3

Science and Technology

Various reports have discussed the critical role of urban meteorological observations and forecasting for various aspects of society, such as public health, public safety and security, transportation, water resource management, storm water runoff, and economic development. Although some progress is evident since these reports, cities still pose a number of difficult challenges for both the scientific and end user stakeholder communities that are not adequately addressed by current meteorological observation, forecasting, and information dissemination technologies. This chapter assesses current capacity, emerging technologies, and future needs related to observing, modeling, and forecasting in the urban environment. A brief review of current urban meteorological knowledge is appropriate to provide context for the issues at hand.

As with broader weather systems (i.e., floods, thunderstorms, blizzards, and hurricanes), urban meteorological processes are directly relevant to societal activities and can significantly affect economic, hazard management, and public-health decision-making. For example, in 2010 and again in 2011, the capital of the United States, Washington, D.C., was crippled by record-breaking snowstorms. Scores of federal workers and even the President experienced transportation and work delays as a result.

During the summer of 2010, the country experienced record-breaking urban flooding in places like Atlanta, Georgia, Nashville, Tennessee, and Oklahoma City (Shepherd et al., 2011). In 2011, the south central U.S. experienced extended periods of very extreme heat (>100 degrees Fahrenheit) for days on end. During this regional-scale heat wave event, urban heat islands in cities like Dallas and Oklahoma City likely caused prolonged heat exposure at night (Figure 3.1). Lessons from the 2003 European Heat Wave and the 1995 Chicago Heat wave (Changnon et al., 1996; Menne, 2003; Beniston, 2004) have indicated that heat mortality is exacerbated by excess heat stored in urban landscapes—the well-known Urban Heat Island (UHI) effect.



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



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

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

OCR for page 57
3 Science and Technology Various reports have discussed the critical role of urban meteorological observations and forecasting for various aspects of society, such as public health, public safety and security, transportation, water resource manage- ment, storm water runoff, and economic development. Although some progress is evident since these reports, cities still pose a number of difficult challenges for both the scientific and end user stakeholder communities that are not adequately addressed by current meteorological observation, forecasting, and information dissemination technologies. This chapter as- sesses current capacity, emerging technologies, and future needs related to observing, modeling, and forecasting in the urban environment. A brief review of current urban meteorological knowledge is appropriate to provide context for the issues at hand. As with broader weather systems (i.e., floods, thunderstorms, blizzards, and hurricanes), urban meteorological processes are directly relevant to societal activities and can significantly affect economic, hazard manage- ment, and public-health decision-making. For example, in 2010 and again in 2011, the capital of the United States, Washington, D.C., was crippled by record-breaking snowstorms. Scores of federal workers and even the President experienced transportation and work delays as a result. During the summer of 2010, the country experienced record-breaking urban flooding in places like Atlanta, Georgia, Nashville, Tennessee, and Oklahoma City (Shepherd et al., 2011). In 2011, the south central U.S. expe- rienced extended periods of very extreme heat (>100 degrees Fahrenheit) for days on end. During this regional-scale heat wave event, urban heat islands in cities like Dallas and Oklahoma City likely caused prolonged heat exposure at night (Figure 3.1). Lessons from the 2003 European Heat Wave and the 1995 Chicago Heat wave (Changnon et al., 1996; Menne, 2003; Beniston, 2004) have indicated that heat mortality is exacerbated by excess heat stored in urban landscapes—the well-known Urban Heat Island (UHI) effect. 57

OCR for page 57
58 58 URBAN METEOROLOGY FIGURE 3.1 Number of days that maximum temperature exceeded 100 degrees (F) from June 1 to Au- gust 31, 2011. Urban heat islands in cities like Dallas and Oklahoma City likely caused prolonged heat exposure at night. SOURCE: NOAA. There have been many reports that highlight the effects of anthropogenic greenhouse gases on future weather and climate and a very clear sign of human alteration to atmospheric processes is evident (NRC, 2010b; NRC, 2011b; IPCC, 2007; USGCRP, 2009). There is also emerging interest in un- derstanding the role of urbanization on the climate system, which includes weather (IPCC, 2007). Cities change properties of the land surface and subsurface, and are also known to influence atmospheric circulation patterns at various spatial scales (Hidalgo et al., 2008; Grimmond et al., 2010b). The high build- ing densities and sparseness of vegetation in cities makes urban surfaces

OCR for page 57
SCIENCE AND TECHNOLOGY 59 typically much rougher and drier than rural surfaces. In addition, the three- dimensional nature of urban environments affects a number of parameters such as evaporation rates, absorption and reflection of solar radiation, and storage of heat, as well as wind and turbulence fields (Figure 3.2). Gaseous and particulate matter emissions (Figure 3.3) also may contribute to land- atmosphere interactions. Classifications for urban land cover zones have been developed (Stewart and Oke, 2009 a, b) that are useful for documenting metadata of urban monitoring sites and for characterizing surface features in urban modeling tools. FIGURE 3.2 Variability of land cover zones in New York City characterized by: differences in surface mor- phology, percentage of surface cover, and sources of heat, water, other gases, and particulates. SOURCE: Sue Grimmond and Bing Maps. R02149 Urban Meteorology Figure 3-2 bitmapped raster image

