5.1 URBAN DISPERSION – CHALLENGES FOR FAST RESPONSE MODELING · URBAN DISPERSION – CHALLENGES FOR FAST RESPONSE MODELING Michael J. Brown Los Alamos National Laboratory, Los Alamos,

URBAN DISPERSION – CHALLENGES FOR FAST RESPONSE MODELING

Michael J. BrownLos Alamos National Laboratory, Los Alamos, New Mexico

1. Introduction

There is renewed interest in urban dispersionmodeling due to the need for tools that can beused for responding to, planning for, andassessing the consequences of an airbornerelease of toxic materials. Although not aneveryday phenomenon, releases of hazardousgases and aerosols have occurred in populatedurban environments and are potentiallythreatening to human life. These releases maystem from on-site accidents as in the case ofindustrial chemical releases, may result duringtransport of hazardous chemicals as in tankertruck or railroad spills, or may be premeditatedas in a chemical, biological, or radiological(CBR) agent terrorist attack.

Transport and dispersion in urban environmentsis extremely complicated. Buildings alter theflow fields and deflect the wind, causing updraftsand downdrafts, channeling between buildings,areas of calm winds adjacent to strong winds,and horizontally and vertically rotating-eddiesbetween buildings, at street corners, and otherplaces within the urban canopy (see review byHosker, 1984). Trees, moving vehicles, andexhaust vents among other things furthercomplicate matters. The distance over whichchemical, biological, or radiological releases canbe harmful varies from tens of meters to manykilometers depending on the amount released,the toxicity of the agent, and the atmosphericconditions. As we will show later, accounting forthe impacts of buildings on the transport anddispersion is crucial in estimating the traveldirection, the areal extent, and the toxicity levelsof the contaminant plume, and ultimately forcalculating exposures to the population.

i) Why fast urban dispersion modeling?

Fast running models are essential forvulnerability studies where many cases must besimulated in a limited amount of time or foremergency response scenarios when an answeris needed quickly.

Most emergency response dispersion modelscurrently in use have little or no building"awareness" and could be misused for urbanapplications in a way that could lead to fatalconsequences. Fast response models willactually get more use day in and day outperforming vulnerability studies or in trainingapplications. Fast response models are neededfor planning and assessment, where many CBagent attack scenarios must be run. They couldbe used in next-generation training withunscripted table-top exercises and provideimmediate feedback. Fast response modelscould be run around the clock at a military baseor high-profile target (e.g., the DC Mall)ingesting local wind measurements in real-time.Fast response urban dispersion codes have alsobe used for determining the optimal placementof CBR agent sensors around buildingcomplexes (e.g., Streit et al., 2004). A fastmodel is needed because thousands of casesneed to be run in order to determine where toplace sensors around a building complex tomaximize the probability of detecting a CB agentattack.

There is a great need - and opportunity - todevelop fast running urban dispersion modelsthat explicitly account for the effects of buildingson transport and dispersion. These modelswould fill a significant void between fast, but lowfidelity conventional plume dispersion modelsand high fidelity, but slow, computational fluiddynamics models. However, due to thecomplexities of flow in urban areas, there arealso significant challenges in developing asimplified fast running model that canadequately account for the impacts of buildings.In this paper, we present an overview ofchallenges for fast response urban dispersionmodeling based on our team’s experience in thisfield. This review is not comprehensive, but ismeant to inform the reader of some of the keyissues with respect to successful implementationof dispersion models in urban areas.

5.1

Corresponding author address: Michael J. Brown, Los AlamosNational Laboratory, Group D-4, MS F604, Los Alamos, NM87545. E-mail: [email protected].

2. Urban Dispersion Modeling Background

In the 70’s and 80’s, there was a strong push todevelop street canyon models for carbonmonoxide hotspot calculations at intersectionsand in street canyons. The APRAC modelallowed for concentrations to be higher on oneside of the canyon as opposed to the other dueto the in-canyon vortex (Dabberdt et al., 1973).Likewise, Yamartino et al. (1989) developed theCanyon-Plume-Box model for computing vehicleemission concentrations in street canyons andutilized a unique segmented Gaussian plumemodel adapted to the in-canyon vortex. In the80’s and 90’s, models that accounted forenhanced mixing due to a single building foraccidental releases at nuclear energy facilitieswere being described in the literature (e.g.,Ramsdell and Fosmire, 1995). Other canyonmodels (e.g., the CAR model (Eereens et al.,1993), the OSPM model (Berkowicz, 2000)),isolated building models (e.g., EPA Prime(Schulman et al., 2000)), and combined models(e.g., ADMS (Robins and McHugh, 2000)) havebeen developed as well to compute localizedconcentrations. Although generally successfulfor their applications, these models were notreally intended for problems beyond a singlestreet canyon or building and, in general, do notcompute wind fields around building complexes.

