turbines will sometimes deliver only about 65 percent of the power generated by the upwind turbine. Therefore, optimizing the layout of the turbines in a wind farm is critical.
The problem of wind farm configuration is further complicated by the effects of local topography. Often the areas with the highest wind potential are those offshore as well as onshore locations with terrain relief. Both types of sites can present special aerodynamic challenges to the layout of the wind farm. Offshore sites are typically impacted by sea breeze circulations,
time-variable surface friction effects, and stable conditions through the lower boundary layer. Wind farms in areas of terrain relief are subject to locally generated mechanical turbulence. In both cases, it is important to also consider the spatial variability of wind and turbulence across the wind park. Such knowledge cannot be gained from ambient wind observations alone, and advanced computational fluid dynamics (CFD) and large-eddy simulation models are needed to locate the wind turbines within the wind farm in order to minimize wake effects from adjacent wind turbines. Models are also instrumental in optimizing the locations of meteorological observing sites, which in turn can provide data assimilated by the models.
The effective operation of a wind park requires wind forecast information on different timescales and across the spatial domain of the wind park (many of which span hundreds of square kilometers of spatially varying topography). Forecasts on the 0- to 48-hour timescale are important for determining the amount of wind energy that can be provided by the wind park, and also determining how much energy will need to be purchased from conventional fossil energy sources. As wind parks become larger or are aggregated onto the electric grid, their sensitivity to wind variations and to wind-forecast accuracy is lessened to a degree.
On the other hand, wind park operations are very susceptible to so-called wind ramp events, which are abrupt, major changes in wind speed over a relatively short period of time (one simple, quantitative definition refers to a wind power change ≥50 percent over an interval of ≤4 hours). The problem of “wind integration” is a significant operational obstacle facing wind energy production. The term refers to the measures energy system operators must take when winds change rapidly (wind ramp events), causing a sudden increase or decrease in wind power generated by turbines. Electric utility companies usually have a specific generation configuration consisting of a mix of coal, oil, gas, hydro, wind, and nuclear power. When winds pick up, the amount of power generated by the wind turbines increases rapidly, causing excess power to be injected into the generation system. Because the system is not capable of handling endless amounts of power, the utility company must quickly ramp down generation from the wind turbines (some wind turbines are able to have their rotational speed controlled) or ramp down other, conventional sources and route excess electricity to neighboring utilities. In the case when the winds may abruptly decrease, the result is a sudden need to import power and turn gas turbines on quickly, often
in minutes. To mitigate the potential for a ramp-up requirement, operating utilities often keep a spinning reserve of gas turbine generators running ready to fill the power gap. This, however, is not an ideal solution because the spinning reserve increases the cost of wind power and can decrease potential carbon savings. There are also costs associated with slowly cycling winds when conventional thermal (coal) units are ramped up and down; in these cases, there are excessive wear-and-tear costs on the thermal units than can be appreciable.
Wind integration costs are difficult to estimate and are not simply the differential between operations with and without wind-generated power (Milligan and Kirby, 2008). In their analysis, they present the example of a system that has sufficient capacity to meet a fast ramp, but in order to meet the ramp, a peaking unit must be utilized. In this case, the baseload unit has an energy price of $10/MWh, but is unable to increase output quickly enough to meet the ramp. A peaking unit is brought online at a price of $90/MWh to meet the ramp. If the marginal unit sets the energy price in a market, the energy price rises during the ramping period because the online base unit is not flexible enough. Had the ramp been sustained during a longer period of time, the base unit could have met the ramp requirement, and the energy price would have remained at $10/MWh. This example points out that ramping can be extracted (at a high price) from the energy market. Further, in a system with significant wind penetration, the ramp scenario can be exacerbated, necessitating even more ramping capability at an even higher price. Milligan and Kirby (2008) further argue that “pooling loads and resources into a larger balancing area holds the promise of allowing additional wind to be integrated into the system at lower cost.”
To help mitigate the wind ramp/integration problem, there is a pressing need to develop very short-range NWP and nowcasting methods that can reliably predict wind ramp events on the 30-minute to 6-hour time frame. Providing accurate and reliable, cost-effective forecasts (with equivalently low false alarm rates) of the onset, duration, and magnitude of wind ramp events that are specific to the location of wind farms may be the single most pressing contemporary meteorological need in support of wind power operations.
Need for Improved Meteorological Observations
Meteorological observations—primarily wind and turbulence, but also temperature—are required to aid in wind park assessment, siting, and design and to assess the performance of individual wind turbines. They are also
needed to initialize NWP models and for data assimilation models that enhance model performance. They are also required to assess (validate) model performance under different meteorological and topographic regimes. Meteorological observations are also necessary for very short-range prediction (i.e., nowcasting) of wind changes on the 0- to 2-hour timescale. Because of the significant vertical extent of large wind turbines and the effects of wind shear and turbulence, it is essential to have detailed knowledge of the mean and turbulent structure of the PBL (and beyond, in some cases), up to heights of about 0.5 km above the land or water surface. Vertical profiles need to be highly resolved both in space (5- to 10-m height resolution) and time (on the order of 5 minutes).