OCR for page 57
60 60 URBAN METEOROLOGY FIGURE 3.3 Urban-atmosphere interactions. SOURCE: adapted from Hidalgo et al., 2008. ©John Wiley & Sons Ltd. Reprinted with permission. R02149 Urban Meteorology Overall, urbanization results in a suite of complex land surface atmo- Figure 3-3 sphere interactions that modify thermodynamic, radiative, dynamic, and bitmapped raster image hydrometeorological processes within the urban area and its surrounding regional footprint, particularly downwind of the city. Extreme weather and climate events in cities typically occur during unfavorable regional-scale conditions exacerbated by global warming trends and local, urban effects (Hunt et al., 2007). As an example, regional-scale heat waves, which are

OCR for page 57
SCIENCE AND TECHNOLOGY 61 expected to become more frequent because of climate change, in com- bination with urban heat island development, exacerbate heat stress for urban populations (Zhou and Shepherd, 2009; Stone et al., 2010). Seto and Shepherd (2009) noted that “the built environment characterized by urbanization is a significant forcing function on the weather climate system because it is a heat source, a poor storage system for water, an impediment to atmospheric motion, and a source of aerosols (e.g., pollutants).” Table 3.1 illustrates several pathways by which urban processes can influence me- teorology or climate. URBAN METEOROLOGY: A SYNOPSIS OF THE SCIENCE The urban heat island’s role on regional and global climate has been the subject of much research (Oke, 1982; Arnfield, 2003; Roth, 2007; Yow, 2007; Hidalgo et al., 2008; Grimmond et al., 2010b) and is one of the most well-studied and familiar manifestations of urban weather modification (Figure 3.4). Research shows that it is spatially correlated with regional land- use and land-use change. During the early phases of urban development, multiple land covers—bare land, vegetated areas, agricultural plots, and built-up areas—emerge in close proximity with each other. As urbanization increases, resulting in reduced vegetated surfaces, the spatial pattern of the urban heat island becomes less scattered and more intense (Imhoff et al., 2010). Stone et al. (2010) recently argued that temperature increases in sprawling urban environments were as likely (or more likely) to be attributed to urban landscape changes as greenhouse gas warming (see Stone abstract in Appendix A). It should, however, be noted that the types of data used to define UHI characteristics play an important role, and the discussion of UHI often lacks important information about data sources and site characteristics (Stewart, 2011). In situ observations of air temperatures from measurement sites within the urban canopy layer provide information about the urban-canopy-UHI. Remotely sensed observations provide information about boundary-layer heat islands (above the urban canopy layer in the atmosphere) and surface heat islands. In general, weather patterns in and near cities depend both on the degree of urbanization—characterized by the above mentioned changes in land surface characteristics, subsurface properties, and chemical com- position of the atmosphere—and on larger scale meteorological conditions (Mestayer and Anquetin, 1995). For situations with moderate-to-high wind speeds, an urban plume with warmer, polluted air is advected downwind of the city (Figure 3.3). Under such conditions, the lowest portion of the urban boundary layer (UBL), the surface layer, can be divided into two main

OCR for page 57
62 62 URBAN METEOROLOGY TABLE 3.1 Various Pathways for Urbanization to Affect the Weather Climate System The column headings represent three ways that urban environments can affect weather and cli- mate. The row headings represent different weather processes affected by each of the three. For example, urban land cover affects/causes an urban heat island because urban land cover modifies the surface energy budget gradients—develops and dominates the RSL wind and turbulence pat- terns. This shear layer also controls the turbulent exchange and ventilation between the UCL and the flow above average roof level, which is typically highly instantaneous and controlled by coher- ent structures (Christen et al., 2007). Overall, the RSL plays a critical role, and its properties need to be properly resolved in numerical models for accurate urban weather and air-quality forecasts. Anthropogenic Greenhouse Gas (GHG) Urban Land Cover Urban Aerosols Emissions Urban Heat Surface Energy Budget Insolation, Direct R adiative Warming Island and Aerosol Effect and Feedbacks Mean Surface Temperature Record Wind Flow, Surface Energy Budget, Direct and Indirect Radiative Warming Dispersion, Urban Morphological Aerosol Effects and and Feedbacks Transport, and Parameters, Mechanical related dynamic/ Turbulence Turbulence, Bifurcated thermodynamic Flow response, Dispersion and Transport Clouds and Surface Energy Budget, Aerosol indirect effects Radiative Warming Precipitation UHI Destabilization, UHI on cloud-precipitation and Feedbacks Meso-circulations, UHI- microphysics, insolation induced convergence effects zones Land Surface Surface runoff, reduced Aerosol indirect effects Radiative warming Hydrology infiltration, less on cloud microphysical and feedbacks evapotranspiration and precipitation processes Carbon Cycle Replacement of Black carbon aerosols Radiative warming high net primary and feedbacks, fluxes productivity (NPP) land of carbon dioxide with impervious Surface Nitrogen Cycle Combustion, Acid rain, nitrates Radiative warming and fertilization, sewage feedback, NOx emissions release, and runoff a SOURCE: Adapted from Seto and Shepherd, 2009.