A range of fast-running urban dispersion modelshave been developed to look at transport overgreater distances. Theurer et al. (1996) modifiedthe Gaussian plume model to account for theplume centerline shift due to channeling. Hall etal. (2000) developed a Gaussian puff modelcalled the Urban Dispersion Model (UDM) foruse on neighborhood to city scales. The puffsinteract with buildings and the model canaccount for building wake mixing and somechanneling. Röckle (1990) derived a uniquemodel that computes flow around buildingsusing empirical equations and massconservation. It has since been incorporatedinto the ABC and ASMUS models and has beenused for dispersion applications at industrialsites (Röckle, 1990). Kaplan and Dinar (1996)report on results using a Lagrangian dispersionmodel linked with their own version of theRöckle model. Our team as well has utilizedthe Röckle concept and made enhancements tothe flow algorithms (e.g., Pardyjak and Brown,2001; Bagal et al., 2004) as well as linked it toan urbanized Lagrangian dispersion model (seeWilliams et al., 2004 this conference). This

package, called the Quick Urban & IndustrialComplex (QUIC) dispersion modeling system,contains a graphical user interface and has beenapplied to neighborhood-scale problems in suchplaces as NYC, Washington DC, Chicago andSalt Lake City (e.g., see Fig. 1).

There are also a handful of groups investigatingthe use of computational fluid dynamics (CFD)models for fast response applications. Ideasrange from coarse resolution simulations usingdrag (e.g., Lim et al., 2001; Chan et al., 2004) tolibrary approaches were a large number ofcases are precomputed and results for specificcases are interpolated from the library (e.g.,Smith and Brown, 2002; Young et al., 2004).

3. Challenges and difficulties

There are both challenges and difficultiesintroduced by the setting (urban areas) and theapplication (dispersal of a toxic contaminantcloud for exposure assessment). The problem ismade more difficult by the constraint of usingfast running dispersion models, i.e., how doesone account for the complexity of urbantransport and diffusion in a simplified model.

This paper is intended to highlight severalchallenges and difficulties and focuses on theimpact of groups of buildings in urban areas.

Figure 1. Plume dispersal calculations by the QUICmodeling system in the vicinity of Madison SquareGarden, New York City. Red dots representcontaminant in the air, while yellow dots representdeposition on street and building surfaces.

release

wind

Excellent reviews on flow around isolatedbuildings and building clusters can be found inCermak (1976), Meroney (1982), and Hosker(1984).

a) Enhanced mixing.

Numerous urban tracer field experiments havefound that near-source dispersion is enhancedin urban areas (McElroy, 1997). For neutralconditions, urban plume spread parameters areabout a factor of two greater than ruralparameters in the lateral direction and a factor of2-5 in the vertical between 1 and 5 km down-wind of the source (Draxler, 1984). However,there is considerable scatter in the data by cityand release location as shown by McElroy(1997) which one could argue is due to localbuilding geometry, relative source location, andmeteorological conditions. Wind-tunnelexperiments have shown that buildingssignificantly impact both lateral and verticaldispersion close to the source and immediatelydownwind (Fig. 2). These experiments indicatethat the arrangement of buildings (e.g., alignedvs. staggered) impact the dispersion rate (e.g.,Davidson et al., 1996; Macdonald et al., 2000).Experiments around nuclear facilities showedextreme sensitivity in the amount of mixing(plume spread) based on wind direction changesand subtle differences in the release location(Start et al., 1977). Accounting for themacroscopic effects of enhanced mixing inurban areas is quite straightforward through theuse of urban plume spread parameters.However, the nature of increased mixing nearthe source depends strongly on the geometry ofthe buildings and their layout, which is moredifficult to parameterize through plume spreadcurves.

b) Importance of plume location & coverage

For applications involving hazardous contam-inant dispersal, we are not only interested inproviding reasonable estimates of the magnitudeof concentration, but also the location of thecontaminant plume. This is because ultimatelythe plume calculation will be used to assessexposures to the population, or to determinewhat areas need to be evacuated, have beencontaminated, and/or need to be cleaned up.Furthermore, in contrast to most air qualityapplications where the maximum concentrationis one of the most important metrics, low valuesof concentration need to be accurately calculat-ed as well because they may significantly impactpopulation exposure assessments. For example,the lethality of low concentrations may not be asgreat, but because the area of the dilute plumecovers more area, it can impact just as manypeople. Plume location and coverage need tobe considered when developing and evaluatingurban dispersion models and pose a verydifficult challenge.

c) Displacement of plume by street channeling.

Experiments have shown that above roof-levelwinds oblique to the main streets can bedeflected at street level and result in a plumeoriginating at street level being shifted offcenterline (e.g., Bächlin and Plate, 1988;Hoydysh and Dabberdt, 1994; MacDonald andEjim, 2002)). As shown in Fig. 3, near the

Figure 3. Top view of wind-tunnel experiment showingtracer being channeled in streets at oblique anglesrelative to the above roof-level wind. From Hoydyshand Dabberdt (1994).

wind

Figure 2. Plan view of smoke dispersal through anarray of staggered cubes (left) and unobstructedfetch (right). From Davidson et al. (1995).

surface the plume can travel up side streets atoblique angles to the above-roof wind and onlyfollows the upper-level wind when it is mixedabove the building tops.