In some cases, research and development is needed to develop measurement systems that better meet the needs of the wind energy industry. In particular, remote sensing wind profilers—optical or ultrahigh frequency (UHF)—are needed to provide long-term and high-resolution profiles of vector velocities up to 300 to 500 m. They need to be cost-effective as well: affordable, easy to install and maintain, and sufficiently robust to operate unattended for years.
The current state of mesoscale observations is inadequate to support the high-resolution modeling needs of the wind energy industry to predict wind energy at, and within, wind parks. This will require dense arrays of surface observations (both wind and state variables) both for nowcasting and moderately dense arrays for mesoscale NWP modeling, but also remote sensing profilers to adequately characterize wind and turbulence through the PBL. The ability to specify the details of the design of these networks is inadequate, and that also is a research area requiring high priority. The mesoscale observing issues and needs discussed in Chapter 3 would well serve the needs of the wind energy enterprise. There it was recommended that humidity, wind, and diurnal boundary-layer structure profiles are the highest priority for a national mesoscale network, the sites for which need to have a characteristic spacing of approximately 150 km but could vary between 50 and 200 km based on regional considerations. Because the most important timescales for mitigating ramp events is short—less than 1 hour—it can be estimated from Taylor’s hypothesis that nowcasting a ramp event with a 15 m s–1 wind speed requires wind observations about 55 km upstream, which is consistent with the previous recommendation. The recommended profiler network would, of course, also be invaluable for producing improved NWP predictions of many wind ramp events on longer timescales.
Need for Improved Modeling
A hierarchy of models, each simulating a range of spatial scales, is required to address the various wind-dependent aspects of wind turbine/park design and operation, including the following:
CFD models are needed to specify turbulence features at, and down-wind of, individual turbines and assess the flow interactions among adjacent turbines.
High-resolution atmospheric (including large eddy simulation) models are needed for the purpose of describing the turbulent flow characteristics and structure of the ambient PBL and interactions with turbulent wakes from individual wind turbines and large arrays of turbines.
High-resolution mesoscale models (grid resolution of 0.5 to 1.0 km) are required to assess the wind energy potential (for prospecting applications) of regions up to 103 km in scale. They are also needed to forecast the wind energy that can be provided by a given wind park—including variations within the park—over forecast periods of 48 to 72 hours. For the latter purpose, ensemble predictions will aid in determining the range of wind power output that can be anticipated, which is useful for assessing the need for alternative power sources. Nowcasting algorithms are needed to provide reliable predictions of very short-range changes in the wind field up to 2 hours.
Need for Improved Collaborations
In spite of the commonalities and feedbacks among research challenges in the four key areas—turbine dynamics, siting and array effects, mesoscale processes, and effects of a changing climate—research in any one area has largely been conducted independent of the research challenges of the other three areas. Enhanced interdisciplinary collaborations among researchers can provide results that are more beneficial, more effective, and timelier. In this way, for example, optimum wind turbine arrays can be designed as a consequence of closer collaboration of fluid dynamicists and mesoscale meteorologists, and improved long-range wind resource planning may result from increased interactions among not only mesoscale and climate researchers but also socioeconomists. In the same way, closer interactions are required among observationalists, experimentalists, modelers, and users. Effort now focused within individual disciplines needs to be augmented by
a flatter, more interdisciplinary approach. This applies both to individual researchers and groups as well as to the governmental agencies that support and prioritize the research.
Partnerships are essential in developing the modeling tools needed to design and operate wind parks. There are important complementary roles to be played by the public, private, and academic sectors, and the three sectors need to find effective ways to collaborate.
Solar power refers to the production of electricity, directly or indirectly, from ambient sunlight. Photovoltaic (PV) systems convert sunlight directly into electricity using arrays of solar cells of various types, such as thin-film, monocrystalline silicon, polycrystalline silicon, and amorphous cells. Arrays of PV panels can provide electricity directly to their hosts (e.g., homes, factories, office buildings, airports) or they can be configured in very large arrays to comprise photovoltaic power stations that provide electricity to the power grid. The Olmedilla Photovoltaic Park in Spain is currently the world’s largest PV plant.18 Built in 2008, the plant has more than 160,000 solar PV panels to generate peak power of 60 MW, which is enough electricity to supply more than 40,000 homes. Concentrated solar power systems produce electricity indirectly by using combinations of mirrors, lenses, and solar trackers to focus sunlight from a large area onto a small area where the concentrated solar energy is used to heat water for use in a conventional thermal power plant. The Solar Energy Generating Systems facility in the Mojave Desert is reported to be the largest solar system in the world, consisting of nine solar power plants that have a combined capacity of 354 MW.19 Yet, solar energy today only supplies about 0.9 percent of the U.S. domestic energy production from all renewable sources (EIA, 2008b; Figure 4.3a).
