Atmospheric Environment Vol. 11. pp. 597-604. Pergamon Press 1977. Printed in Great Britain. URBAN-RURAL WIND VELOCITY ROBERT D. BORNSTEIN and Department of Meteorology, DIFFERENCES* DOUGLAS SCOTT JOHNSON San Jose State University, San Jose, California 95192 (Received 2 August 1976, in final form 9 December 1976) Abstract-Wind speeds along a streamflow line through New York City are found to be decreased below (increased above) those at sites outside of the city during periods with regional wind speeds above (below) about 4 m/s. The decrease is attributed to increased values of the surface roughness parameter in the city, as compared to values in nearby non-urban regions. The increase is associated with accelerations produced by a well-developed urban heat island. Cyclonic turning of the flow along the streamflow line through the city during the high wind speed periods results from decreases in the value of the Coriolis force associated with the increased urban frictional drag. During periods of acceleration over the city, either cyclonic or anticyclonic turning was found, due to a combination of two effects. The origin of the flow was found to influence the direction of the turning over the city. during night-time heat island hours. For easterly flows the origin is a highly stable rural region, while for westerly flows it is most likely one of the less stable, but aerodynamically smooth, water bodies around New York City. INTRODUCTION ( turn in air passing over a city, thus increasing the speed of the flow. Urban areas affect the wind flow pattern and hence - The speed of the undisturbed flow is also important the transport of contaminants in the atmosphere. Few in determining the magnitude of the urban effect, as air pollution studies have had sufficient wind observa- a heat island tends to be weak with high wind speeds tions to establish the time and space variation of the (Bornstein; 1968). For example, Georgii (1968) transport velocity in and around an urban area. reported that during clear, calm nights in Frankfurt, Most investigations into the effects of urban areas urban-induced circulations reached a speed of from on regional flow patterns have used observations 2-4 mls only when the prevailing regional speed was from two towers or two'W eather Service stationsless than 3-4 m/s. Chandler (1965) compared wind one within the city and one at an "airport" site out- speeds from one upwind rural- station and one Lonside of the city. Such studies, e.g., Landsberg (1956), don Ul'ban station, and found that the' nighttime Frederick (1964), and Graham (1968), for New York urban wind speed was greater than the rural speed City, Nashville, and Fort Wayne, respectively, have only when the regional speed was less than 5 m/s. generally shown wind speeds over the urban area to .This paper investigates the effect of regional wind be less than those -outside of it. This decrease has speed, time of day, and regional wind direction on been a.ttributed to the greater aerodynamic surface changes of wind direction, as well as speed, caused roughness of a city compared to its surrounding rural by the roughness and warmth of New York City. environs. These changes are evaluated at various distances Chandler (1965), however, demonstrated that the upwind and downwind of the city, using data from distribution of surface wind speeds in and around a denser network of anemometers than previously London, England, depends on such factors as upwind available. Some of these results have been previously rural wind speed, season, and time of day. During presented by Bornstein et al. (1972) and by Johnson periods with a well-developed "urban heat island" he and Bornstein (1974). . found that the surface wind speed over London was METHOD OF ANALYSIS actually greater than at nearby rural sites. There are two ways in which the urban heat island A mesoscale network of anemometers was estabaffects the wind flow over a city. First, the urban tem- lished in and around New York City as part of the perature excess creates a convergent "sea breeze" type NYUfNYC Air Pollution Dynamics Project of circulation cell, with flow near the surface directed 1964-68. This network, described by Davidson (1967), into the warm city, and with the cell superimposed Druyan (1968), and Bornstein et al. (1976), consisted on any existing mean flow. This effect has been of 97 sites located in a rectangle centered on the west observed by Okita (1960,1965) and Pooler (1963), in side of mid-town Manhattan. The rectangle was Asahikawa (Japan) and Louisville, respectively. 