ebook img

Using GIS and landscape metrics in the hedonic price modeling of the amenity value of urban ... PDF

13 Pages·2007·0.65 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Using GIS and landscape metrics in the hedonic price modeling of the amenity value of urban ...

LandscapeandUrbanPlanning79(2007)240–252 Using GIS and landscape metrics in the hedonic price modeling of the amenity value of urban green space: A case study in Jinan City, China Fanhua Konga,∗, Haiwei Yinb, Nobukazu Nakagoshia aGraduateSchoolforInternationalDevelopmentandCooperation,HiroshimaUniversity,Kagamiyama1-5-1,Higashi-Hiroshima739-8529,Japan bDepartmentofUrbanandResourcesSciences,NanjingUniversity,Nanjing210093,China Received7November2005;receivedinrevisedform11January2006;accepted14February2006 Availableonline2May2006 Abstract Urbangreenspaceshaveimportantamenityvaluesthatincludeprovisionofleisureopportunitiesandaestheticenjoyment.However,mostof thesevalueslackamarketprice.Consequently,theyareusuallyignoredorunderestimatedbyurbanplanningpolicy-makers,withtheresultthat remnanturbangreenspacesarebeinggraduallyencroacheduponbyurbansprawl.Asaresult,quantitativeinformationregardingtheimplicit, non-marketpricebenefitsfromurbangreenspaceisurgentlyrequired.Propertiesboughtandsoldonthemarketarecompoundcommoditiesthat embodyamenityvaluesandpeoplearewillingtopaytoliveintheproximityoflocalamenityenvironment.Thushedonicmodels,whichuse suchpropertiesasproxies,canoftenbeemployedtoquantifyenvironmentalamenities.InChina,residentialhousingreform(inplacesince1998) hasterminatedthetraditionalresidentialwelfaresystem,andmadeitpossibletoquantifythemonetaryvalueofgreenspaceamenitiesbasedon hedonicpricingmodels.ThisstudywasconductedinJinanCity,andwillhelpaddressthepreviousabsenceoftheapplicationofhedonicprice modelstothevaluationofurbangreenspaceamenitiesinmainlandChina.GISandlandscapemetricswereusedindetermininghedonicprice modelvariables.Asexpected,theresultsprovedthatthehedonicpricingmodelperformedwellusingthisapproach,andaccordinglyitwasfurther improved.Resultsalsoconfirmedthepositiveamenityimpactofproximateurbangreenspacesonhouseprices,andhighlightedthepreferences ofhomeownersinJinanCity.Greenspaceamenityvariablesthatwerestatisticallysignificantatthe5%levelincludedthesize–distanceindexof sceneryforest,accessibilitytoparkandplazagreenspacetypes,andthepercentageofurbangreenspace.Inaddition,land-usepatchrichness,the locationsectorandtheeducationenvironmentalsoprovedtobehighlysignificantvariables.Theresultsofthestudyshouldprovideinsightsto policy-makersinvolvedinurbanplanning. ©2006ElsevierB.V.Allrightsreserved. Keywords: Hedonicpriceanalysis;Greenspaceamenity;Housing;Geographicalinformationsystem;Landscapemetrics 1. Introduction andOsgood,2003).Intheabsenceofanexplicitmarketpricefor aunitofenvironmentalamenity,thebenefitsareusuallyignored The development of environmental awareness has resulted or underestimated by urban planning policy-makers, with the inastrongdemandbyurbanresidentsforgreenspaceforvar- consequence that remnant urban green spaces have shrunk in ious purposes, including aesthetic enjoyment, recreation, and sizeandhavebeengraduallyencroacheduponbyurbandevel- access to clean air or a relatively quiet environment (Miller, opmentandsprawl(Lockeretz,1989;Englishetal.,1990;Leiva 1997; Tyrva¨inen and Miettinen, 2000). However, amenity val- andPage,2000).Suchconflictingtrendsraisetheneedforgreen uesattachedtourbangreenspacesarenon-marketprice,envi- spaceprotectionandallocation,whichinturnrequiresestimates