The CAT model: Predctng r temperture n cty streets on the bss of mesured reference dt Evytr Erell 1 nd Terry Wllmson 2 1 The J. Blusten Insttute For Desert Reserch, Ben-Guron Unversty of the Negev, Sde Boqer Cmpus, Isrel 2 The Unversty of Adelde, Adelde, Austrl ABSTRACT: Meteorologcl mesurements requred for the development of desgn scheme tht responds to the locl envronment re generlly recorded by the wether servce n sttons tht re ssumed to be representtve of the surroundng regon. However, no ccount s tken of the chnges n condtons cused by n development, even though dfferences between eorologcl condtons wthn ctes compred wth djcent rurl res (the het slnd effect hve been noted s erly s Romn tmes, nd my be substntl. A new computer model (CAT provdes dt representng relstc ste-specfc r temperture n cty street bsed only on dt from stndrd wether sttons such s those operted t locl rports nd descrptors of the two stes. The CAT model hs been tested on feld dt mesured n montorng progrm crred out n Adelde n 2000-2001. After clbrtng the model, predcted r temperture correlted well wth mesured dt n ll wether condtons over extended perods. Conference theme: Archtecture nd the envronment Keywords: n mcroclmte; computer modellng; wether sttons; Adelde INTRODUCTION Dfferences between the cty nd the rurl surroundngs occur on two dstnct scles (Oke, 1976: The n cnopy lyer conssts of r contned between the n roughness elements, typclly buldngs. Its clmte s domnted by the nture of the mmedte surroundngs, especlly ste mterls nd geory. The n boundry lyer s tht porton of the plnetry boundry lyer whose chrcterstcs re ffected by the presence of n n re t ts lowest boundry, generlly consdered to be pproxmtely t roof level. Severl models hve been developed over the yers n n ttempt to descrbe mcroclmtc condtons n the n cnopy lyer. A comprehensve revew of n clmte smulton models (Bornsten, 1984 noted tht " potentlly mportnt use of such models could be n plnnng of n development so s to crete n clmtes tht re more helthy or n dwellers." Indeed, Clrke (1993 dentfed mcroclmte nd ste specfc dt s one of the problem res defnng boundry condtons for buldng performnce smultons. Despte ths mpertve few comprehensve n clmte smulton models exst whch focus drectly on the ssues of mportnce for buldng therml performnce smulton. Shrln & Hoffmn (1984 nd Swd & Hoffmn (1990 descrbed the development of smulton model bsed on CTTC (Cluster Therml Tme Constnt prer to predct n tempertures s functon of some bsc n geory. The objectve of the reserch reported here ws to nvestgte whether reltvely smple uncoupled model of the temperture n the n cnopy,.e. model n whch there s no feedbck from the mcro-scle ste to the meso-scle n boundry lyer, cn nonetheless predct the temperture t gven ste wth suffcent ccurcy to be useful tool for reserch on the effects of n form on r temperture n the n cnopy lyer. 1. SIMULATION The CAT (Cnyon Ar Temperture model predcts r temperture n n n cnyon on the bss of mesured eorologcl prers from reference stton locted t nerby ste exposed to the sme meso-scle clmtc condtons. Model nputs re lmted to eorologcl prers montored routnely t stndrd wether sttons. Ths restrcton hs mportnt mplctons tht wll be dscussed n detl n the relevnt sectons below. In the trdtonl pproch to numercl modellng of the mcroclmtc chrcterstcs of the n cnopy lyer, wether dt re used to provde geogrphcllydjusted ntl nd boundry condtons to plnetry boundry lyer (PBL model, such s URMICLEM (Th, 1978. The meso-scle eorologcl condtons predcted by such model, n conjuncton wth detled ste-specfc descrptors, re used to drve n n clmte model cpble of provdng mcro-scle tempertures t the ste under consderton. Ths pproch s bsed on complete nlyss of the energy fluxes, but requres detled clmtc