Figure 4 illustrates the importance that channel-ing can have on longer range plume transport.As shown in the two simulations using the UrbanDispersion Model, one without building effectsand the other with building effects, the ground-

level concentrations are significantly displacedfrom one another. With the prevailing wind fromthe west, the case without buildings shows theplume traveling eastward leading to dangerouszones (denoted by the red areas) across theriver to the east. With buildings, the plume firstgets channeled to the northeast and then oncethe material gets above rooftop it travelseastward with the prevailing wind. However, thedangerous zones are now in a completelydifferent place, to the west of the river and to thenortheast of the release point.

d) The direction of winds in street channel.

As the wind shifts from being perpendicular tothe street canyon axis to oblique angles, the flowpattern in the canyon is thought to switch froman in-canyon vortex to channeling. For idealizedstreet canyons, the direction and strength of thechannelized wind is often assumed to beproportional to the component of the wind aloftparallel to the street canyon (e.g., Yamartino etal., 1989). But for more complicated buildingarrangements (i.e., more realistic) theseassumptions may not hold. Tall buildings candeflect the flow and result in ground-level windsin the opposite direction to that expected. Forexample, Fig. 5 shows a simulation with aprevailing wind from the southeast. In the east-

Figure 4. Simulations of plume dispersion in downtownPortland using the Urban Dispersion Model without(top) and with (bottom) buildings near the releaselocation. With prevailing westerly winds, the plume isdisplaced to the northeast due to channeling when theinfluence of buildings are considered.

Figure 5. Simulation of winds in downtown Salt LakeCity using the QUIC model. For southeasterly inflow,the tall building deflects some of the flow downwardsinto the street canyon resulting in westerly flow nearstreet level, opposite to what is normally expected.

Tall building

westerly flow

west running canyons, the flow at street level isexpected to be from the east. The results showwinds in the opposite direction in one of theeast-west running streets. This occurs becauseof flow blockage due to a tall building on thedownstream side of the canyon which deflectsthe flow downwards into the street canyonresulting in westerly flow. This exampleillustrates that simple rules for channeling maynot work in areas where building heights varysignificantly. Models may need to keep somephysics, e.g., mass conservation, in order toaccount for flow distortion and deflection byrealistic building layouts.

e) Channeling – an intermittent phenomenon?

Channeling of winds at street level by buildingsmay be an intermittent phenomenon. That is,channeling may turn on and off and switchdirections on short time scales resulting fromsubtle changes in the prevailing winds. Figure 6shows measurements from a street canyon indowntown Oklahoma City where the windsperiodically switch directions. This samephenomenon has been found in wind time seriesmeasured in a New York City street canyon(Reynolds, 1994). If this proves to be a commonfeature of street canyon winds, this will bedifficult to account for in fast response modelsdue to the intermittent nature of the process andthe difficulty of specifying above-roof winds.

f) Conditions of in-canyon vortex formation.

Many studies have confirmed the existence of amean flow in-canyon vortex when the ambientwind is close to perpendicular to the streetcanyon axis (Fig. 7). The in-canyon vortexrotates vertically with an axis parallel to thestreet canyon and has been found to influence

within street canyon concentrations andventilation out of the street canyon (e.g., Georgiet al., 1967; Hoydysh et al., 1974). The literaturesuggests that the in-canyon vortex maydisappear (or at least be secondary inimportance to channeling) when the above-roofwind is within 30 degrees of parallel to the streetcanyon axis (Ludwig and Dabberdt, 1972;Yamartino et al, 1989). It is likely that this angleis dependent on street canyon length, the widthof the street canyon and side streets, and theheight (and relative heights) of buildings, amongother things. For a street canyon in Chicagowith a height-to-width ratio of about 1.4, Depauland Sheih (1986) also found that the in-canyonvortex does not exist below a roof-level windspeed of 1.5-2.0 m/s. The conditions underwhich the in-canyon vortex forms and does notform needs to be accounted for in fast responsedispersion models in order to accuratelycompute concentration distributions within streetcanyons.

g) Deep-canyon mixing.

Computational fluid dynamics modeling (e.g.,Mestayer et al., 1995; Baik and Kim, 2000) andlaboratory experiments (e.g., Baik et al., 2000)suggest that there may be counter-rotatingvertically-stacked vortices in deep streetcanyons. The stacked vortices alter the natureof vertical transport and dispersion. However,to date, there has been no confirmatoryevidence of this in real street canyons. Severalpresentations at this meeting may address thisissue (e.g., Eliasson et al., 2004).

h) Real cities vs. idealized building geometries.