OCR for page 57
SCIENCE AND TECHNOLOGY 63 FIGURE 3.4 A qualitative description of typical rural and urban surface energy balance processes. The values are in units of kW h m2 day−1. Figure courtesy of R. Sass, Rice University. SOURCE: Shepherd, 2005. © Earth Interactions. Reprinted with permission. R02149 Urban Meteorology sublayers (Figure 3.5): the roughness sublayer (RSL) and the inertial sublayer Figure 3-4 (ISL). The RSL typically extends from the surface to a height equivalent to bitmapped raster image 2-5 times the average building height (Raupach et al., 1991; Rotach, 1999; Kastner-Klein and Rotach, 2004). The layer below average roof level, the lowest portion of the RSL, is often referred to as urban canopy layer (UCL). Within the UCL, atmospheric patterns are spatially inhomogeneous, strongly influenced by local effects, and very hard to predict (Klein et al., 2007; Vardoulakis et al., 2003). At the same time, the UCL is the region where most of the urban anthropogenic emissions of atmospheric pollutants occur and where people spend most of their time, and is thus of great relevance. In the upper part of the RSL, above average roof level, a strong shear layer—a layer with high wind velocity gradients—develops and dominates the RSL wind and turbulence patterns. This shear layer also controls the turbulent exchange and ventilation between the UCL and the flow above average roof level, which is typically highly instantaneous and controlled by coherent structures (Christen et al., 2007). Overall, the RSL plays a critical role, and its properties need to be properly resolved in numerical models for accurate urban weather and air-quality forecasts.

OCR for page 57
64 64 URBAN METEOROLOGY FIGURE 3.5 Sketch of the urban boundary layer structure indicating the various (sub)layers and their names. SVF in box c) stands for sky view factor. SOURCE: Rotach et al., 2005, modified after Oke, 1987. ©Springer-Verlag Wien. Reprinted with permission. R02149 Urban Meteorology Figure 3-5 bitmapped raster image

OCR for page 57
SCIENCE AND TECHNOLOGY 65 For large-scale weather patterns with weak dynamic forcings (i.e., weak winds), urban flow patterns are primarily thermally driven, and a dome-like circulation pattern develops in cities that are located in flat terrain and away from large water bodies. Canopy-layer UHI signatures are typically strongest under such conditions, with the highest values recorded at night. Historical and current literature has also persistently shown that UHI destabilization, urban surface roughness, and pollution can independently or synergisti- cally initiate, modify, or enhance precipitation cloud systems. The so-called “urban rainfall effect” is clearly established in the literature; the majority of studies note an enhancement of rainfall in warm season convection (Chang- non et al., 1981; Bornstein and Lin, 2000; Shepherd et al., 2010). However, there is discourse about the sign of precipitation change (i.e., the increase or decrease of precipitation) associated with urbanization and what physi- cal mechanisms are of primary significance (Ashley et al., 2011; Shepherd et al., 2010). Studies continue to verify that urbanization may also modify lightning (Rose et al., 2008), freezing rain (Changnon, 2004), and snowfall climatologies (Shepherd et al., 2010) as well. Further, urban land use accel- erates hydrologic response through surface runoff variability and stresses on conveyance systems (Shepherd et al., 2011; Villarini et al., 2010a; Reynolds et al., 2008) and thus amplifies urban flooding risks. Human activities in urban areas (e.g., transportation, energy, and in- dustrial processes) result in the production of “urban” aerosols or pollution which is associated with increased greenhouse gas emissions. Although urban areas have significantly higher carbon dioxide concentrations than in rural areas, greenhouse gas emissions per capita may be lower for urban dwellers than those for rural dwellers (Dodman, 2009). Jacobsen (2010) has recently discussed the implications of urban carbon domes on public health and the climate system. Urban modification of winds, temperature, and turbulence (including mixing height) affect the concentration, disper- sion, and transport of atmospheric pollutants which, in addition to higher emission rates, contributes to poor air quality in cities (Grimmond et al., 2010b). Numerous studies have focused on predicting and reducing urban air pollution, yet deficiencies in both observational and modeling capacity are still evident (NRC, 2004). Ozone and particulate matter (up to 2.5 and 10 micrometers; PM2.5 and PM10) concentrations still exceed the National Ambient Air Quality Standards (NAAQS) in many metropolitan areas, and millions of people in the U.S. live in so-called non-attainment areas (EPA, 2012). Beyond conventional air quality concerns, accidental and intentional releases of chemical, biological, or radiological pollutants in cities remain as potential threats and pose particular challenges. Emergency-response