220 km long in the E-W direction and 110 km long Second, the reduced nighttime stability of the urban in the N-S direction. Data from the network were atmosphere increases the downward flux of momenused to study the effects of synoptic features on mesoscale flows (Scudder, 1965) and mesoscale trajec* This work is part of the M.S. Thesis of the second author. tories (Druyan, 1968). 597 598 ROBERT D. BORNSTEIN and DOUGLAS SCOTT JOHNSON The anemometer network consisted of 14 airport stations (National Weather Service and Federal Aviation Administration); 4 military bases (Air Weather Service and Naval Weather Service); 10 Coast Guard bases; 15 utility companies; 14 industrial sites; 29 public agencies and institutions (Public Health, sanitation, schools, etc.); and 11 sites established by New York University. Wind data were collected from each of the sites during twenty observational periods, each three to five days in duration and randomly selected during the years from 1964-1967. The data were averaged over an hour, centered on the hour, except for those from the airport, military, and Coast Guard stations, which were standard hourly synoptic observations. During a shakedown period, wind speeds were corrected to a height of 30.5m by use of power law profiles. However, since the magnitude of the cOITections was generally less than 1 mps, the procedure was discontinued. The hourly-averaged wind speed and direction data were then plotted onto maps, and streamflow and isotach analyses were made (see Fig. 1 for one such analysis). A streamflow line is here defined to be everywhere parallel to the flow; however, the spacing between adjacent lines is not proportional to flow speed as it is with stream lines. When constructing these maps over the complex urban teITain, it was necessary to ensure that the wind at each site was representative of the flow in its vicinity. In addition to missing data (about 25% for a given map), various problems were encountered during these analyses, e.g.,instrument orientation problems, local channeling effects, and inCOITectlyplotted directions. A summary of the various problems encoun74"45' 30' 15' 74°00' tered at each site is given in Bornstein et al. (1976). Most of the problems were transitory in nature, and the vast majority of the data could be included in the analyses, even under low wind speed conditions. The influence of time of day on the urban-rural wind velocity was determined by separately considering nighttime (midnight and 2 a.m.) and daytime (noon and 2 p.m.) observations. At each of these times, on every day in the study, the streamflow line passing closest to the U.S. Weather Service station at Central Park (CP) in Manhattan was determined. The wind speed and direction at the Park were compared with those at distances of ten and twenty miles (16 and 32 km) upwind and downwind from the Park along the streamflow line, if an observation was available within one mile of the specified distance. There were three daytime and four nighttime convergent wind direction cases, in which no single streamflow line passed through New York City, as illustrated by Fig. 1. In these cases the rural wind speed was always nearly calm, the main circulation being induced by the urban heat island. Because of the problem in finding a single streamflow line passing through the city in these cases, they were not included in the analyses. URBAN-RURAL WIND SPEED DIFFERENCES The Central Park station was initially taken to be the "most urban" location, and deviations between the wind speed at the Park and at each of the other locations along the streamflow line were determined. The deviations were averaged for the day and night cases separately, and these values were used to com.', ; 45' 30' . 