ronmental benefits (Robinette, 1972; Grey and Deneke, 1978; oftherecreationalvalueofgreenspaces.Consequently,proxy Miller, 1997; Tyrva¨inen and Va¨a¨na¨nen, 1998) that cannot be measuresforhowtheamenitiesareperceivedmustbedeveloped. directlytradedonanopenmarket(Moreetal.,1988;Sengupta Thedeterminationofthevaluetosocietyofsuchnon-market pricedrecreationresourcesisnotanewconcepttoeconomists (Price, 2000). Several methods such as travel cost models, the ∗ Correspondingauthor.Tel.:+819075088072;fax:+81824246931. contingentvaluationmethod,andhedonicpricingmodelshave E-mailaddress:[email protected](F.Kong). beendeveloped,andwereimprovedinrecentdecades. 0169-2046/$–seefrontmatter©2006ElsevierB.V.Allrightsreserved. doi:10.1016/j.landurbplan.2006.02.013 F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 241 Travelcostmodelshavebeenwidelyusedtodeterminethe realestatevaluationcanonlybefullytakenintoaccountbyusing environmentalvalueofrecreationresources.Thebasicpremise thedescriptiveframeworkofaGIS.Oneofthemostbasicadvan- ofthetravelcostmethodisthatthetimeandtravelexpensesthat tagesofaGISistopositionpropertiesonalocalmapinterms peoplespendtovisitasiterepresentthe“price”ofaccesstothe oftheirgeographiccoordinates(Dinetal.,2001).Spatialstatis- site.Travelcostmodelsareappropriatewhenpeopletravelvari- ticswithinaGIS,basedondigitizedremotesensingdata,have ousdistancestoreacharecreationresource.Althoughtheyhave madepossiblethedevelopmentofaccurate,consistent,andunbi- beenusedinurbansituations(Dwyeretal.,1983;Petersonetal., asedexplanatoryvariables,suchasaccessibilitytopublicgreen 1983),theyaregenerallynotthoughttoworkwellforneighbor- spaces,inafastandefficientmanner.Thesecanthenbeusedto hoodrecreationresourcesbecauseoftheverysmalldifferences measure the environmental characteristics of properties better, betweenusersindistancetraveled(Moreetal.,1988). increasingtheunderstandingofhousepricingvariationsascom- Contingent valuation method is the most prevalent method pared to previous studies (Wyatt, 1996; Can and Megbolugbe, usedtoestimateaneconomicvalueforenvironmentalgoods.By 1997;Geogheganetal.,1997;Lakeetal.,2000;Brasingtonand creatingahypotheticalmarketinwhichindividualscanexpress Hite, 2005). Supported by GIS, landscape ecology has expe- their willingness to pay for an environmental good, contin- riencedunprecedentedrapiddevelopmentsincethemid-1980s gentvaluationmethodemployssurveytechniquestoaskpeople (Wu et al., 2000). The creation and development of landscape about the amount they would pay for non-market commodi- metricsbylandscapeecologistshavemadequantificationofspa- tiesifmarketsdidexist(MitchellandCarson,1989;Tyrva¨inen tialstructurepossible.Jointecologicalandeconomicsmodeling and Va¨a¨na¨nen, 1998; Xu et al., 2003). However, because the based on GIS will potentially make a large contribution to the responses do not involve actual market purchases, a distinct valuationofenvironmentalamenitiesandtheunderstandingof drawback of this method is that responses are based on hypo- humanimpactsonurbanecosystems(Geogheganetal.,1997). thetical rather than actual behavior (Tyrva¨inen and Va¨a¨na¨nen, Todate,hedonicpricingmodelshavenotbeenusedtoeval- 1998). uate implicit prices for the amenity of urban green spaces in Since the application of hedonic pricing model to environ- mainland China, nor for any other non-priced environmental mental economics found a statistically significant relationship resources associated with house prices. This was mainly the between