OCR for page 57
66 66 URBAN METEOROLOGY forecasting necessitates models that capture essential features of urban flow and dispersion processes and provide fast exposure predictions. The NRC (2003b) concluded that no model system exists that fulfills all critical requirements for emergency response. Aerosols affect climate, directly and indirectly, through radiative forcing (Kaufman et al., 2005). The “direct” radiative effect of aerosols is to scatter, reflect, or absorb solar radiation. Most aerosols, including sulfates found in urban environments, promote a cooling effect in the radiative budget; however, carbon-based aerosols absorb solar radiation and may warm the atmosphere and surface. Such warming can affect the atmospheric stability profile and thereby alter cloud and precipitation development. Climate- aerosol interactions are quite complex and beyond the scope of this discus- sion, but it is clear from emerging literature that the negative and positive effects associated with the urban production of aerosols must be placed in the context of scale: local (or urban), regional, and global. For example, aerosols augment UHI-effects (mainly in the surface or boundary layer) on temperature through direct interactions with solar radiation (Jin et al., 2011). Anthropogenic aerosols also act as condensation nuclei or “seeds” for cloud microphysical processes (Rosenfeld et al., 2008). This so-called “indirect effect” of aerosols further perturbs the radiation budget, cloud distribution, and precipitation variability. Souch and Grimmond (2006) and Grimmond et al. (2010b) provide very comprehensive assessments of current knowledge, gaps, and needs within the urban weather climate community. Expanding on these assessments, the current state of urban observational and modeling techniques in research and operations, emerging technologies, and remaining needs and challenges for urban meteorology were discussed at the workshop. The following sec- tion summarizes these discussions and relevant literature. ADVANCES IN URBAN FORECASTING AND MONITORING TECHNIQUES Meteorological observations and modeling are tightly coupled and require continual emergence of new understanding, measurements, and technology (Dabberdt abstract in Appendix A). The measurements needed for particular meteorological processes are typically a function of the spatio- temporal scale of the process, the latency requirements of the application, and technological capacity. Shorter-range forecasts (< 60 minutes) may rely on heuristic methods involving extrapolation, but beyond this time period

OCR for page 57
SCIENCE AND TECHNOLOGY 67 (and even inclusive of it), numerical weather prediction is the primary mechanism for forecasting. These models require detailed four-dimensional representation of the atmosphere provided by surface in situ measurements, surface-based remote sensing systems, upper air soundings, and satellite data (often through data assimilation techniques). In cities, meteorological observations and forecasting become even more complex because of the high spatial variability, unique physical characteristics of the urban canopy and its impacts on various processes, and challenges with model initialization. Although such challenges exist, over time there have been several key advances in urban forecasting and monitoring. Some of the key advances that have emerged in the past 10-20 years in the obser- vational and modeling communities are highlighted here. Monitoring and Observations Urban Campaigns The need for urban data sets has been recognized by the scientific community. A number of major field campaigns have been successfully completed in the United States (e.g., Salt Lake City, Allwine et al., 2002; and Oklahoma City, Allwine et al., 2004) and Europe (e.g., London, UK, Ar- nold, et al., 2004; Basel, Switzerland, Rotach, 2005; and Marseille, France, Mestayer et al., 2005). Grimmond (2006) provides an overview of progress in urban observations which includes references to additional studies. Most of the urban observation studies were short-term measurement campaigns that provided data sets for a limited range of environmental conditions. Im- proving models to predict dispersion of hazardous material within the urban atmosphere was the major objective for many of these studies (Hanna et al., 2007; Hanna and Baja, 2009), but the data sets have also served for evalua- tion studies (Liu et al., 2006; Lemonsu et al., 2009) of urban surface-energy balance parameterizations and urbanized mesoscale models. In addition, these studies have provided new insights about the structure of the UBL, the impact of atmospheric stability on mean and turbulent processes within the UBL, and fundamental properties of turbulence within the RSL (Christen et al., 2007; Klein and Clark, 2007; Nelson et al., 2007b, 2011). These efforts further document the significance of urban observations in the urban surface layer and the whole boundary layer for advancing research and operations in urban meteorology. It is critical that efforts be undertaken to sustain and ease access to these datasets and to promote initiatives for future urban studies.