Most of our understanding of flow and dispersionin and around building complexes comes fromwind-tunnel experiments where the inflow wind

Figure 6. Time series of wind direction measured ina street canyon illustrating periods of rapid variationin flow direction.

Figure 7. Wind-tunnel measurements showing In-canyon vortex in an array of wide buildings. Flowperpendicular to building face and building height-to-street width ratio of one. From Brown et al. (2000).

is constant and building arrangements aregenerally idealized. For example, the ideas wehave regarding the in-canyon vortex forskimming flow (H/W < 1) and counter-rotatingvortices for wake interference flow (H/W ~ 2/3)originate from wind-tunnel experiments of widebuildings that are of equal height and the meanflow is more-or-less two-dimensional. For a realcity, with buildings of different heights andshapes, and spacing between buildings varyingin space (see Fig. 8, for example), it may behard to find regions where simple straight-forward in-canyon vortex flow exists. That is,the flow field will most likely be a complexcombination of 2D and 3D vortices (bothhorizontally and vertically rotating) andchannelized flow.

i) Rapid vertical mixing.

It is well documented that a building that sticksup above surrounding buildings can result inenhanced vertical mixing (e.g., Wedding et al.,1977). It is hypothesized that intermittent spiralvortices develop on the downwind side of tallbuildings exposed to the ambient flow. Asshown in the snapshot in Fig. 9, these spiralvortices can transport material from near theground rapidly to building top. This will reducestreet-level concentrations locally near therelease and change the downwind timing of thepassage of the plume. Intermittent phenomenonthat are non-local in nature are difficult toaccount for in fast response models. For aGaussian puff type model, one commonapproach would be too force the puff to be equalto the building height when it interacts with the

building. However, because it is an intermittentphenomenon this would most likely result in anoverestimation of vertical mixing.

j). Upstream transport.

Few fast response dispersion models accountfor upstream transport induced by buildingrecirculation zones. Many experiments haveshown that contaminants can be transported inthe opposite direction of the prevailing wind dueto the eddies and mean flows induced by build-ings (e.g., see Hosker, 1987). This can occuraround isolated buildings when the contaminantgets caught in sidewall, rooftop, and/or cavityrecirculation zones and can be even morepronounced in groups of buildings wherecontaminants can travel upstream in the cavityof one building to the sidewall eddy of anotherbuilding and so forth. As shown in the HIGRADLarge Eddy Simulation in Fig. 10, material fromthe release point is transported upstream inrelation to the ambient wind due to the cavityflow, sidewall recirculation and channelinginduced by the upwind buildings. This figureillustrates how initial lateral plume spread isgreatly enhanced. In some cases, contaminatedareas and potentially exposed population can bedramatically changed by upstream transport.Dispersion models that do not account for thesenear-source effects could provide misleadinginformation if applied in cities.

Figure 8. Building footprints for downtown LosAngeles. Color scale represents building height inmeters.

Figure 9. Snapshot of smoke mixing upwards on thebackside of the building. The smoke release is atstreet-level in the canyon and the prevailing wind iscoming out of the picture perpendicular to the buildingface. This intermittent mixing phenomenon is thoughtto occur due to intermittent spiral vortices that developon the downwind-side of the building. Photo from theUSEPA Meteorological Wind Tunnel courtesy of BobLawson.

k) Intersections.

Winds in intersections and at street ends arethought to be very complicated. Winds at eachof the four corners can all be blowing in differentdirections. This is illustrated in Fig. 11 whichshows wind directions measured within and nearan intersection during the Oklahoma City JointUrban 2003 field experiment. Intermittenthorizontally-rotating eddies that spiral upwardsare thought to exist near the end of the streetcanyon and have been identified in wind tunnelsmoke visualization experiments (e.g., Hoydyshet al., 1974). These corner vortices are thoughtto change the nature of transport and dispersionin street canyons (e.g., Cermak et al., 1974;Hoydysh and Dabberdt, 1988; and Hayden etal., 2002). In addition, subtle changes to thewinds in the intersections may impact the natureof the flow in the adjoining street canyon (e.g.,intermittent channeling, in-canyon vortexdevelopment). Hoydysh and Dabberdt (1998)found that concentrations within the same streetintersection but on opposite sides of the streetcan vary by an order of magnitude. It is clearthat transport and dispersion in intersections andnear street canyon ends are very complicatedand will be difficult to parameterize.

l) Impact of roof features.

Wind-tunnel experiments have confirmed thatsmall-scale building features can influence

transport and dispersion. For example,Rafailidis (1997) showed that peaked roofsresulted in significantly different concentrationsin the street canyon as compared to those withflat roofs and had a stronger influence thanbuilding spacing. Kastner-Klein and Plate(1999) showed that peaked roofs interspersedwith flat roofs – and which was upstream of theother - dramatically changed the concentrationpatterns in the street canyon. A wind-tunnelexperiment by Cermak et al. (1974) showed thata rooftop structure on one corner of a streetintersection changed street-level concentrationsby a factor of two. Few fast response urbandispersion models currently have the fidelity toaccount for these effects.

m) Turbulence specification in the urban canopy.