property values and air quality (Ridker and Henning, resultofthepreviousrealestatemarketsystem.Inthetraditional 1967),economistshavepaidincreasingattentiontotheinforma- residentialwelfaresystem,thestate,ratherthantheindividual, tionalcontentofhouseprices(McLeod,1984;Geogheganetal., heldrightstohouses.Theoldsystemrestrictedthesaleofres- 1997).Propertyisacomplicatedcommoditywithmanydimen- idential houses in the market as real goods. It was impossible sions,anddifferencesinthesellingpricesofhouseswillbedic- to access and verify house transaction prices. Thus far, quan- tatedbyanumberoffactors,includingthequalityofthehousing titative research regarding urban green space amenities based structure,neighborhoodcharacteristics,accessibilitytothecen- onhedonicpricingmodelshasbeenlimited(WangandHuang, tralbusinessdistrict,aswellastheenvironmental(dis)amenities 2005).However,thenatureofwelfarehousing(FuLiFang)has associatedwiththeproperty.Hedonicpricingmodelsuseamar- beenfundamentallychangedsinceeconomicreformunderthe ketgood(thehouse)asaproxy,withinwhichanon-marketgood, policy of “housing monetarization” (Shang Pin Fang), which suchasanenvironmental(dis)amenity,isimplicitlytraded.They was promulgated in 1998 (Zhu and Guo, 2004). Since then, have been frequently applied in America after being devel- housing has been sold under free market conditions with con- opedbyRosen(1974)(McLeod,1984;CheshireandSheppard, sumersgiventherighttoentertheresidentialmarketandchoose 1995;Tyrva¨inen,1997).Otherempiricalstudieshavealsobeen housingattributesdependingontheirtaste(Yang,2001;Wang conducted in European countries (Garrod and Willis, 1992; and Huang, 2005). This provides an opportunity to conduct a Tyrva¨inen,1997;Tyrva¨inenandMiettinen,2000).Urbangreen study and fill the gap in this research field in mainland China spacesproducebenefitsthatpresumablymakeaneighborhooda by developing and performing a hedonic pricing estimation of niceplaceinwhichtolive,andthesebenefitsarereflectedinthe the benefits deriving from urban green space amenities. How- pricesofsurroundingproperties.Consequently,theamenityof ever,theChineseresidentialmarketisinembryonicform,and urbangreenspacescouldbevaluedinmonetarytermsaccord- needs to be further improved. For example, work units (Dan ingtohowmuchpeoplepayforsuchbenefitsintheirhousing. Wei) bought commodity housing for subsequent allocation to Byusinghedonicpricingmodel,thevalueofanenvironmental their workers still existed. In 2000, only 87.29% of total was amenitycanbeestimatedfromthepricesofrelatedactualmar- soldtotheendusers(ChinaStatisticsBureau,2000).Inaddition, kethousetransactions.Themethodisusuallytermeda“revealed housing prices are a business secret for real estate developers, preferencemethod”inordertodistinguishitfromthestatedpref- whodonotwanttomakepublicdetailsofhousingtransactions. erencemethodssuchascontingentvaluationmethod,whichare Hence,datacollectionregardinghousepricesremainsdifficult. basedonintendedratherthanactualbehavior(Moreetal.,1988; In this paper, real estate transaction prices were captured only Tyrva¨inenandMiettinen,2000). for the year 2004 in Jinan City. Based on the hedonic pric- In recent decades, the development of geographic informa- ing model, we attempted to indirectly evaluate the amenity of tion systems (GIS) has gradually made hedonic pricing model urbangreenspaces.Atthesametime,implicitpricesforother apowerfultool,butatpresent,itisstillunderutilizedinurban environmentalcharacteristicsrelatedtohousing,suchastheedu- andenvironmentaleconomics(BrasingtonandHite,2005).The cationalenvironmentandenvironmentaldisamenities,werealso commondictumthatlocationisthemostimportantparameterfor estimated. 