15' 73°00' 4,°15' 41°15' 4'°00' 41°00' 40°45' 40030' 40°45' 40" 30' c~\...~ , "--2.5 AUGUST 0600 40°15' 74°45' 30 15' 74000' 45 30' 15' 7,1964 E.S.T. 40°15' 73°00' Fig. 1. Analyzed streamflow pattern showing a convergent heat island induced circulation southwest of Manhattan (shaded area). Urban-rural . 10 -------. DAY --- --- J: a.. ::2E B ~ 6 .:::-./"/"\. 20UP 10UP DISTANCE FROM . CP 10 DOWN 20 DOWN CENTRAL PARK (MILES) Fig. 2. Day and night wind speeds at various distances from Central Park. pute the average values for the non-Central Park locations, Overall 599 wind velocity differences Speed Differences The average distribution of the daytime and nighttime wind speeds across New York City is shown in Fig. 2. During the day, the average speed decreases as the wind approaches the city, as might be expected with increasing surface roughness. The surface roughness parameter increases from as little as 0.01 em over mudflats to several meters over built up urban areas (Oke, 1974). However, as the surface roughness is again reduced downwind of the city, the wind speed should return to approximately its upstream rural value. The fact that it does not is due to the afternoon (easterly) sea breeze, which is opposed to the direction of the regional flow on about two-thirds of the study days. Statistical summaries of all of the results in the current study are given in Johnson (1975). The peak in the nighttime curve of Fig. 2 is caused by the nocturnal Urban heat island, the effect of which can overshadow the effect of the increased aerodynamic roughness of the city. The maximum speed occurs at ten miles (16 km) downwind, rather than at Central Park, due to a combination of the large size of the city and the frequent westerly flow direction. As will be explained below, the maximum speed should occur near the downwind edge of an urban area. Eff~ct of Wind Speed Landsberg (private communication, 1971) suggested that there is a critical prevailing regional wind speed below which wind speed is a maximum over an urban area, and above which wind speed is minimum over a city. To test this hypothesis, the magnitude of the upwind rural speed at which the average deviation changed sign was determined. This value was 8-9 mph (about 4 m/s) for the nighttime cases and 7-8 mph (about 3 m/s) for the daytime cases. Thus, the data were separated into high- and low-speed categories, as shown in Fig. 3, where missing points contained fewer than five observations. The high-speed daytime curve parallels the daytime overall curve because. of the frequency of high daytime wind speeds. Conversely, the daytime wind speed maximum oyer the city in the low-speed case is due to accelerations produced by the heat island. In general, the daytime heat island is weak, but it is strongest when the prevailing regional speed is low. The low-speed nighttime curve of Fig. 3 parallels the nighttime overall curve because of the high frequency of low nighttime wind speeds. The wind speed maximum for the nighttime low-speed cases occurs at the same location as in the low-speed daytime case. The' high-speed nighttime decrease in speeds over the city is also similar to that found in the corresponding daytime cases. However, at night this decrease is followed by a sharp increase ten miles (16 km) downwind of the Park. This difference between the night and day high-speed cases arises because the daytime sea breeze is far stronger than the nighttime land breeze. The daytime and nighttime low-speed cases were further divided into two subgroups (0 to 3 and ~ 4 mph, or 0 to 1.4 and ~ 1.8m/s), but only the nighttime subgroups had enough da,ta to be analysed. The maximum speed occurs over the city for all three night low-speed categories (Fig. 4), but the percent increase at Central Park is greatest for the 0-3 mph DAY 12 '~., ". ")~' 10 'ALL" J:S a. ::E ::>6 4 .- '>':::---:"- .~. --" ::~" -- . ./...---. "~/ 14 .~" NIGHT. 12 10 ~"--./ I a. ::EB ::> 6 4 :~:?:?:~" Fig. 3. Day and night wind speeds at various distances from Central Park, grouped according to upwind rural wind speed. 600 ROBERT D. BORNSTEIN and DOUGLAS SCOTT JOHNSON NIGHT 8 /'~ 4-8 6 I 0.. :::;: ::>41 2 ./' ~~' ?