OCR for page 57
82 82 URBAN METEOROLOGY and satellite data (Carter et al., 2011; Jeyachandran et al., 2010). Martilli (2009) discusses the derivation of input parameters for urban canopy models from such data sets. The National Urban Database and Access Portal Tool (NUDAPT; Ching et al., 2009a, b), developed by the Environmental Protec- tion Agency, is an important effort to provide building datasets, including computed input parameters for the urban canopy models (such as rough- ness length, building area fraction, and mean building height for each grid cell), for major U.S. cities. Such a database is an important resource for the urban modeling community. It is also important that efforts be undertaken to secure access and future development of NUDAPT. Coupling of Atmospheric Models from Global Down to Urban Scales Grimmond et al. (2010a) noted that most operational and climate mod- els still fail to resolve urban areas and their associated atmospheric impacts. Jin and Shepherd (2005), Jin et al. (2007), and Oleson et al. (2008a, b) have argued that emerging climate modeling systems must adequately represent urban processes. McCarthy et al. (2010) discussed results from global cli- mate simulations after an urban land surface scheme (Best et al., 2006) was implemented within the Hadley Centre Global Climate Model (HadAM3, Pope et al., 2000). The notion of “convergence” is a critical issue. Global climate model (GCM) spatial resolution is rapidly improving (that is, grid cells are becoming smaller) with increasing computational capacity, while urban footprints are enlarging. This leads to a convergence effect whereby urban-related interactions will become increasingly relevant for proper characterization of atmospheric circulations, fluxes, and weather and cli- mate prediction. Lamptey (2010) established that current and future climate modeling at regional scales will have to properly resolve urban contributions and modifications to the surface energy equation. “Urbanization” of GCMs and regional climate models will involve pa- rameterization of urban processes in the land surface components of most modeling systems. The models are most effective when they are simple enough to ensure structural compatibility and computational efficiency (Ole- son et al., 2008a), yet complex enough to capture key urban-atmosphere interactions. Current urbanized GCMs resolve the spatial attributes of the urban landscape as well as the three-dimensional morphology (“urban can- yon structure”). This construct captures many of the radiative, dynamical, and flux processes adequately. Jin and Shepherd (2005) offered a discussion of how emerging satellite data may provide critical information on urban surfaces (e.g. emissivity, temperature, albedo, vegetation fraction) for GCMs.

OCR for page 57
SCIENCE AND TECHNOLOGY 83 Oleson et al. (2008b) found that of the atmospheric and surface conditions they considered in their study, heat storage and sensible heat flux were most sensitive to uncertainties in the input parameters. They recommend that “attention be paid not only to characterizing accurately the structure of the urban area (e.g., height-to-width ratio) but also to ensuring that the input data reflect the thermal admittance properties of each of the city surfaces.” Advanced Exposure Assessments Ambient concentrations from centrally located monitoring stations have been widely used as exposure surrogates (Sarnat et al., 2001; Burke et al., 2001; Ozkaynak et al., 2008). However, errors may occur depending on the study design and scale at which individual exposure to air pollutants from central pollution monitors are analyzed (Gamble and Lewis, 1996; Zeger et al., 2000; Brauer et al., 2008; Sarnat et al., 2006). The characterization of exposures in an epidemiology study may be improved by the use of high-resolution air pollution models as the basic input for human exposure models, along with integration of human factor data (Burke et al., 2001; Ott et al., 1988, Zartarian et al., 2000). There has been a large number of studies that have characterized empirically intra-urban spatial gradients in urban air pollution using saturation sampling and descriptive statistical models (e.g., Henderson, et al., 2007). These approaches have been necessitated by emissions and meteorology data that are not well resolved spatially and by the limitations of deterministic models. In their review of models used to determine air pollution exposures, Jerrett et al. (2005) concluded that integrated dispersion and meteorologi- cal models are becoming more widely used to assess health effects related to air pollution. However, because classical Gaussian dispersion models, which often predict imprecise urban concentration patterns, are enshrined in government regulations and policies, they are most widely used for such dispersion-model-based exposure assessments. Jerret et al. describe “unreal- istic assumptions about pollutant transport” as a major limitation for health studies based on such models. Although coupled mesoscale meteorological and chemical transport models such as the CMAQ (Byun and Schere, 2006) deliver more reliable concentration predictions for a wide variety of different species of gaseous and aerosol pollutants, the intra-urban variability of concentrations is often not well represented with such models, as the typical spatial resolution in the smallest grid is 1-4 kilometers. Isakov and Ozkaynak (2008) proposed a hybrid approach in which local impacts of mobile and stationary sources