242 F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 This paper is structured as follows: in Section 2 theoreti- andPollakowski(1979)recommendedtheuseoftheBox–Cox calmodelsofhedonicpricingmethodaredeveloped,including flexiblefunctionalforminhedonicanalysis,whichwouldmea- linear and semi-log models; Section 3 describes the data and sure best performance with goodness-of-fit. This functional explanatory variables; in Section 4 the valuation technique is form is commonly applied in the hedonic literatures (Kau and applied to the amenity of urban green spaces and the benefits Sirmans, 1979; Shultz, 2001; Smith et al., 2002). However, are estimated; finally some conclusions are drawn in Section CasselandMendelsohn(1985)havepointedoutadrawbackof 5. The main objectives of this study were to: (1) attempt to thismethodology:thatusingabest-fitcriteriontochoosefunc- apply hedonic pricing models to the valuation of urban green tionalformsdoesnotnecessarilyleadtomoreaccurateestimates spaceamenitiesinJinanCity;(2)searchforsuitablevariablesin ofcharacteristicprices.Followingpreviousstudies(Gretherand hedonicmodelsusingGISandlandscapemetrics;(3)presentthe Mieszkowski,1980;AcharyaandBennett,2001;DesRosierset monetaryvalueofurbangreenspaceamenitiestourbanplanners al., 2002), this paper employs common functional forms (lin- anddecision-makersforoptimizingurbandevelopmentprojects ear and semi-log linear hedonic price functions) with multiple and greening planning. The results of this paper constitute the regressionanalysis. firstempiricalstudyinmainlandChinaaboutquantitativevalua- tionofurbangreenspaceamenitiesbasedonthehedonicpricing 2.1. Linearmodel models. Thelinearhedonicmodelestimatedinthispaperisasfollows: 2. Thehedonicpricingmodel P =α+Eβ+Gγ +Lη+ε (2) Beforeobtainingestimatesusingthehedonicpricingmodel, where P is an (n×1) vector of mean housing cluster prices; theformforthehedonicpricefunctionmustbechosen.Hous- E is an (n×1) matrix of neighborhood environmental charac- ingisnotahomogeneouscommodity(McLeod,1984).Ahouse teristics; G is an (n×1) matrix of accessibility characteristics buyer purchases both a dwelling and a set of site character- (spatial character measured by GIS); L is an (n×1) matrix of istics. The price at which a house is traded is determined by landscapeindices(spatialcharactermeasuredbylandscapemet- a host of factors: housing structure, location, neighborhood, rics);α,β,γ,andηareassociatedparametersvectors;andεisan and environmental characteristics. The household must find (n×1)vectorofrandomerrorterms.Thenotationintheabove the dwelling with the best combination of features at the best equation is slightly different from that in Eq. (1), with inde- price (O’Sullivan, 2002). The implicit prices of spillover ben- pendentvariablesgroupedintothreetypes.Table1providesa efits from the neighborhood or environmental characteristics detaileddescriptionofeachvariableusedinthehedonicpricing canbeinferredbasedonthehedonichypothesis(Rosen,1974). model.Themodelingprocedureinvolvedenteringallvariables Consequently, the traditional hedonic pricing model takes the of a specific group into an ordinary least squares regression, followingform: andexaminingtheadjustedR2.Theanalyzedresultsareshown P =f(x1,x2,...,xn) (1) in Table 2. However, as Rosen (1974) points out, there is no