At some stage, all but the most simple modelsrequire information on the turbulent velocityand/or length scales (e.g., u*, lm). Plume or puffdispersion models often utilize the frictionvelocity to estimate plume spread parameters orto compute velocity profiles, random-walkmodels need to have the normal and shearstresses defined, and computational fluiddynamics models must satisfy boundaryconditions which are often parameterized usinga law-of-the-wall which is u* dependent. Ingeneral, fast response models need aparameterized form of these turbulent scales.Recently, there has been focused research doneon whether similarity theories hold above the

Figure 10. Concentration contours produced byHIGRAD LES illustrating upstream transport at point ofrelease. Dispersion is dramatically enhanced laterallydue to this upwind transport. From DeCroix andBrown (2002).

release

wind

Figure 11. Wind direction measured at six differentpoints in or near the Park Ave. – Broadwayintersection in Oklahoma City. Note that sensors allwithin a street width of one another are measuringvery different wind directions.

urban canopy (e.g., Roth, 2000), but much lesshas been accomplished within the canopy layer.Exceptions include, for example, the work byChristen et al. (2002), Kastner-Klein and Rotach(2004), and Nelson et al. (2004). Further studiesare needed in this area in order to advance thenext generation of urban dispersion models.

n) Gradient-mixing vs. non-local mixing.

Many fast-response dispersion models utilizethe concepts of gradient mixing, either implicitlyor explicitly. For transport and dispersionaround buildings, large eddies can act tophysically transport contaminants from one pointin space to another. For example, in the lee of abuilding, a cavity develops that intermittentlyflops laterally from side-to-side, resulting insignificant mixing of materials released in thedownwind cavity. Or, for a release in a streetcanyon, material intermittently is ejected out ofthe canyon resulting in periods of low and high

concentrations in the canyon (Fig. 12). This“non-local” mixing is not adequately accountedfor by gradient mixing theories. In the world ofCFD, the distinct advantage of large-eddysimulation is the fact that it can account for non-local mixing since it explicitly models theseintermittent vortices (large eddies) that developaround buildings. For fast response modeling,non-local mixing theories, however, are not wellestablished and much uncertainty remains intheir implementation. Initial work on implemen-tation of non-local mixing ideas into random-walk modeling can be found in Williams et al.(2004) at this conference.

o) Tree effects.

In many cities, trees account for a large fractionof the canopy cover and can alter the heat andmoisture budget (e.g., Oke, 1989). In addition,trees can slow down winds through drag, trapcontaminants below leaf level, etc. Tree trunksmay actually increase turbulence intensity nearthe ground (Gayev and Savory, 1999). Anoutdoor dispersion experiment around anisolated building could only be replicated by aCFD model when the drag from a row ofEucalyptus trees were accounted for in thesimulation (Calhoun et al., 2000). In someinstances, fast response models may need toaccount for trees to improve urban dispersionprediction capabilities.

p) Traffic effects.

Moving vehicles create turbulence and may alterthe near-surface winds. Wind-tunnel experi-ments by Kastner-Klein et al. (2003) have shownthat traffic can significantly impact ground-leveland near-wall concentrations within a streetcanyon. Gaussian dispersion models weredeveloped to account for vehicle-induced mixingthrough an initial vehicle wake plume spreadparameter (e.g., Benson, 1992). As discussedin Kastner-Klein and Clark (2004), the schemeof Di Sabatino et al. (2003) is beingimplemented into the QUIC urban dispersionmodel. Although in principal it may be possibleto approximate the effects of traffic ondispersion, it will be difficult to obtain data ontraffic density and average vehicle speed forspecific streets in a city of interest.

q) Real-time behavior.

As alluded to in prior sections, the real-timenature of the flow around buildings is less

Figure 12. Snapshots of concentration contours fromlarge eddy simulation of plume dispersal in streetcanyons. Release is in the 1st canyon near the back-wall and winds are from the left. Material is intermit-tently ejected and injected into street canyons by largebuilding-induced eddies. From Smith et al. (2002).

understood (and more difficult to model) than thesteady-state behavior. For many airbornehazardous release scenarios, intermittentphenomena (e.g., peak concentrations) may beextremely important due to the health, lethality,or flammability properties of particular agents.

r) Atmospheric stability and heating of buildingsurfaces.

Another key issue is atmospheric stability.Although cities have been shown to minimizestability effects, dispersion measurements showthat plume spread is dependent on stability (e.g.,McElroy, 1997). It is not clear whether this is amesoscale feature due to the upstream airadvected into and above the city, or if this ismicroscale feature resulting from different ratesof heating (and cooling) of urban surfaces.CFD modeling studies have shown that heatingof individual building walls radically changes thevertical structure of the flow within a streetcanyon (e.g., Mestayer et al., 1995; Kim andBaik, 1998) and around an isolated building(Smith et al., 2000). Large eddy simulations bySmith (1999) of dispersion during the daytime inthe Washington DC Mall area showed largeimpacts due to heating at the surface. Theplume was lofted into the air by convectiveeddies, whereas when the surface heatingscheme was deactivated the plume travelednear the ground (Fig. 13).