~. 20UP DISTANCE .~.' / ~'~'V' 10UP \ CP 10 DOWN 20 DOWN FROM CENTRAL PARK (MilES) Fig. 4. Night wind speed at various distances from Central Park during low-speed conditions, grouped according to upwind rural wind speed. rural speed: 212% at 20 miles (32 km) upwind and 112% at both 10miles upwind 20 miles downwind. The corresponding values for the 4-8 mph (1.8 to 3.6 m/s) cases are only 19, 5, and - 3%, respectively, demonstrating that the urban heat island effect is most intense when the rural wind speed is very small. Effect of Wind Direction tude for the sea breeze. The pronounced maximum' speed at 10 miles (16 km) downwind for the daytime low-speed, west wind cases (Fig. 5) is, of course, due to the daytime urban heat island, while the sea breeze is responsible for the quick recovery to the upstream value at 20 miles downwind. The nighttime east and west overall and low-speed curves of Fig. 6 show that the maximum speed is always found at the downwind edge of the city, the location of which changes with flow direction. The greater smoothness of the curves of Fig. 6, compared to those of Fig. 4, is due to the separation of the data into east and west wind flow cases in the former figure. The large magnitude of the nighttime urban maxima, as compared to the daytime urban maxima of Fig. 5, is due to the stronger nighttime urban heat island. The night-time cases do not show any effect of a land breeze (which would act to cause the upwind rural speed to be larger than the downwind rural speed), because the land breeze is far weaker than the sea breeze. The percent increase at Central Park for the 0-3 mph (0-1.4 m/s) west wind case is again extremely large, i.e., 350% at 20 miles (32 km) upwind, 200% at 10 miles upwind, and 99% at 20 miles downwind. The corresponding values for the 4-8 mph With easterly flow, Central Park is near the downwind edge of the city, but with westerly flow; the edge " " "->8 MPH' WEST 12 of the city is about ten miles (16 km) downwind of ';. Central Park, i.e., in Brooklyn or Queens. The effect of wind urban-rural wind speed differences has been ;",., Ii >.~.,. --. ,. :--',0.' ---....... evaluated by dividing the wind data into westerly and 10 ...... .,.< ,. . '. easterly flow cases:Westerly.cases are those for which the wind direction at the first upwind rural station ~,," .":; .~ .,:", :t: was between 181 and 360°, and easterly cases correa. 8 .: ::;: ~ '" spond to directions between 1 and 180°. These data ::> were used to construct the curves in Fig. 5 and 6, ' in which categories having fewer than five compari/" " 6 sons for two or more locations are again omitted. /' " / '\ The daytime east and west wind curves of Fig. 5 / '\ differ markedly, because the sea breeze tends to re//\ 4 inforce easterly flows. Thus, the daytime east wind 0-7 ./ cases include mostly highspeed cases. The wind speed decreases over the city in the two daytime easterly EAST 13 . - -,,>8 MPH flow curves because of the increased roughness over the city, and the lack of penetration of the sea breeze flow. Since the sea breeze rarely reaches areas down:t: wind of the city at these times, the flow does not a. :::;: completely recover. to the large upwind rural values. :5 \ / However, with a strong westerly flow, the opposing sea breeze prevents the speed at 20 miles (32 km) 9 downwind from returning to the upwind rural value. In fact, the speed at 20 miles downwind during these periods is actually slightly less than at 10 miles downwind. The effect of the sea breeze is also demonstrated by the fact that the speed decrease during high speeds Fig. 5. Day wind speed at various distances from Central for easterly winds is double that for westerly winds. Park for west and east wind cases, grouped according to upwind rural wind speed. The difference implies a 2-3 mph (about 1 m/s) magni- " , ' .---- '. " .~. ", ~'\ / :\, /. ~:V Urban-rural WEST /; ~. 8 ' , -- - 4 8 ALL"" ,,/ ::E "" ° 0 ,,/ 0"'" 0-8 MPH 54 "/" --~:/"" 6 ° :I: a.. --~- '" ,,_0; / , / 2 0-3 .--- / / / I I ", '0 /,-0 / ,\ ,\0 \ " '- ° \ \ II \ \ ° 6 E/.