reasontoexpecttherelationshipbetweenthepriceandtheenvi- where P is the market price of the housing and x , x , ..., x ronmental variable to be linear. In fact, non-linearity is to be 1 2 n are the characteristics contained in the property. These char- expected, because purchasers cannot treat individual housing acteristics can be categorized into three broad groups of vari- attributes as discrete items from which they can pick and mix ables.Structuralvariablesrelatetothedirectcharacteristicsof untilthedesiredcombinationofcharacteristicsisfound(Garrod theproperty;accessibilityvariablesdefinetheeasewithwhich and Willis, 1992; Morancho, 2003). Thus, semi- or double- amenitiescanbereachedfromtheproperty;andneighborhood logarithmic specifications are frequently formulated, although environmentalcharacteristicsvariablesdescribethequalityofits linearmodelsarestillinusebecauseoftheeasewithwhichthe surroundings,suchasindustrialpollutionandindicatorsregard- parameterscanbeinterpreted(AcharyaandBennett,2001;Des ingthetypeandstructureofland-use(Freeman,1993;Lakeet Rosiersetal.,2002;Huntetal.,2005). al., 2000). In this paper, 17 variables were chosen in the con- structionofhedonicpriceindices.Thevariables,withdetailed 2.2. Semi-logmodel descriptions,canbefoundinTable1. The hedonic pricing model employed in the evaluation of Inthispaper,thesemi-logmodelwasspecifiedbyanatural recreation resources typically uses multiple regression tech- log transformation of the housing price in the hedonic regres- niquestorelatepropertypricedetailstodiversecharacteristics sions.Themodelisasfollows: ofdifferingproperties,andtosortoutthedifferentcontributions. Manyeconomistshavestressedthateconomictheorydoesnot ln P =α+Eβ+Gγ +Lη+ε (3) suggestanappropriatefunctionalformforhedonicpriceequa- tions(Rosen,1974;Freeman,1979;HalvorsenandPollakowski, Inadditiontothedirectlinearandsemi-logregressionmodel- 1979; Cassel and Mendelsohn, 1985). Consequently, it is rea- ing(Table2),anotheranalysiswasperformedusingastepwise sonable to try several functional forms and utilize a multiple regressiontechniqueintheSPSSsoftwaretoeliminatecollinear regressionequation(CasselandMendelsohn,1985).Halvorsen variables(Table3). F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 243 Table1 Variabledescriptionsandtheirexpectedeffectsonhousetransactionprices Independentvariable Description Expectedsigna S-DPARK Ameasureoftheintegratedimpactofthenearestparksize(Spa)andstraight-linedistancefromthepropertyto + theboundaryofthenearestpark(Dpa),S-DPARK=ln(Spa/Dpa) S-DPLAZA Ameasureoftheintegratedimpactofthenearestplazasize(Spl)andstraight-linedistancetotheboundaryofthe + nearestplaza(Dpl),S-DPLAZA=ln(Spl/Dpl) S-DSCEN Ameasureoftheintegratedimpactofthenearestsceneryforestsize(Ssc)andstraight-linedistancetothe + boundaryofthenearestsceneryforest(Dsc),S-DSCEN=ln(Ssc/Dsc) TYPEGR Typeofthenearesturbanpublicgreenspace.Plaza,park,andsceneryforestareweightedwith1,2,and3, + respectively ACPARK Ameasureoftheeaseofaccessibilitytoapark;accumulatedtimecostfromthecentroidofthehousingcluster − tothenearestpark(inmin) ACPLAZA Ameasureoftheeaseofaccessibilitytoaplaza;accumulatedtimecostfromthecentroidofthehousingcluster − tothenearestplaza(inmin) ACSCE Ameasureoftheeaseofaccessibilitytoasceneryforest;accumulatedtimecostfromthecentroidofthehousing − clustertothenearestsceneryforest(inmin) ACCBD AmeasureoftheeaseofaccessibilitytotheCBD;accumulatedtimecostfromthecentroidofthehousing − clustertothecentralbusinessdistrict(inmin) PLGR