OCR for page 57
84 84 URBAN METEOROLOGY are modeled with the Atmospheric Dispersion Modeling System (AERMOD) (Cimorelli at al., 2005), while the regional transport and background con- centrations are simulated with CMAQ. Application of Weather and Climate Models for Urban Planning At scales ranging from global to local, models are providing guidance on mitigation and adaptation decisions. Recent studies (Oleson et al., 2010a; Akbari et al., 2009) have presented compelling evidence that increasing the albedo of cities on a global scale can not only reduce summertime tempera- tures but also offset carbon emissions through reduced power demand. To take advantage of this effect, some cities are deploying large scale greening plans (e.g., MillionTreesNYC; see Chapter 2). Models are needed to help quantify the expected benefits and identify unintended impacts of these programs. For example, by how much will planting trees and/or creating urban green spaces reduce urban temperatures (Lynn et al., 2009; Zhou and Shepherd, 2009; Solecki et al., 2005)? Better vegetation models are thus required in the current urban canopy models— a need that is not only of relevance for urban planning studies but also for improving urban canopy parameterizations and weather predictions in general. Recent efforts to couple building energy models with urban canopy parameterizations implemented within NWP models (Salamanca et al., 2010, Salamanca and Martilli, 2010) provide better tools to study feedbacks between building energy use and urban climate, as well as to test the impacts of various urban growth scenarios on future urban climate. However, these models require further validations, and the output was shown to be quite sensitive to the details in information about urban structures and morphol- ogy (Salamanca et al., 2011). Coupling of dynamic urban meteorological models with socioeconomic models is important for long-term planning and prevention in areas of high vulnerability. Risk assessment ultimately needs this type of coupling and will allow taking a more proactive approach. The urban modeling com- munity has not inspired confidence among end user groups concerned with sustainability of cities in the ability to accurately model the complex effects impacting quality of life issues in the urban environment—the more complex models considering street, road, and building impacts have not consistently been able to improve on simple model forecasts.

OCR for page 57
SCIENCE AND TECHNOLOGY 85 Data Assimilation and Probabilistic Forecasting Techniques In general, data assimilation and probabilistic forecasting techniques represent clear advancements over the last decade, but their application for urban areas is still largely unknown. These techniques have been used primarily for research applications rather than for end users. This is partly because of the need for greater computing power, but also because end us- ers are not typically educated on how to properly interpret model-generated probabilistic information, and modelers struggle with how best to com- municate it. Data Assimilation Liu et al. (2006) evaluated a multi-scale, rapid-cycling, real-time, four- dimensional data-assimilation and forecasting system to assimilate high density data collected during the Joint Urban 2003 field project in down- town Oklahoma City. Using a mesoscale modeling framework, they found improved characterization of the boundary layer, atmospheric dynamics, and thermal structure, which led to reduced biases in forecasting wind speeds and a more realistic boundary layer structure. Baklanov et al. (2009) have noted the need for improved assimilation of surface characteristics into urban scale NWP and air quality models. Even though some high-resolution satellite or airborne datasets are available for urban landscapes, a new gen- eration of algorithms for assimilation of surface temperature, albedo, snow, and other key urban variables will be required. A generation of two-and three-dimensional land information systems (LIS) have emerged that can be used to assimilate urban surface features into appropriate coupled modeling systems. Kumar et al. (2006) have described a land surface modeling framework that “integrates various community land surface models, ground measurements, satellite-based observations, high performance computing, and data management tools.” They also use a sequential data assimilation extension of their LIS that uses many land surface models, observational sources, and assimilation techniques. LIS is one of several types of various land data assimilation systems (LDAS) that are emerging for constraining land-atmosphere simulations involving weather and climate models (Ghenta et al., 2011; Jimenez et al., 2011; Bosilovich, 2008; Rodell et al., 2004). Chen et al. (2007) present an evaluation of the uncoupled high-resolu- tion land data assimilation system (HRLDAS) which was developed to initial- ize land-state variables of the coupled WRF-land surface model (Noah) for

OCR for page 57
86 86 URBAN METEOROLOGY high-resolution applications. An offline approach was chosen because soil moisture spin up (to reach the equilibrium state) can take up to several years, which is difficult to handle within the computationally expensive WRF. For urban applications, HRLDAS is urbanized by running the coupled Noah/ur- ban model in an offline mode, which then provides the initial soil moisture, soil temperature, snow, vegetation, and wall/road/roof temperature profiles (Chen et al., 2011). Ross et al. (2009) have developed a land information system method for assimilating heterogeneous spatial and georeferenced information into three-dimensional urban models. Many opportunities for enhanced assimilation exist within the air quality forecasting community, such as the NAQFC (described earlier in this chap- ter). Pagowski et al. (2010) have recently demonstrated the role that assimi- lation of surface ozone and fine aerosols can have on air quality forecasts. They applied the WRF-CHEM model and Grid-point Statistical Interpolation, a three-dimensional variational (3D-Var) assimilation tool. This relatively simple approach in chemical data assimilation of ozone and fine particulate matter leads to improved skill of the chemical model forecasts. Other studies using the four-dimensional variational data-assimilation framework (4D-Var) and ensemble Kalman filter (EnKF) data-assimilation frameworks and strate- gies show equal promise for future air quality forecasting (Constantinescu et al., 2007a-d; Elbern and Schmidt, 1999 and 2001; Elbern et al., 1997). Probabilistic Forecasting Dabberdt and Miller (2000) illustrated the direction in which probabi- listic simulations of air quality can aid the user community. Delle Monache et al. (2006) presented results from ensemble ozone forecasts using both meteorology and emission perturbations. Wilczak et al. (2006) conducted a similar study but used seven different models, with their own meteorology, emissions, and chemical mechanisms, to create the ensembles. While both studies concluded a gain in forecast skill through the ensemble approach, they also identified a number of issues. Delle Monache et al. (2006) con- cluded that complex relationships between perturbations to ozone precur- sors and meteorological drivers can result in systematic forecast errors. A survey of the literature suggests that at the urban scale, many ap- plications of probabilistic forecasting methods are primarily found within the hydrometeorological context (Villarini et al., 2010b; Schellart et al., 2009). A deterministic forecast provides the user with an “illusion of cer- tainty” (Villarini et al., 2010b). In a recent discussion of probabilistic flood forecasts, Villarini et al. (2010b) cited an American Meteorological Society