4. Conclusions

Fast running urban dispersion models areneeded for applications involving toxic agentreleases in cities. In order to obtain accuratepredictions of both plume concentrations and

plume location, explicitly accounting for theeffects of buildings is important. The complexflows that develop in and around buildingspresents a serious challenge for fast responsemodel developers, i.e., having a simplified modelwith enough fidelity to account for thecomplexities of transport and dispersion in cities.Specific challenges and difficulties werehighlighted in this paper, including issuesregarding turbulent mixing, mean flow prediction,and building geometry effects. There arecertainly other challenging issues for successfulurban dispersion modeling (e.g., indoor-outdoormodeling, porosity of parking structures,interaction with meteorological scales, impact ofmoisture and rain). A number of groups aremaking progress in this area, but much workremains to be done.

Acknowledgements. This work was supportedunder the Department of Homeland Security’sBiological Countermeasures program.

5. References

Bächlin and E. Plate, 1988: Wind tunnelsimulation of accidental releases in chemicalplants, pp 291-303 in EnvironmentalMeteorology, Kluwer.

Bagal, N., E. Pardyjak, and M. Brown, 2004:Implementation of a rooftop recirculationparameterization into the QUIC fastresponse urban wind model, 5th AMS UrbanEnv. Conf., Vancouver, B.C.

Baik, J.J. and J.J. Kim, 1999: A numerical studyof f low and pol lutant dispersioncharacteristics in urban street canyons, J.Appl. Meteor., 38, 1576-1589.

Baik, J.J., R.S. Park, H.Y. Chun, J.J. Kim, 2000:A laboratory model of urban street-canyonflows, J. Appl. Meteor., 39, 1592-1600.

Benson, P., 1992: A review of the developmentand application of the CALINE 3 and 4models, At. Env., 26B, p 179.

Berkowicz, 2000: OSPM – A parameterizedstreet pollution model, Env. Monitor. &Assess., 65, 323-331.

Brown, M., Lawson, R., DeCroix, D., & Lee R.,2000: Mean flow and turbulence measure-ments around a 2-d array of buildings in awind tunnel, 1 1th AMS Appl. of Air Poll.Meteor. Conf., Long Beach, Ca.

Cermak, J., 1976: Aerodynamics of Buildings,Annual Review of Fluid Mechanics, vol. 8,75-106.

Figure 13. Iso-surfaces of plume for case of nosurface heating (top) and surface heating turned on(bottom) in HIGRAD large eddy simulation of a linesource release (red line).

wind

Cermak, J., D. Lombardi, and R. Thompson,1974: Applications of physical modeling tothe investigations of air pollution problems inurban areas, 67th APCA Meeting, Denver,CO.

Chan, S., T. Humphries, and R. Lee, 2004: Asimplified CFD approach for modeling urbandispersion, Symp. on the Urban Zone, 84th

Annual AMS Meeting, Seattle, WA.Christen, A., R. Vogt, M. Rotach, and E. Parlow,

2002: First results from BUBBLE I: Profilesof fluxes in the urban roughness sublayer,4th AMS Urb. Env. Symp., Norfolk, VA.

Dabberdt, W., F. Ludwig, and W. Johnson,1973: Validation and applications of anurban diffusion model for vehicularpollutants, At. Env., v 7, pp 603-618.

Davidson, M., K. Mylne, C. Jones, J. Phillips, R.Perkins, J. Fung, and J. Hunt, 1995: Plumedispersion through large groups of obstacles- a field investigation, At. Env., 29, 3245-3256.

Davidson, M., Snyder, W., Lawson, R. & Hunt,J., 1996: Wind tunnel simulations of plumedispersion through groups of obstacles. At.Env., 30, pp. 3715-31.

DeCroix, D. and M. Brown, 2002: Report onCFD model evaluation using URBAN 2000Field Experiment data: IOP 10, Release 1,26 October, 2000, LA-UR-02-4755, 80 pp.

DePaul, F. and C. Sheih, 1986: Measurementsof wind velocities in a street canyon, At.Env., 20, pp. 455-459.

Di Sabatino, S., P. Kastner-Klein, R. Berkowicz,R. Britter, and E. Fedorovich, 2003: Themodelling of turbulence from traffic in urbandispersion models – Part I: Theoreticalconsiderations. Env. Fluid Mech., 3, 129-143.

Draxler, R, 1984: Diffusion and TransportExperiments, Near Obstacles. AtmosphericScience & Power Production, DOE/TIC-27601, D. Randerson, Ed., 367-422. U.S.Dept. of Energy, Wash., DC.