nformton tht s generlly not vlble n typcl wether sttons. The CAT model dopts smpler pproch: the rw wether dt tself serves s the nput to the n clmte model, scrfcng comprehensve nlyss n the nterests of prctclty nd ese of pplcton. Ths pproch ws the bss of the modfed CTTC computer code (Elnhs nd Wllmson, 1997, n nlytcl model ncorportng cluster therml tme constnt for predctng r temperture vrtons n the n cnopy lyer. Itself n extended verson of prevous model orgnlly proposed by Swd & Hoffmn (1990, ths model predcted r temperture t n n pont bsed on mesured temperture t rurl reference ste, modfed by the contrbuton of chrcterstclly n condtons. The conceptul bss for ths converson ws stted by Elnhs & Wllmson (1997 s follows: 210
T = T + T " T (1 b b sol sol lw T = T + T " T (2 lw where T (t refers to r temperture t gven locton, T b the meso-scle bse temperture, nd T sol nd T lw the net short wve nd long wve rdtve contrbuton to the locl temperture. The subscrpts nd refer to n nd eorologcl (reference loctons, respectvely. Combnng the bove expressons yelds: T = T + ( Tsol " Tsol " ( Tlw " Tlw (3 The bove procedure ttrbuted the vrton n r temperture to dfferences n short wve nd long wve energy blnces between the n ste nd the eorologcl (reference ste. However, the sme hod my be employed to express dfferences resultng from other modfers of locl mcro-clmtc condtons, such s, for exmple, nthropogenc het, the hydrologcl blnce or the effects of vegetton, whch my ffect ltent het. The CAT model dopts smlr conceptul pproch, but employs dfferent hodology to compute T b, the meso-scle bse temperture, nd then to predct cnyon r temperture. The process my be summrsed s follows: An energy blnce s clculted from eorologcl tme seres for the reference ste nd for ech surfce of the n cnyon, tkng nto ccount modfctons to the energy exchnge processes resultng from the specfc chrcterstcs of both stes. The effect of sensble het flux t the surfce on r temperture t the reference ste s then clculted on the bss of the surfce het exchnge coeffcent h c nd new emprcl coeffcent of tulent het trnsfer between the cnopy lyer nd the roughness sublyer bove. The tme-dependent contrbuton of the locl sensble het flux to the r temperture t the reference ste s then deducted from the mesured vlue to gve T b, the meso-scle bse temperture. The sme hodology s then used to clculte the ensemble contrbuton to the locl cnyon r temperture of sensble het flux from ll cnyon surfces, smlrly modfed by (dfferent coeffcent of tulent het trnsfer between the r n the cnyon nd the roughness sublyer bove. Ths contrbuton s then dded to the bse temperture to gve the predcted r temperture n the cnyon. Inputs requred by the model comprse two seprte ctegores: Tme seres descrbng eorologcl condtons t the reference ste, ncludng the dry bulb temperture of r, reltve humdty, wnd speed, globl nd dffuse solr rdton on horzontl plne nd cloud cover. Descrptors of both the reference ste nd the n ste beng smulted, ncludng: geogrphc locton (lttude nd longtude; buldng geory, ncludng street cnyon spect rto, buldng densty nd rto of mssve externl wlls to ste re; mterl propertes, ncludng surfce lbedo nd the coeffcents of the objectve hysteress model (OHM of therml storge (Grmmond et l., 1991; nd hydrologcl fctors ffectng the prcellng of tulent het flux nto ltent het nd sensble, ncludng the coeffcents of the LUMPS model by Grmmond nd Oke (2002. The georc representton of the n envronment s two-nd--hlf-dmensonl: cnyon dmensons re gven only n cross-secton (length n the drecton of the mn longtudnl xs s consdered sem-nfnte. However, tretment of n-cnyon wnd s fully threedmensonl, s s the ncorporton of solr poston n the computton of short-wve rdnt exchnge. The CAT model trets ech of the followng fctors components of the