OCR for page 57
SCIENCE AND TECHNOLOGY 87 policy statement on flooding, noting that an important research challenge was “quantifying forecast uncertainty by providing probabilistic forecast guidance” (AMS, 2000). A 2008 policy statement addressing probabilistic forecasting (AMS, 2008) noted several challenges that must be overcome before the user community can take advantage of probability forecasting, many of which are directly applicable to the urban meteorology problem. Advanced Sensing Techniques for the Atmospheric Boundary Layer Although there have been important recent advances in observation techniques that have resulted in clear improvements of weather forecasts, there is an obvious need for better observations, particularly to obtain high- resolution profiles of atmospheric variables within the atmospheric/urban boundary layer. For the emergency response community, timely, high spa- tial and temporal information in and above the urban canopy (i.e., within minutes at sub-km horizontal resolution) of meteorological fields impacting transport and dispersion of hazardous contaminants in an urban environ- ment (i.e., vertical shear of winds, temperature, and moisture) is essential. There are a number of new and emerging sensing techniques, including differential absorption lidars, which provide promise for urban boundary layer/planetary boundary layer (UBL/PBL) profiling of moisture and bound- ary layer structure at markedly reduced cost (perhaps an order of magnitude lower than previous norms). These lidars are currently being developed at the University of Montana in collaboration with NCAR and also in the commercial sector. Micropulse lidars and ceilometer instruments provide low-cost profiling technologies for aerosol structure within the PBL and can be used to detect the mixed layer height and structure. Studies (Emeis et al., 2005) have been done to successfully demonstrate their value, but they have largely not been used for these purposes in operational settings in the U.S. Doppler-lidar systems can provide high-resolution planetary bound- ary layer wind profiles. Sodars, which are quite a bit cheaper, are often of limited use for urban applications due to concerns about noise pollution caused by the instrument and also background noise contamination of the sodar signals. TAMDAR (Tropospheric Airborne Meteorological Data Report- Tropospheric Airborne Meteorological Data Report- ing) and MDCARS (Meteorological Data Collection and Reporting System) aircraft-based samplings are promising ways of obtaining medium-resolution PBL observations in and around urban areas, albeit at nonuniform sampling intervals throughout the day.

OCR for page 57
88 88 URBAN METEOROLOGY Non-Traditional Sensor Networks The unique form, population density, and attributes of urban environ- ments make them particularly challenging from the standpoint of traditional meteorological observations. However, the same challenges associated with urban environments present opportunities as well. Nontraditional “human or social” monitors may provide critical real-time meteorological, hazard, or emergency response data through media such as Twitter, Facebook, YouTube, and text messaging alerts (RSS [Really Simple Syndication] or SMS [Short Message Service] feeds). Personal contributions to meteoro- logical assessment and prediction are common. For example, the Weather Channel now uses an interactive twitter feed.3 BreakingNews, an MSNBC. com property, consolidates verified updates from various news sites, wire services and social networks, allowing people to easily log onto one site and see what other traditional and nontraditional news organizations are reporting.4 Many public meteorological reports consist of input from voice over phone systems, amateur radios, and controlled-access websites (Ferree et al., 2009). Ferree et al. also note the potential value of social media to NWS forecasters or emergency managers; yet they also note key challenges or gaps in effective integration into the assessment and deci- sion making process. Key questions that the use of nontraditional sensor networks raises in- clude the following: (1) Are there ways to aggregate real-time information from public sources with at least some quality control? If so, then how is this information distributed in a timely manner to official sources for emergency response, the broadcast media, the general public, and back to those us- ing these social media? (2) How is the information filtered, and by whom? (3) What are the roles of the NWS and the private sector in gathering and redistributing such information? Mass (2011) documented the great value in geotagged information avail- able from social media sites for nowcasting of weather and for providing warnings. Mass also noted the rapid proliferation of smartphone applications (“apps”) and potential roles that they play in urban meteorological dissemi- nation and emergency response. He noted that several available or emerging apps leverage Global Positioning System (GPS) and cell phone tower infor- mation to determine location specific information. Next-generation apps might deliver site- or condition-specific forecasts or nowcasts. 3 ttp://www.weather.com/social/national. h 4 ttp://www.breakingnews.com/. h