~o :I: a.. ~ ~ 4 ALL 0, "o~ 0-;,0 /,"&-8 MPH 0" 20 UP 10UP "'<~: CP 10DOWN 20 DOWN DISTANCE FROM CENTRAL PARK (MilES) Fig. 6. Night wind speed at various distances Park for west and east wind cases, grouped upwind rural wind speed. from Central according to (1.8-3.6 m/s) case are only 20, 11, and 4%, respectively. URBAN-RURAL WIND DIRECTION to increase with increasing drag (Blackadar, 1962). The southeast to northwest directed sea breeze, found downwind of the city (for the westerly flows making up two-thirds of the present cases), would explain the increased cyclonic turning seen in the figure at that location. In contrast, Angell et ai. (1972), found that downstream of Los Angeles, where the sea breeze effect is generally in the same, direction as the mean flow, daytime flow directions quickly readjust. Effect of Wind Speed \ \ 0/ 601 wind velocity differences DIFFERENCE The deviation of wind direction' 8.Iong each available streamflow line was determined relative to the wind direction at the first upwind rural site (either 10 or 20 miles (16 or 32 km) upwind). A positive deviation indicates anticyclonic turning, while a negative value indicates cyclonic turning. The average deviation at each point downwind (of 20 miles upwind from Central Park) was computed separately for day and night. The average direction deviations were corrected for those cases in which the first upwind rural observation was at 10miles (16 km) upwind instead of 20 miles upwind. The average turning between these two distances was computed (if five or more comparisons were available), and then used to recompute new values of the average deviations along those streamflow lines not originating at 20 miles upwind. Direction changes for the high- and low-speed upwind rural subgroups are plotted in Fig. 8. Similar amounts of frictionally induced cyclonic turning are shown for the first 40 miles of the low- and high-speed daytime cases. However, while the high-speed case shows significant additional cyclonic turning at 20 miles (32km) downwind, the low speed curve does not. For the predominantly westerly flow cases, this turning is due to a passage through the sea breeze front (into the marine air). Why this pattern does not appear in the low-speed cases is unclear, except that perhaps the value at 20 miles do~wind is based on , only five comparisons. The low night-time speed curve of Fig. 8 shows that the air turns anticyclonically as it passes from the very stable rural regions into the less stable urban atmosphere. This is consistent with the idea that cross-isobaric flow angles are reduced as stability is decreased(Yamamoto et ai., 1968):In the high-speed case, the increased friction again causes a turning toward lower 'pressure over the city. However, since the turning at 20 miles (321m) downwirid is less cyclonic than that at 10 miles downwind, recovery to the upwind flow direction has l;Iegun. ,'U Effect 'of Wind Direction The high- and low-speed wind data were further separated into cases with easterly or westerly flow, as was done for the wind speed data. The daytime east and west wind direction curves (Fig. 9) are similar to the curve for the overall daytime data (Fig. 24 20 / 16 " .. ~ 12 oS ~ z 6 t;; 4 , Fig.7, Corn" ve"io~ p'p",on","od wm"" ve"ioo, .. Overall Direction Differences The overall direction differences are summarized in Fig. 7 (in which the wind direction is that relative to 20 miles (32 km) upwind). There is almost no turning over the city for the average night-time case, although there is a small amount of anticyclonic turning downwind of the city. The cyclonic turning over the city for the daytime cases is due to the increased frictional drag associated with the increased urban surface roughness, as cross-isobar flow angle tends 15 ...J .. 0 .. origi",1 / ........ ---- ,""'- NIGHT ~ DAY / I / ,,' "" ,/' // / ffi 4 .... .. ...J 0 "4 ~ --- ----- 20 UP 10 UP CP 10 DOWN 20 DOWN DISTANCE FROM CENTRAL PARK (mile.) Fig. 7. Day and night streamflow lines through New York City. 602 ROBERT D. BORNSTEIN and DOUGLAS SCOTT JOHNSON 28 DAY :; I 24 I ,, 20 16 12 / 8 'E ~0 / / Z .J NIGHT , --- -- ----- ~ LOW ' , ,,~ , :. , -,-, . 