het blnce s follows: 1.1. Rdtve exchnges Net rdton (Q* my be mesured drectly, but s not usully montored n stndrd eorologcl stton. Furthermore, net rdnt exchnge depends on the orentton of the relevnt surfce, ts ngle of tlt wth respect to the horzon nd ts degree of exposure to drect solr rdton, to the sky nd to other terrestrl surfces. The CAT model therefore utlses more commonly vlble dt - globl solr rdton nd dffuse solr rdton, both on n unobstructed horzontl surfce, nd cloud cover, ll recorded t the reference stton. Net rdnt exchnge t the reference stton nd on ech surfce n the n cnyon s then clculted on the bss of solr geory nd the respectve vew fctors of the prtcptng surfces. Incomng drect norml (bem solr rdton s clculted from mesured globl rdton nd dffuse rdton, tkng nto ccount solr elevton. Bem rdton s then used to clculte the drect component on buldng surfces, wth respect to orentton nd mutul shdng by other cnyon surfces. The proporton of ech surfce exposed to drect rdton s clculted usng model tht llows nput of seprte (dfferent vlues for ech cnyon wll. In ddton, the heght nd setbck of djcent buldngs my be nput to ccount for obstructons creted by buldngs tht re much tller thn those formng the prmry cnyon perer, or for tower blocks constructed on podum tht comprses the ctul cnyon wlls. Cnyon surfces not exposed to drect solr rdton, ncludng portons of exposed surfces shded by djcent wlls, re consdered s beng llumnted by dffuse rdton nd by reflected rdton only, n proporton to the respectve vew fctors for ech of the source res. The tme-dependent totl solr rdnt flux on the th cnyon surfce recevng drect solr rdton s well s reflected rdton from j other cnyon surfces s thus gven by the followng expresson: I dr+ df j(1 " m j (# " j j I = I (1 " PSA( t + I # + dr df (4 where I (t s the men hourly totl solr rdton ncdent on the surfce (W m -2, PSA s the prtl shded frcton of the surfce, I dr(t the hourly men unobstructed drect component of the solr rdton, I df(t the hourly men unobstructed dffuse component of the solr rdton, I dr+df(t the combned drect nd dffuse components of the solr rdton, the vew fctor between two surfces, m j the solr bsorptvty of surfce j (W m -2 o K nd s the men sky vew fctor for surfce. Reflected rdton, s noted bove, tkes nto ccount prmry reflectons only. Incomng long wve rdton (L" s clculted from the dry bulb temperture of the r nd from tmospherc humdty for cler sky condtons v the emprcl correlton proposed by Brutsert (1982 for tmospherc emssvty, 1 = 1.24( e T 7 (5 where e s the prtl wter vpour pressure of r (mb nd T r temperture (K. The prtl wter vpour pressure of r my be clculted from reltve humdty nd r temperture followng the procedure set out n ASHRAE (1989. Sky rdton s clculted usng the Stefn-Boltzmnn reltonshp L #= " T 4 (6 s 211
where # s the Stefn Boltzmnn constnt (5.67*10-8 W m -2 K -4, $ s sky emssvty nd T r temperture (K. A correcton fctor ncorportng the effect of clouds on L" s ppled followng Mrtn (1989: 2 3 L ( n = (1 + 0.0224n " 0.0035n + 0.00028n L cler (7 Compred wth other emprcl correctons for the effect of clouds, such s the one proposed by Oke (1987, ths expresson scrfces ccurcy for smplcty of pplcton: t requres no nformton on cloud type, whch my not lwys be vlble n stndrd eorologcl records. The pproch dopted n the CAT model for clcultng outgong long wve rdton s compromse between detled modellng of surfce temperture nd the use of mbent r temperture s surrogte for t. The former s ccurte, but requres detled descrpton of the (tme-chngng therml propertes of the sold, wheres the ltter s smple but less ccurte. Surfce temperture s pproxmted by the sol-r temperture, whch s defned s the equvlent outdoor temperture whch wll cuse the sme rte of het flow t the surfce nd the sme