The availability of electricity from solar energy systems, like that from wind farms, is highly variable in space and time, which poses significant challenges for a power grid that requires stability and predictability. Developers of large solar plants and utilities that utilize distributed PV systems that
encompass residential and commercial installations have common needs for reliable observations, historical and real-time databases, and accurate forecasts on a variety of scales.
Existing historical, national solar radiation databases are available from the DOE National Renewable Energy Laboratory (NREL, 1995, 2007), the State University of New York at Albany (Perez et al., 2007), and NASA.20 They are based on in situ measurements from a few tens of surface-based radiometers together with satellite cloud imagery and analytical interpolation algorithms. However, the accuracy of hourly estimates of surface solar irradiance from satellite imagery is only ±20 percent. Additionally, the temporal resolution of 1 hour does not meet user needs for 15-minute data (some users may require temporal resolution as fine as 1 minute); nor is the spatial resolution (≥10 km) of these databases consistent with many user needs for data resolution of 1 km. It is notable, however, that an excellent mesoscale solar and meteorological measurement network and database exists at the Southern Great Plains site of the DOE Atmospheric Radiation Measurement program (Ackerman and Stokes, 2003), which can provide an invaluable resource for testing and evaluating alternative measurement, analysis, and forecasting methods.
Utility operators require solar resource forecasting on several timescales: ≤3 hours for dispatching to enable a steady power supply to the grid; 24 to 72 hours for system operations planning; and seasonal to interannual forecasting for economic analyses and system planning. However, today there does not exist an operational solar insolation forecasting capability that meets user needs.
The needs of users and the shortcomings of existing measurement networks, solar databases, and forecast models were addressed in a recent AMS community workshop (Weatherhead and Eckman, 2010) and a recent NREL workshop and subsequent extensive needs study (Renné et al., 2008); these findings reflect their conclusions:
There is a pressing need for high-resolution (15 minutes or less) solar resource data derived from the hourly model results, and/or from additional high-quality measurements. This involves obtaining more reliable site-specific data from extrapolation methods, more spatially refined satellite-
derived solar resource estimates, and on-site measurements of downwelling solar radiation (all components) and downwelling infrared radiation.
Reliable, operational solar forecasting is not available today. Users need 12- to 72-hour solar resource forecasts, as well as very short term (≤3 hours) and seasonal-to-interannual forecasts, for use by system operators in system planning and load-following operations.
An interactive archive of solar resource information is needed so that developers, utilities, system operators, and system planners can access solar resource information needed for specific analyses and applications.
Lower-cost solar resource measurements and assessments are needed for key locations (e.g., distributed PV systems).
Improved satellite datasets are required that encompass improved estimates of aerosols, better detection of snow cover, and higher spatial resolution.
Relationship to Other Weather Research Needs
Improved observations, simulations, and predictions of wind and turbulence and solar radiation are needed with high spatial and temporal resolution and accuracy to optimally locate, design, and operate wind and solar energy facilities. These efforts will require a focused, high-priority national research and R2O program that would be carefully and closely integrated with the observing and predicting initiatives and socioeconomic actions recommended in Chapters 2 and 3 of this report. In this way, previous recommendations pertaining to improved NWP with global nonhydrostatic models (page 58), improved quantitative precipitation estimation and forecasting (page 69), improved hydrologic prediction (page 70), and improved mesoscale observing systems (page 87) will all positively impact the information needs of the renewable energy enterprise as they pertain to electricity generation from wind and solar (but also from hydro resources). By the very nature of these applications, the efforts recommended here need to be a true public–private partnership with significant involvement of academia as well. To be successful, these efforts will require the formation of effective collaborations and partnerships among power system designers, operators, grid managers, observationalistsresearchers, forecasters, and modelers. It is encouraging that the DOE has interacted closely with academia and the private sector, as exemplified by the community workshops the DOE has organized to facilitate improved weather-related support for wind and solar energy production. And NOAA has recently indicated that its Next Generation Strategic Plan (scheduled for release in summer 2010)
will explicitly include a focus on support for energy production from renewable resources.21
Recommendation: The effective design and operation of wind and solar renewable energy production facilities require the development, evaluation, and implementation of improved and new atmospheric observing and modeling capabilities, and the decision support systems they enable. The federal agencies should prioritize and enhance their development and support of the relevant observing and modeling methods, and facilitate their transfer to the private sector for implementation.