20 UP 10 UP CP fO DOWN 20 DOWN DISTANCE FROM CENTRAL PARK (miles) .J --""---'- 20 UP 10 UP CP 10DOWN 20 DOWN DISTANCEFROMCENTRAL PARK (miles) Fig. 8. Day and night streamflow lines through New York City, grouped according to upwind rural wind speed. 7), as both showing cyclonic turning. The sea breeze penetration is seen as a 40° cyclonic shift at 1Omiles (16 km) upwind in the east wind data, while in the west wind case it appears as a 30° cyclonic shift at 10 miles downwind. For the night-time flows of Fig. 9, only the west wind case shows the expected anticyclonic turning over the city consistent with a decrease in atmospheric stability. The east wind case of Fig. 9 shows cyclonic turning consistent with an increase in surface roughness. The difference between the two curves is due to dissimilarities in the air entering New York City from east and west directions. In the former, the air is highly stable, as it is coming from a rural DAY ///" 24 _20 ';16 ]12 UJ / ,// // / Fig. 10. Night streamflow lines for westerly and easterly flows grouped according to upwind rural wind speed. region. In the latter, it is most likely to be less stable than that from the east, as it originates over one of the water bodies around New York City. Hence, for westerly flow the stability difference between the origin and the urban sites is great, while for easterly flow the roughness difference between the origin and the urban sites is great. The near,.zero average turning of the night-time curve of Fig. 7 resulted from averaging the anticyclonic turning of th'e west wind cases and the cyclonic turning of the east wind cases. 'The east and west wind direction data were separated into high- and low-speed wind subgroups. The night-time west.wind, high-speed curve jnFig. 10 shows the expect~d frictionallyjnduced cyclonic turning, while the. single easterly, high-speed. qase is not shown. The night-time west' wind, ,low-speed curve shows the expected stability induced anticyclonic turning over the city, and the corresponding east wind curve shows the expected roughness induced cyclonic turning. The magnitude of the night-time west wind, low-speed anticyclonic turning is greater than that for the overall west wind curve in Fig. 9, as the highspeed cases with their cyclonic turning have been / = / t 24 cO EAST,/ --<i16 ffil2 fj 8 4 0 '* 0r E-4' -' ,/ ~.. --0; .J « 0 0:: ~ 8 « ..J / EAST . 20 UP DISTANCE , , 10 UP CP 8 tn 4 15 ~-8 ..J 4 /-----, I ~ 12 w NIGHT """""'--- ..~ 0'! ---ffi ' I :; 16 ------------- WES"-- I 20 / WEST UJ !;;: ..J 4 ~ ~ .. ---- -4 -8 ~ '-- 20 UP 10 UP CP 10DOWN 20 DOWN DISTANCEFROMCENTRALPARK(miles) 15 ~ .J 0 ~ 0 ---~-- Iii """', '" ~ 4 ! HIGH ~-_/- LOW 0 I- ----- ---- ~----- [LOW ~-12 ..J L(}N - - -~/ 8 ------------ -==-.L UJ _/-~~----- C 0 ..J .. a:: UJ !;;: 12 ..J ~ -8 /"", ;:! 4 VJ 0 ~ 15-4 ~---- ~ t ~~~ -'1 16 :§ 12 ~ z 8 HIGH g 0'-- //-.,;/ ...- - WEST 8 4 4 0 , 10 DOWN 20 DOWN FROMCENTRAL PARK (miles) Fig. 9. Day and night streamflow lines through New York City during east and west wind cases. / / // HIGH .-/ --- ---~-- -- LOW ---~ 20 UP I I I -........ 10UP CP 10DOWN 20 DOWN DISTANCEFROM CENTRALPARK (miles) Fig. 11. Day streamflow lines for westerly flow, grouped according to upwind rural wind speed. Urban-rural ,603 wind velocity differences removed. In addition, the magnitude of the night-time east wind, low-speed cyclonic turning is less than the cyclonic turning of its overall curve (Fig. 9), as its high-speed case had extremely large cyclonic turning. The daytime east wind cases did not have enough data to subdivide into high- and low-speed cases, but the daytime west wind, high-speed curve of Fig. 11 shows the expected roughness induced cyclonic turning over the city, followed by extreme cyclonic turning as the flow encounters the sea breeze at a point 10 miles (16 km) east of Central Park. The daytime west wind, low-speed curve turns cyclonically over the city, as have all previous daytime curves (Figs. 8 and 9). SUMMARY Analyses of surface wind speed data obtained from the New York University mesoscale anemometer network for the metropolitan New York City area showed that urban-rural surface wind velocity differences between the city and surrounding non-urban areas depend