temperture dstrbuton throughout the mterl s results from the outdoor r temperture nd the net rdton exchnge between the surfce nd ts envronment (the commonly used notton hs been modfed to conform to conventons used n clmtology: K #" + sl* Tsol = T + hc (8 where T sol s gven n K, K" s ncomng short wve rdton (W m -2, % s short wve bsorptvty of the surfce, $ s s the long wve emssvty of the surfce, L* s net long wve rdton t the surfce (W m -2 nd h c the surfce convectve het exchnge coeffcent. The outgong long wve rdton (L& s tself functon of the sol-r temperture, so ts vlue s obtned by tertve clculton of the expresson bove untl T sol s obtned to n ccurcy of less thn 0.5K. The contrbutons of the ground surfce nd vertcl surfces of the cnyon re clculted seprtely tkng nto ccount ther respectve vew fctors to the sky. 1.2. Het storge (Qs Het storge s modelled usng the Cmuffo & Bernrd (1982 formulton: Q * " Q s = 1Q * + 2 + 3 t (9 where t s tme; prer 1 ndctes the overll strength of the dependence of the storge het flux on net rdton; 2 determnes whether the curves Q S nd Q* re exctly n phse ( 2=0, or Q S precedes Q* ( 2>0; nd 3 s the sze of Q S when Q* becomes negtve (postve vlues ndctes Q S becomes negtve before Q*. The vlues of coeffcents 1, 2 nd 3 correspond to typcl vlues proposed by Grmmond & Oke (2002. 1.3. Tulent het flux (Q H nd Q E The vlue of h c, the surfce het exchnge coeffcent, s ffected by the reltve tempertures of the surfce nd the djcent r lyer, by the orentton of the surfce wth respect to the horzontl, nd most mportntly, by the speed of r movement cross the surfce. Snce h c cnnot be determned by nlytcl hods, CAT employs the followng emprcl correltons for horzontl nd vertcl surfces, obtned expermentlly on buldng surfces t n n ste (Hgshm nd Tnmoto, 2003: 2 2 2 = 3.96 u + v + w 6.42, nd h c + ( hor 2 2 2 h c = 10.2 u + v + w + 4.47 (10 ( vert where u, v nd w re wnd vectors (m s -1 nd h c s gven n W m -2 K -1. In the reference ste, wnd speed ner the surfce (20cm heght s derved from dt mesured t 10m heght ssumng stndrd logrthmc profle. Vlues for the zero-plne dsplcement (d nd the roughness length (z 0 re estmted n ccordnce wth the erodynmc chrcterstcs of the reference ste. In the cse of typcl eorologcl stton, d s on the order of few mllres nd z 0 bout 1cm. u * cn then be clculted usng vlue of k=0.4 for the von Krmn constnt: vz u* = k " ln( z d z0 (11 Usng the resultng vlue for u *, the expresson s then solved for v z to yeld wnd speed ner the surfce. The smple logrthmc profle s not pproprte n the n cnyon. Wnd speed ner cnyon surfces s clculted on the ssumpton tht vortex s formed, usng the Hotchkss nd Hrlow (1973 model, quoted n Ymrtno nd Wegnd (1986, to clculte wnd speed ner ech of the surfces from wnd speed bove the cnyon top. The trnsverse (horzontl component of the flow (u nd the vertcl component (w t ny pont n two-dmensonl rectngulr notch of depth H nd wdth B (n effect n n cnyon re gven by the followng expressons: u = u 0 (1" # "1 [ $(1+ ky " #(1" ky /$]sn(kx (12 # 1 w = # u0ky(1 # " [ # " / ]cos( kx (13 where k="/b: '=exp(-2kh; (=exp(ky; y=z-h; nd u 0 s the trnsverse wnd speed bove the cnyon (t the pont x=b/2, z=h. The trnsverse flow, whch s ndependent of the longcnyon longtudnl flow, hs logrthmc profle: & z + z # & + # 0 zr z0 (14 v( z = vr log$ log$ % z0 " % z0 " where v r s the long-cnyon component of the wnd speed t reference heght r bove the cnyon nd z 0 s the roughness length. In the bsence of ccurte locl mesurements, the roughness length s estmted s beng one-tenth the verge heght of the cnyon wlls (Grmmond et l., 1998. Sensble het flux s prersed for ech cnyon surfce from the net rdnt blnce (Q* nd the storge flux (Qs (De Brun nd Holtslg, (1982, n the revsed form gven by Grmmond & Oke (2002 for n condtons: (1 $ # + (" / s QH = ( Q * $% Q s $ 1+ (" / s (15 where s s the slope of the sturton vpour pressureversus-temperture curve; ( s the psychrorc constnt; nd % nd ' re emprcl prers. % depends on the sol mosture sttus, nd ccounts for the strong correlton of Q H nd Q E wth Q*-Q S, wheres ' ccounts for the uncorrelted porton. 