Ameasureofpercentageofthelandscapethatisurba(cid:2)n(cid:3)greenspace(cid:4)ina300mradiuswindow,indicatingthe + greeningratioofthehousingcluster,PLGR=Pg= nj=1agj/A ×100,whereagjisthegreenspaceareaof patchj,andAisthewindowsize NPGR Ameasureofthefragmentationofurbanpublicgreenspacesina500mradiuswindow.NPGR=Ng,whereNgis ? thetotalnumberofgreenspacepatchesinonewindow PRGR Patchrichnessoftheneighboringurbanpublicgreenspacesina500mradiuswindow,PRGR=m,wheremis ? thenumberofurbangreenspaces(classes)presentinonewindow AIGR Ameasure(in%)ofaggregationofpublicgreenspacesina500mradiuswindow,AIGR=(gii/maxgii)×100, ? wheregiiisthenumberoflikeadjacencies(joins)betweenpixelsofpublicgreenspaces(i);maxgiiisthe maximumnumberoflikeadjacencies(joins)betweenpixelsofpublicgreenspaces(i),basedonthesingle-count method NPLA Ameasureofpatchdensityoftheneighboringland-useina500mradiuswindow,NP=Nl,whereNlisthetotal ? numberofland-usepatchesinonewindow PRLA Patchrichnessoftheneighboringland-useina500mradiuswindow,PRLA=m,wheremisthenumberof ? land-usepatchtypes(classes)presentinonewindow (cid:2) (cid:3) (cid:4) EDM Ameasureofenvironmentdisamenity,EDM=(n/d)= n2/ ni=1di ,wherenisthenumberoffactories − withina1000mbuffer,diisthedistancefromthecentroidofthehousingclustertofactoryi;orEDM=1/D, whereDisthenearestdistancefromthecentroidofthehousingclustertoafactorywhenthereisnofactory withinthe1000mbuffer (cid:2) (cid:3) (cid:4) EE Ameasureofeducationenvironment,EE=(n/d)= n2/ ni=1di ,wherenisthenumberofuniversities + withina500mbuffer,anddiisthedistancefromthecentroidofthehousingclustertouniversityi;orEE=1/D, whereDisthenearestdistancefromthecentroidofthehousingclustertoauniversitywhenthereisno universitywithinthe500mbuffer DLOC Ameasureofthelocationsectorofhousingcluster.Thesamplesaredividedintofourparts:north,west,east, + andsouthandweightedwith1,2,3,and4,respectively a +and−representincreasinganddecreasingeffects(respectively)onthehousetransactionprice;?indicatesanaprioriundeterminedsign.Definitionsoflandscape metricsarebasedonMcGarigaletal.(2002). 3. Dataanddefiningexplanatoryvariables up areas have expanded from 24.6km2 in 1949 to more than 190km2in2003(JinanStatisticsBureau,2003).JinanCityowns The characteristics or variables to be included in the esti- atotalofapproximately105.6km2 publicgreenspaces(classi- matesmustbedeterminedbeforeobtainingsuchestimatesusing fiedassceneryforests,plazas,andparks)(KongandNakagoshi, hedonic pricing model. This decision may involve recognition 2005),amountingtoabout19.62%ofthetotalstudyarea.There ofrelevantvariablesthatareomittedfromtheanalysisanddeter- arecomparativelyfewpublicgreenspacesinthebuilt-uparea, minationofhowtocapturethechosenvariables(Cheshireand and the majority (91.1%) are found in the urban fringe, espe- Sheppard,1995). cially on the south and southeast hills. Jinan City is known as “CityofSprings”forthereareatleast72famousspringsinthe 3.1. Housepriceandstructurecharacteristicsdata city. The sources of these springs are the southern and south- easternsceneryforestareas(Fig.1).Hence,conservationofthe ThisstudywasconductedinJinanCity,thecapitalofShan- southern and southeastern scenery forest is a key issue (Kong dong Province in China. Jinan City has experienced dramatic andNakagoshi,2005).However,greenspaceamenitiesattract population growth (from 3.19million in 1952 to 5.75million migrants, and the rapid urban development has encroached on in 2002) and has sprawled greatly in the last 50 years. Built- muchoftheurbangreenspaceinrecentdecades,especiallyinthe 244 