Eerens, Sliggers, and van den Hout, 1993: TheCAR Model: the Dutch method to determinecity street air quality, At. Env., v 27B, p 389.

Eliasson, I., B. Offerle, C.S.B. Grimmond, and S.Lindqvist, 2004: Wind fields and turbulencestatistics in an urban street canyon, 5th AMSUrban Env. Conf., Vancouver, B.C.

Gayev, Y. and E. Savory, 1999: Influence ofstreet obstructions on flow processes withurban canyons, J. Wind Eng. & Ind.Aerodyn., 82, 89-103.

Georgi, H., E. Busch, and E. Weber, 1967:Investigation of the temporal and spatialdistribution of the emission concentration ofcarbon monoxide in Frankfurt/Main, ReportNo. 11, Inst. Meteor. & Geophysics, Univ.Frankfurt/Main.

Gross, G., 1997: ASMUS- Ein numerischesModell zur Berechnung der Stromung undder Schadtstoffverteilung im Bereiche i n z e l n e r G e b a u d e . I I :Schadstoffausbreitung und Anwendung.Meteorol. Zeitschrift, 6, 130-136.

Hall, D., A. Spanton, I. Griffiths, M. Hargrave, S.Walker, 2000: The UDM: A model forestimating dispersion in urban areas, Tech.Report No. 03/00 (DERA-PTN-DOWN).

Hayden, R., W. Kirk, G. Succi, T. Witherow, andI. Bouderba, 2002: Modifications of highwayair pollution models for complex geometries –Vol. II – Wind tunnel test program. FHWA-RD-02-037.

Hosker, R.P. Jr., 1984: Flow and Diffusion NearObstacles. Atmospheric Science & PowerProduction, DOE/TIC-27601, D. Randerson,Ed., Ch. 7, 241-326. U.S. Dept. of Energy,Wash., DC.

Hoydysh, W., R. Griffiths, and Y. Ogawa, 1974:A scale model study of the dispersion ofpollution in street canyons, 67th APCAMeeting, Denver, CO.

Hoydysh, W. & W. Dabberdt, 1988: Kinematicsand dispersion characteristics of flows inasymmetric street canyons. Atm. Env., 22,677-89.

Hoydysh, W. and W. Dabberdt, 1994: Concen-tration fields at urban intersections: fluidmodeling studies, At. Env., 28, pp 1849-1860.

Kaplan, H. and N. Dinar, 1996: A LagrangianDispersion Model for Calculat ingConcentration Distribution within a built-updomain, Atmos. Env., Vol. 30, No. 24, pp.4197-4207.

Kastner-Klein, P. and J.V. Clark, 2004: Modelingof flow and dispersion characteristics intypical building configurations with the fast-response model QUIC, 5th AMS Urban Env.Conf., Vancouver, B.C.

Kastner-Klein, P. and E. Plate, 1999: Wind-tunnel study of concentration fields in streetcanyons, At. Env., 33, 3973-3979.

Kastner-Klein, P., M. Ketzel, R. Berkowicz, E.Fedorovich, and R. Britter, 2003: Themodelling of turbulence from traffic in urbandispersion models – Part II: Evaluationagainst laboratory and ful l -scale

concentration measurements in streetcanyons. Env. Fluid Mech., 3, 145-172.

Kastner-Klein P. and M. Rotach, 2004: Meanflow and turbulence characteristics in anurban roughness sublayer, Bound. LayerMeteor. (in press).

Kim, J. and J. Baik, 1998: A numerical study ofthermal effects on flow and pollutantdispersion in urban street canyons, 2ndAMS Urban Env. Symp., Albuquerque, NM.

Lim, D.W., D.S. Henn and P.A. Hookham, 2001:Preliminary assessment of linked CFD-operational dispersion models on the urbanscale, 3rd Int. Symp. on EnvironmentalHydraulics, Tempe, AZ.

Ludwig, F. and W. Dabberdt, 1972: Evaluation ofthe APRAC-1A urban diffusion model forcarbon monoxide, Final Report, SRI Project8563.

Macdonald, R. and C. Ejim, 2002: Hydraulicflume measurements of mean flow andturbulence in a scale model array of 4:1aspect ratio obstacles, 5th UK Conf. on WindEng., Nottingham, UK.

Macdonald, R., S. Carter Schofield, P. Slawson,2001: Measurements of mean plumedispersion in simple obstacle arrays at 1:200scale, Univ. Waterloo Thermal Fluids Rep.No. 2001-2.

McElroy, J.L., 1997: Diffusion from low-levelurban sources: reexamination using recentlyavailable experimental data, J. Appl.Meteor., 36, 1027-1030.

Meroney, R.N., 1982: Turbulent Diffusion nearBuildings. In Engineering Meteorology (E.J.Plate, ed.), Ch. 11, 481-525. Elsevier,Amsterdam.

Mestayer, P., J. Sini, and M. Jobert, 1995:Simulation of wall temperature influence onflows and dispersion within street canyons,Air Pollution 95, Porto Carras, Greece.