1.4. Anthropogenc het (QF In the CAT model, nthropogenc het flux Q F s dded to the sensble het flux Q H before the temperture modfcton procedure s crred out. The energy s treted s f t s relesed from pont source n the centre of the cnyon. Monthly men vlues of totl nthropogenc het flux densty re multpled by correcton fctor relted to the tme of dy to djust for the typcl durnl pttern of energy consumpton (Slor nd Lu, 2004. 212
1.5. Temperture predcton from sensble het flux The locl effect of the sensble het flux t the surfce on r temperture n the cnopy lyer depends on the exchnge of energy wth the mxed lyer bove roof level, nd s ffected by two clsses of fctors (Kstner- Klen nd Plte, 1999: The geory of the n cnyon, especlly ts spect rto; nd the morphology of the buldngs t s defned by, especlly the desgn of the roof. For gven cnyon, the effect of these fctors my be consdered fxed. The chrcterstcs of the rflow bove the roofs, n prtculr wnd speed nd drecton wth respect to the cnyon xs, nd the mgntude of the tulence. A further fctor s the stte of tmospherc stblty ner the surfce. All of these fctors re constntly chngng. An ccurte evluton of ths exchnge my be crred out numerclly by CFD. However, CFD requres very detled nput, nd s stll too demndng on computer resources to smulte extended perods. Anlytcl models, though less strngent n ther requrements, nevertheless requre nput of mesurements suffcent to chrcterze tmospherc stblty by prers such s the Rchrdson number, whch cnnot be derved from typcl wether sttons. CAT expresses the exchnge of sensble het between the cnopy lyer nd the mxed lyer bove by mens of bulk erodynmc formulton for fluxes, of the generl form: Q H = T r t (16 where Q H s the sensble het flux, obtned from the sum of the contrbutons of the ndvdul surfces comprsng the envelope of the ste, weghted by ther respectve res s proporton of the overll pln re of the ste. T s the temperture dfference between the cnopy lyer nd the bse temperture n the reference lyer, whch wll vry over tme, nd r t s resstnce to het trnsfer cross the nterfce between the two lyers. An expresson for r t ws developed whch ncorportes the effects of ste geory, nd whch ccounts for the effects of tulence, nd n prtculr, tmospherc stblty: r t = ( 1" m h c (17 where s the rto of the cnyon surfce re to the pln re of the ste, h c s the coeffcent of het exchnge t the surfce (whch s functon of wnd speed, nd m s coeffcent of tulent exchnge tht vres wth stblty nd wnd speed. The vlue of m ws obtned emprclly for four seprte clsses of flow regmes, whch were defned on the bss of the temperture dfference between the sol-r temperture of the surfce nd the dry bulb temperture of the r t the reference ste, s follows: Neglgble mxng when T<-2K; lght mxng -2K<T<0K; moderte mxng 0K<T<5K; vgorous mxng 5K<T. Tble 1 below gves vlues of m for tulent exchnge between n n cnyon wth n spect rto of pproxmtely 1.3 nd the roughness sublyer bove compred to those for n open reference ste, for ech of the mxng clsses. Tble 1: Vlues of the coef fcent of tulent exchnge m for df ferent mxng regmes mxng regme open ste n cnyon (H/W=1.3 neglgble mxng 0.15 0.1 lght mxng 0.5 0.3 moderte mxng 0.7 0.5 vgorous mxng 0.8 0.65 Cnyon r temperture s clculted s the sum of the r temperture t the reference stton nd the net chnge n temperture resultng from the locl chrcterstcs of the reference ste nd the n cnyon: T = T + ( TQ H " TQ H (18 2. FIELD DATA Feld mesurements were