on time of day, wind direction, and upwind rural wind speed. During both daytime and night-time hours there exists a critical upwind rural wind speed below (above) which air is accelerated (decelerated) as it flows over the rough, warm city. The acceleration is due to horizontal pressure gradients directed inward to the center of a city, and to the decreased stability of the urban atmosphere. Both of these effects are associated with a well-developed urban heat island. The deceleration, on the other hand, is due to the large value of the urban surface roughness parameter, as compared to that of the surrounding non-urban regions. The critical speed was found to be 8 mph (3.6m/s) for daytime and 9 mph (4.1m/s) for night-time, in agreement with results obtained by Chandler (1965) for London. The results also agree with those predicted by the numerical heat island model of Bornstein (1975) in that the fastest winds over New York City were found at its downwind edge. The effect of the daytime easterly sea breeze was demonstrated by the low-speed winds downwind of New York City with prevailing westerly flow, and by the high-speed winds upwind of the city with prevailing easterly flow. Wind direction is also affected by the city. When air decelerated as it flowed over the city, it generally turned cyclonically. The wind direction is also affected by the sea breeze, which caused daytime cyclonic turning downwind of the city with westerly flows and upwind of the city with easterly flows. During periods of acceleration over the city, either cyclonic or anticyclonic turning was found, due to a combination of .two effects. The origin of the flow was found to influence the direction of the turning over the city during night-time heat island hours. For easterly flows the origin is a highly stable rural region, while for westerly flows it is most likely one of the less stable, but generally very aerodynamically smooth water bodies around New York City. For westerly flows the stability difference between the origin of the flow and the city was great, and the flow turned anticyclonically over the city, i.e., the cross-isobaric flow angles decreased in the less stable air. While for the easterly flows, the roughness difference between the origin and the urban site was greater, and the flow turned cyclonically over the city. However, for daytime periods' of acceleration the flow only turned cyclonically, as the thermal stability-difference effects is small during such periods. It should be noted that the stability-difference argument presented above is only completely valid for an infinitely-wide city, with flow perpendicular to its boundaries. This model corresponds to the solution obtained from a two-dimensional, vertical plane, urban boundary layer model, such as that of Born. stein (1975). For a city of finite size, the surface crossisobaric flow angle would also be affected by convergence into the warm city. This pattern is consistent with that of three-dimensional urban heat island, as illustrated by the wind tunnel measurements of SethuRaman and Cermak (1974). Depending on the wind direction, and on whether the flow originates above or below the center of the heat island, either cyclonic or anticyclonic turning may occur over the city. The present study is based on changes along a single streamflow line, so nothing can be said about the effect of stream line convergence, even though the results of SethuRaman and Cermak (1974) show that this effect should be present downwind of the urban center and not over its center, due to the effects of advection. Additional observational studies are needed to determine the conditions under which convergence becomes significant. It is interesting to note, however, that the region of convergence in Fig. 1 is found downwind of the warmest section of New York City, i.e., downwind of Manhattan Island. The predicted trajectory and size of a puff of pollution passing over New York City were found by Johnson and Bornstein (1974) to be greatly affected by the varying wind speed, direction, and stability induced by the city. Thus, an urban air pollution transport and diffusion model which