OCR for page 57
SCIENCE AND TECHNOLOGY 89 These capacities are primarily enabled by smart technologies such as GPS or mobile GIS capabilities, which are fairly ubiquitous in urban regions. Together, the combination of social media and personalized technology (via GPS receivers, smartphones, etc.) has ushered in the potential era of dynamic, ever-evolving “personal sensor webs.” Nontraditional sources, however, are not limited to personal space. Other data sources that could be of value for urban meteorological forecasting, hazard response, or dissemination include commercial aircraft, road vehicles, traffic monitoring systems, electronic reader boards, among others. Weather is the number one cause of nonrecurrent traffic congestion, which leads to spikes in greenhouse emissions. A relatively new U.S. Depart- ment of Transportation initiative on Connected Vehicles offers the promise to obtain atmospheric state information (e.g. temperature, pressure) from state, fleet, and possibly private citizen passenger vehicles (Mahoney et al., 2010; Drobot et al., 2010). With roughly 250 million vehicles on the nation’s roads, most of which are in urban areas, this system could be a paradigm shift in our ability to monitor the surface atmosphere. REMAINING NEEDS AND FUTURE CHALLENGES Despite the many emerging meteorological observation and forecasting technologies described here, several key needs, challenges, and opportuni- ties have been identified during the course of this study: The “Urban” Signal and Climate There has been inadequate emphasis on providing the necessary obser- vations and modeling activities in the urban environment that are required to meet the needs of a broad cross-section of users of urban meteorology in- formation. In the quest to create unbiased global-scale temperature records, meteorological services around the world routinely adjust the meteorologi- cal record to exclude the urban heat signal from climatological records (Karl et al., 1988; Peterson, 2003), and monitoring stations affected by urban development are being relocated to remote rural areas. Given that the type of data required to study and serve the needs of cities is mismatched with what is currently routinely monitored, it is important that the complemen- tary functions of global and urban climate records be recognized. It is also optimal to have quality-controlled long-term urban observations. In addition, there should be a two-way interaction between broader green- house-gas based climate change and urban climate change communities.

OCR for page 57
90 90 URBAN METEOROLOGY Stone et al. (2010) have revealed that large urban regions are warming faster than smaller cities and rural regions. This raises critical questions about whether adaptation and mitigations strategies are properly scaled to address this “hyper-warming” in large cities. Cities are also poorly characterized (if they are included at all) in GCMs, yet urban footprints and associated aerosol loads are growing as GCM grid size is decreasing. Früh et al. (2010) have recently explored mechanisms to downscale GCM data to the urban scale using a cuboid method. In their method, urban heat load and the frequency of air temperature threshold exceedances were simulated using eight microscale urban climate runs for each appropriate wind direction as well as the time series of daily meteoro- logical parameters either from regional climate projections or observation. While experimental, this methodology rightly notes the need to “downscale” climate impacts onto the “urban climate” signal. It is also essential to address how the urban climate signal scales up to impact regional to global climate in terms of temperature, precipitation, and cloud systems. Integration of Research Knowledge into Operational Framework Although the problem of converting what is learned in research into on- the-ground applications is common (i.e. the research to operations “valley of death” problem), the committee recognizes some opportunities within the urban meteorological community. For example, research has established that the shape and size of a city and the prevailing wind can affect the orientation of the UHI signal (Basara et al., 2010). With such knowledge, forecasters could provide very detailed minimum temperature forecasts, for example, around a city. Likewise, the literature has clearly established that precipitation and convective activity may be affected by urban regions. Most operational organizations are not aware of this research or may be skeptical, even though an emerging consensus is developing (Shepherd et al., 2011) on urban rainfall effects. This is complicated by uncertainty on the sign of the effect and under what conditions the effects are most evident prompting the need for further research. Dynamics of Cities Cities evolve constantly, and modeling and monitoring systems need to be flexible enough to adjust to these changes and input parameters need to be constantly updated. Challenges for the future will be (1) how to address

OCR for page 57
SCIENCE AND TECHNOLOGY 91 current deficiencies in model frameworks, physics, and dynamics while avoiding a plethora of poorly validated models that are not compatible or easily implemented in operational settings because of their complexity and the number and type of input parameters needed, (2) advancing air chemistry models and better coupling with weather models to improve predictions of particulates and other harmful atmospheric pollutants, and (3) implement- ing end-to-end physical social models with appropriate feedbacks and error characterization.

OCR for page 57