crred out t two n cnyons n Adelde Chesser Street, whch runs northsouth, nd French Street, whch hs n est-west xs. Dt were smpled t 15-second ntervls, nd 15- mnute verges were computed by electronc dt loggers nd stored for subsequent retrevl. The dt from the n stes ws compred wth dt recorded t reference ste t the Adelde Cty Councl Nursery, north of the Torrens Rver, nd wth records provded by the Bureu of Meteorology from the Kent Town stton. The reference ste s locted n green belt surroundng the centrl busness dstrct of Adelde, pproxmtely 2100m northwest of the cty centre. The n stes re bout 500m from the centre of the busness dstrct, nd bout 1500m southest of the reference ste (Fgure 1. The re n the photo s the core of the Adelde ropoltn re, whch extends bout 20kms est from the Gulf of St. Vncent to the Adelde Hlls, nd bout 25kms from north to south. Fgure 1: Aerl photo of centrl Adelde Dt representtve of the whole of the study re ncluded globl solr rdton; dffuse solr rdton, net ll-wve rdton nd effectve sky temperture. The reference stton montored ste-specfc dry bulb temperture, reltve humdty, wnd speed nd drecton, sol temperture nd sol mosture. At the n stes mesurements were mde of r temperture t severl ponts n ech street cross-secton, reltve humdty, wnd speed nd drecton. 3. MODEL VALIDATION The vldton of the model s bsed on comprng predcted tempertures wth observed dt ssembled n the expermentl phse of the project. An obvous dffculty n pplyng stndrd goodness of ft tests to the predcted nd observed temperture records n the current experment s tht there s hgh degree of utocorrelton between the vrbles. Thus use of the 213
Fgure 2: Comprson of predcted nd observed r temperture n Adelde, My 1999 correlton coeffcent (r or ts squre (r 2 nsuffcent. Therefore, n ddton to comprng the observed nd predcted vrtes respectve mens (O, P nd stndrd devtons (s o, s p; the ntercept nd slope of lest squres lner regresson between the vrtes; the errors descrbed by the root men squred error (RMSE, nd ts systemtc nd unsystemtc components, two ddtonl ndctors were used: The Wllmott ndex of greement (Wllmott, 1981: d N = 1 = 1# N = 1 ( P # O [ P" + O" ] 2 2 (19 where P re predcted vlues, O observed ones, nd P" = P O nd O" = O O. In the expresson bove, the denomntor of the mn term represents the potentl error of the smple whle the numertor represents the porton of ths error left unexplned by the smulted vrte. The ndex of greement vres between 0 nd 1, 1 ndctng perfect greement: The Wllmson confrmton fctor (Wllmson, 1995: If the vlue v of n nput vrble cn provde frly close pproxmton to the mesured prer m, model my only be consdered useful f the estmted vlue of the prer n queston e s closer to the observed vlue thn ths trvl pproxmton. The expresson C s=u(m,v-u(e,m where U(m,v s Thel s nequlty coeffcent between the mesured vlue nd the trvl vrble estmte, nd U(e,m s Thel s nequlty coeffcent between the estmted vlue nd the mesured vlue of the prer n queston, s n ndctor of the mgntude of the error n predcton: The mxmum vlue of the C s s U(m,v, when U(e,m s equl to zero, ndctng perfect correlton between the mesured nd estmted vlues. Dvdng the confrmton fctor C s by U(m,v normlzes ll possble vlues of ths fctor, gvng degree of confrmton D hvng vlue between zero nd unty. U(m,v-U(e,m D = (20 U(m,v The model ws clbrted on the bss of dt for the month of June 1999. Emprcl vlues were obtned for the coeffcents of the het storge expresson (equton 9; for the sol mosture (equton 15 nd for the coeffcent of tulent exchnge t the cnyon top (Tble 1, so s to mnmze the error n predcton of cnyon r temperture. The smulton ws then run for three other month-long perods usng constnt vlues for these coeffcents. Vsul nspecton of the grph for the frst ten dys of My 1999 (Fgure 2 s ressurng, snce t demonstrtes tht the mn trends, such s the nocturnl n het slnd, re descrbed frly ccurtely. Vrous sttstcl mesures evlutng the qulty of the predctons re shown n Tble 2. The predctons for the months of My nd Mrch show good greement wth observed tempertures. The results for November re dsppontng. Although the bsolute error s not very lrge, the vlue of the Wllmson confrmton fctor s low becuse durng ths perod, wether condtons n the n ste nd n the reference ste were nerly dentcl, so rndom errors n ny of the nputs could led to n error n predcton tht ws s lrge s the dfference between the stes. Tble 2: Vldton sttstcs for dfferent tme perods (June dt were used for clbrton. June My November Mrch Men error [ o C] -0.03-0.05-0.12-0.22 Stndrd devton [ o C] 1.08 0.88 1.38 1.19 Systemtc MSE [ o C] 0.10 0.02 0.02 0.06 Unsystemtc MSE [ o C] 1.08 0.77 1.90 1.40 Wllmott ndex of greement 0.95 0.97 0.98 0.99 Wllmson confrmton fctor 0.51 0.58-0.14 0.33 4. DISCUSSION Accurte determnton of the surfce het exchnge coeffcent h c hs drect berng on the ccurcy of the CAT model. The dffculty of ssessng ths prer n envronments wth complex erodynmc morphology, such s the n cnyon, mens tht n prctce, ccurte predctons of the effect of sensble het flux on r temperture requre frly ccurte modellng of wnd speed ner the surfce. The model ncorported n CAT s currently lmted to n confgurtons where smple rotor vortex s formed, nd s npplcble where the cnyon s too shllow (resultng n solted roughness flow, ccordng to Oke s clssfcton or where t s very deep (resultng n very wek couplng of street-level wnd wth r bove the cnyon. Assessment of the sensble het flux s senstve to the vlue of the coeffcent n equton15 bove. Ths coeffcent, whch descrbes the vlblty of mosture n dfferent envronments, my rnge n vlue between zero n hyper-rd loctons to bout 1.4 over ocens (Grmmond nd Oke, 2002. In most n stes t s lkely to be n the rnge 0.4-1, but more ccurte determnton my requre clbrton of the CAT model to vrety of loctons on the bss of expermentl dt from on-ste montorng. Untl such dt re vlble, clbrton my be crred out to ft predctons to dt from the reference ste n eorologcl condtons tht re known to result n neglgble n effects, such s wndy, overcst wether. Ths procedure my only be crred out f the contrbuton of nthropogenc het to temperture dfferences between the loctons s smll. The CAT model, n common wth mny models of the n mcroclmte, ssumes tht net dvectve het trnsfer wthn the n re s neglgble. It s therefore pplcble where buldng densty, typology nd ground cover n the source re for energy fluxes re unform. However, the presence of n prks or lrge vrtons n buldng heght nd street sectons n close proxmty to the cnyon beng modelled my result n errors whose 214
mgntude s ffected by ther dstnce from the cnyon, the drecton of the wnd nd the ntensty of the fluxes. The CAT model s lso ncpble of nlysng dfferences between the reference ste nd the n cnyon resultng from meso-scle propertes of the stes such s topogrphy, elevton, dstnce from lrge body of wter (f pplcble etc. CONCLUSIONS The CAT model demonstrtes tht reltvely smple nlytcl model of the n cnyon cn provde estmtes of r temperture tht re substntl mprovements over the trvl cse of usng unprocessed dt from reference wether stton. However, the model n ts current form s stll lmted, especlly wth respect to the types of n morphology t my be ppled to. It hs lso not been tested gnst feld dt from loctons other thn Adelde, whch my hve substntlly dfferent mcro-clmtes. It therefore remns to be seen whether the emprcl clbrton fctors deduced from the Adelde dt my be ppled elsewhere. If these restrctons cn be overcome, the CAT computer model s seen s hvng severl prctcl pplctons: Evluton of the mcro-clmtc effects