Nelson, M., M. Brown, E. Pardyjak, and J.Klewicki, 2004: Turbulence within and abovereal and artificial urban canopies, 5th AMSUrban Env. Conf., Vancouver, B.C.

Oke, T., 1989: The micrometeorology of theurban forest, Phil. Trans. R. Soc. Lond. B,324, pp. 335-349.

Pardyjak, E. and M. Brown, 2001: Evaluation ofa fast-response urban wind model –comparison to single-building wind-tunneldata, 3rd Int. Soc. Environ. Hydraulics,Tempe, AZ, Dec. 2001, LA-UR-01-4028, 6pp.

Rafailidis, S., 1997: Influence of building arealdensity and roof shape on the wind

characteristics above a town, Bound.-LayerMeteor., 85, 255-271.

Ramsdell and Fosmire, 1995: Atmosphericdispersion estimates in the vicinity ofbuildings, Rep. No. PNL-10286, PacificNorthwest National Laboratory.

Reynolds, M., 2004: personal communication.Robins, A. and C. McHugh, 2001: Development

and evaluation of the ADMS building effectsmodule, Int. J. Env. & Poll., 16, 161-174.

Röckle R., 1990: Best immung derStromungsverhaltnisse i m Bereichkomplexer Bebauungsstrukturen, Ph.D.dissertation, Vom Fachbereich Mechanik,der Technischen Hochschule Darmstadt,Germany.

Roth, M., 2000: Review of atmosphericturbulence over cities, Q. J. R. Meteor. Soc.,126, pp 941.

Schulman, L., D. Strimaitis, J. Scire, 2000:Development and evaluation of the PRIMEplume rise and building downwash model, J.Air & Waste Manage. Assoc., 50, 378-390.

Smith, W. S., 1999: personal communication.Smith, W. S. and M. Brown, 2002: A CFD-

generated wind field library feasibility study:maximum wind direction interval, 4th AMSSymp. Urban Env., Norfolk, VA, May 20-242002, LA-UR-02-1100.

Smith, W. S., M. Brown, and D. DeCroix, 2002:Evaluation of CFD simulations usinglaboratory data and urban field experiments,4th AMS Symp. Urban Env., Norfolk, VA,May 20-24 2002, LA-UR-02-1099.

Smith, W. S., Reisner, J. M., and Kao, C.-Y. J.,2001: Simulations of flow around a cubicalbuilding: comparison with towing-tank dataand assessment of radiatively-inducedthermal effects, Atm. Env. 35(22), 3811-3821.

Start, G., J. Cate, C. Dickson, N. Ricks, G.Ackermann, and J. Sagendorf, 1977:Rancho Seco wake effects on atmosphericdiffusion, NOAA Tech. Memo. ERL-ARL-69.

Streit, G., M. Williams, S. Linger, and M. Brown,2004: The National Cathedral, WashingtonD.C.: BioWatch Site Protection, LA-CP-04-0585.

Theurer, W., E. Plate, and K. Hoeschele, 1996:Semi-empirical models as a combination ofwind tunnel and numerical dispersionmodeling, At. Env., v 30, pp 3583-3597.

Wedding, J., D. Lombardi, and J. Cermak, 1977,“A wind-tunnel study of gaseous pollutantsin city street canyons”, J. Air Poll. ControlAssoc., v 27, pp 557-566.

Williams, M., M. Brown, D. Boswell, B. Singh,and E. Pardyjak, 2004: Testing of the QUIC-PLUME model wi th wind- tunnelmeasurements for a high-rise building, 5th

AMS Urban Env. Conf., Vancouver, B.C.Yamartino, R., D. Strimaitis, and T. Messier,

1989 :Modification of highway air pollutionmodels for complex site geometries, vol. 1,Data analyses and development of the CPB-3 model, Rep. no. FHWA-RD-89-112.

Young, T., J. Boris, S. Cheatham, J. Fulton, C.Lind, K. Obenschain, and G. Patnaik, 2004:CT-analyst: fast and accurate CBRemergency assessment, 5th AMS UrbanEnv. Conf., Vancouver, B.C.

LA-UR-04-

Los Alamos National Laboratory is operated by the University of California for the United States Department of Energy under contract W-7405-ENG-36.

TITLE: URBAN DISPERSION – CHALLENGES FOR FAST RESPONSEMODELING

AUTHOR(S): Michael J. BrownLos Alamos National Laboratory

SUBMITTED TO: 5th AMS Symposium on the Urban EnvironmentVancouver, B.C., August 2004

By acceptance of this article, the publisher recognized that the U S Government retains a nonexclusive, royalty-free license to publish or reproducethe published form of this contribution or to allow others to do so for U S Government purposes.

The Los Alamos National Laboratory requests that the publisher identify this article as work performed under the auspices of the U S Department ofEnergy.

Los Alamos Los Alamos National LaboratoryLos Alamos, New Mexico 87545