attempts to compute concentration distribution, based solely on the standard meteorological observations taken as a single U.S. Weather Service "airport" station located outside of an urban area, will produce erroneous results. . Acknowledgements~ This work was supported by NSF Grant GA-41886. The preliminary draft was typed by Susan Johnson, while the final draft was typed by Linda F. LaDuca. Professor Al Miller reviewed preliminary versions of the final manuscript. Final preparation' of the figures used in this report was carried out at Brookhaven National Laboratory. REFERENCES Angell J. K.. Pack D. H., Machta L.. Dickson C. R. and Hoecker W. H. (1972) Three-dimensional air trajectories 604 ROBERT D. BORNSTEIN and DOUGLAS Scm, determined from tetroon flights in the planetary boundary layer of the Los Angeles Basin. J. Appl. Meteorol. 11, 451-471. Blackadar. (1962) The vertical distribution of wind and turbulent exchange in a neutral atmosphere. J. Geophys. Res. 67, 3095-3102. Bornstein R. D. (1968) Observations of the urban heat island effect in New York City. J. Appl. Meteorol. 7, 575-582. Bornstein R. D. (1975) Two-dimensional, non-steady numerical simulations of night-time flows of a stable planetary boundary layer over a rough warm city. J. Appl. Meteorol. 14, 1459-1477. Bornstein R. D. et al. (1972) Recent observations of urban effects on winds, temperature and moisture in New York City. Preprints of the Amer. Meteor. Soc. Conference on the Urban. Environment, Philadelphia, Pa., Oct. 31-Nov. 2, 1972. Bornstein R. D. et al. (1976) NYU/NYC urban air pollution project of 1964:-1969: The data. Dept. of Meteorology, San Jose State University Report No. 76-02, 141 pp. Chandler T. J. (1965) Climate of London. W. Heffer & Sons, Ltd., Cambridge, England. Davidson B. (1967) A summary of the New York University urban air pollution dynamics research program. Air Poll. Cont. Assoc. J. 17(3); p. 154. Druyan L. M. (1968) A comparison oflow-level trajectories in an urban atmosphere. J. Appl. Meteorol.' 7(4), '583-590. Frederick R. H. (1964) On the representativeness of surface wind observations using data from Nashville, Tennessee. lilt. J. Air Wat. Pollut. 8, 11. Georgii H. W. (1968) The effects of air pollution on urban climates. Presented at W.M.O. Symposium on Urban Climates and Building Climatology, Brussels, Belgium, October 1968. JOHNSON Graham I. R. (1968) An analysis of turbulence statistics at Fort Wayne, Indiana. J. Appl. Meteorol. 7, 90-93. Johnson D. and Bornstein R. D. (1974) Urban-rural wind velocity differences and their effects on computed pollution concentrations in New York City. Preprints of Amer. Meteor. Soc. Symposium on Atmospheric Diffusion and Air Pollution, Santa Barbara, Calif., 267-272. Johnson D. S. (1975) Urban-rural wind velocity differences in New York City and their effects on the transport and dispersion of pollutants. M.S. Thesis, Dept. of Meteorology, San Jose State University, 58 pp. Landsberg H. E. (1956) The climate of towns. In: Man's Role in Changing the Face of the Earth. W. L. Thomas, ed., Univ. of Chicago Press, Chicago, Illinois, 584-606. Landsberg H. E. (1971) Personal communication. Oke T. R. (1973) Review of Urban Climatology 1968-1973. W.M.O. Report No. 383, TN No. 134, 132 pp. Okita T. (1960) Estimation of direction of air flow from observation of rime ice. J. met. Soc. Japan 38, 207. Okita T. (1965). Some chemical and meteorological measurements of air pollution in Asahikawa. Int. J. Air Wat. Pollut. 9. 323. Pooler F. (1963) Airflow over a city in terrain of moderate relief. J. Appl. Meteor. 2, 446-456. Scudder B. E. (1965) Diagnosing the mesoscale wind field over an urban area by means of synoptic data. M.S. Thesis, New York Univ., New York. SethuRaman S. and Cermak J..E. (1974) Physical modelling of flows and diffusion over an urban heat island. Adv. in Geophys. 18b,223-240. Slade D. H. (1968) Meteorology and Atomic Energy. U.S. Energy Comm., Oak Ridge, Tenn. Yamamoto G., Yasuda W. and Shimanuhi A. (1968) Effect of thermal stratification on the Ekman layer. J. met. Soc. Japan, Ser. 2, 46, 442-445.. .. ",' ;;f"t; ,..i,U' '", "', ,,', . " ;
© Copyright 2026 Paperzz