of proposed plnnng regultons concernng lnd use nd buldng densty n new or exstng n development; Evluton of chnges to the mcro-clmtc condtons t specfc n loctons s result of proposed chnges such s new constructon; Use s reserch tool for exmnng vrous n confgurtons n terms of ther effect on pedestrns nd on consumpton of energy n buldngs. Provson of dt representng relstc ste-specfc eorologcl condtons tht my then be nput to buldng therml smulton softwre to produce more ccurte modellng of energy use. ACKNOWLEDGEMENTS The uthors would lke to thnk Mr. McLchln nd Mr. Tor for llowng eorologcl equpment to be nstlled n ther buldngs t Chesser nd French Streets for the durton of the experment; the Adelde Cty Nursery, whch provded ste for the reference stton; Dr. Cly nd Mr. Wld of the Dept. of Physcs t Adelde Unversty for provdng the IR detectors; nd the Cty Councl of Adelde for ther support of the project. REFERENCES Bornsten, R. D. (1984."Urbn clmte models: nture, lmttons nd pplctons", Urbn Clmtology nd ts Applctons wth Specl regrd to Tropcl res, Mexco D.F., World Meteorologcl Orgnzton, Genev Brutsert, W. (1982. Evporton nto the Atmosphere. Dordrecht / Boston / London, Kluwer Acdemc Publshers. Cmuffo, D. nd Bernrd, A. (1982. "An observtonl study of het fluxes nd ther reltonshps wth net rdton", Boundry Lyer Meteorology 23: 359-368. Clrke, J. A. (1993. "Assessng buldng performnce by smulton", Buldng nd Envronment 28(4: 419-428. De Brun, H. nd Holtslg, A. (1982. "A smple prerzton of the surfce fluxes of sensble nd ltent het durng dytme compred wth the Penmn- Monteth concept", Journl of Appled Meteorology 21: 1610-1621. Elnhs, M. M. nd Wllmson, T. J. (1997. "An mprovement of the CTTC model for predctng n r tempertures", Energy & Buldngs 25: 41-49. Grmmond, C. S., Cleugh, H. A. nd Oke, T. R. (1991. "An objectve het storge model nd ts comprson wth other schemes", Atmospherc Envronment 25B: 311-326. Grmmond, C. S. B., Kng, T. S., Roth, M. nd Oke, T. R. (1998. "Aerodynmc roughness of n res derved from wnd observtons", Boundry Lyer Meteorology 89: 1-24. Grmmond, C. S. B. nd Oke, T. R. (2002. "Tulent het fluxes n n res: Observtons nd Loclscle Urbn Meteorologcl Prerzton Scheme (LUMPS", Journl of Appled Meteorology 41: 792-810. Hgshm, A. nd Tnmoto, J. (2003. "Feld mesurements for estmtng the convectve het trnsfer coeffcent t buldng surfces", Buldng nd Envronment 38: 873-881. Kstner-Klen, P. nd Plte, E. J. (1999. "Wnd-tunnel study of concentrton felds n street cnyons", Atmospherc Envronment 33: 3973-3979. Mrtn, M. (1989. "Rdtve coolng". Cook, J., Ed. Pssve Coolng. Cmbrdge, Mss., MIT Press 593 p. Oke, T. R. (1976. "The dstncton between cnopy nd boundry-lyer n het slnds", Atmosphere 14: 268-277. Oke, T. R. (1987. Boundry Lyer Clmtes. London & New York, Methuen. Slor, D. nd Lu, L. (2004. "A top-down hodology for developng durnl nd sesonl nthropogenc hetng profles for n res", Atmospherc Envronment 38: 2737-2748. Shrln, N. nd Hoffmn, M. E. (1984. "The n complex s fctor n the r-temperture pttern n Medterrnen costl regon", Energy & Buldngs 7: 149-158. Swd, H. nd Hoffmn, M. E. (1990. "Predcton of n r temperture vrtons usng the nlytcl CTTC model", Energy & Buldngs 14: 313-324. Th, H. G. (1978. An Urbn Mcro-Clmte Model for Ste-Specfc Buldng Energy Smulton: Boundry Lyers, Urbn Cnyon nd Buldng Condtons. Dept.of Archtecture. Berkeley, Unversty of Clforn: 135. Wllmson, T. J. (1995."A confrmton technque for therml performnce smulton models", Buldng Smulton '95, Mdson, Wsconsn, U.S.A. Wllmott, C. J. (1981. "On the vldton of models", Physcl Geogrphy 2(2: 184-194. Ymrtno, R. J. nd Wegnd, G. (1986. "Development nd evluton of smple models for the flow, tulence nd pollutnt concentrton felds wthn n n street cnyon", Atmospherc Envronment 20: 2137-2156. 215