F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 Table2 Regressionresultsoflinearmodelandsemi-logmodel,dependentvariable:inprice Independentvariable Linearmodel(R2=0.691,adjusted Independentvariable Semi-log model (R2=0.698, R2=0.642) adjustedR2=0.650) Coefficients t-Ratio Coefficients t-Ratio Constant 4205.716 3.602 Constant 8.436 24.365 S-DPARK 43.309 0.753 S-DPARK 1.837E−03 0.108 S-DPLAZA 213.253 1.894 S-DPLAZA 4.612E−02 1.382 S-DSCEN 86.160 1.479 S-DSCEN 1.873E−02 1.084 TYPEGR 280.159 1.562 TYPEGR 1.539E−02 0.289 ACCBD −22.226 −1.222 ACCBD −6.023E−03 −1.117 ACPARK −11.873 −0.368 ACPARK −1.137E−02 −1.189 ACPLAZA −37.478 −1.074 ACPLAZA −9.529E−03 −0.921 ACSCE −15.937 −0.622 ACSCE −7.304E−03 −0.961 PLGR 62.326 4.537 PLGR 2.030E−02 4.983 AIGR 5.403 1.356 AIGR 1.876E−03 1.588 NPGR 429.778 2.971 NPGR 0.138 3.228 PRGR −1130.393 −2.922 PRGR −0.351 −3.061 PRLA −377.081 −4.088 PRLA −0.118 −4.317 NPLA −4.129 −0.662 NPLA −2.720E−03 −1.471 DLOC 194.435 2.078 DLOC 7.476E−02 2.695 EE 69.402 2.640 EE 1.722E−02 2.208 EDM −18.518 −0.796 EDM −4.246E−03 −0.616 southern scenery forest area. Proper management is needed to ture and public services (Ministry of Construction, PR China, preventencroachmentontogreenspacesandtoreallocatethem, 1998). butoptimizingthespatialpatternofurbangreenspacerequires Atthestartofthestudy,variablesrelatingtostructuralchar- evaluationoftheeconomicvaluesofurbangreenspaces.There- acteristics were designed, including the age of the house, and fore, quantitative information in monetary terms concerning whetherithadgas,internet,bathroom,orheating,ashasbeen greenspacebenefitsisurgentlyneededforfutureurbanplanning donebymostpreviousresearchers(McLeod,1984;DesRosiers inJinanCity. etal.,2002).However,aparticularweaknessofthehouseprice Housing prices have fluctuated considerably since reform surveywasthatonlylimitedinformationwasavailableregard- of the real estate market in 1998, so the data collection sur- ing differences between property structures. These properties vey was only carried out in August, 2004. All the surveyed wereconstructedinthelasttwoyears,andthosefromthesame houses were built between 2003 and 2004, and sold in 2004, housingclusterwerebuiltbythesamerealestatedeveloper.Con- at a time when local house prices were most stable. A single sequently,thesurveyedpropertiesarelikelytoexhibitstructural year of data was used to avoid problems with time variations, similarities, not only in terms of their age but also in terms of but this resulted in a relatively small sample. In the end, 124 theirlayoutandinteriorandexteriordesignfeatures.Ingeneral, housingclusterswereselected.Thehousingclustersarelocated property structure characteristics appear to differ mostly with withintheurbanareaandcomparedbyroads,andareoftenbuilt size.Somerelatedpapershavedemonstratedthatpropertysize byasinglerealestatedeveloper,withvaryingbasicinfrastruc- has a positive relationship with selling price (McLeod, 1984; Table3 Regressionresultsaftereliminatingcollinearvariables,dependentvariable:inprice Independentvariable Linearmodel(R2=0.648, Independentvariable Semi-log model (R2=0.673, adjustedR2=0.620) adjustedR2=0.647) Coefficients t-Ratio P-value Coefficients t-Ratio P-value Constant 3735.850 4.666 0.000 Constant 8.172 37.838 0.000 S-DSCEN** 141.094 3.275 0.001 S-DSCEN** 4.919E−02 4.094 0.000 PRLA** −417.878 −5.319 0.000 PRLA** −0.123 −5.084 0.000 PLGR** 63.546 4.997 0.000 PLGR** 2.055E−02 5.702 0.000 ACPLAZA* −51.631 −1.737 0.085 ACPLAZA* −1.849E−02 −2.571 0.011 EE** 71.813 2.964 0.004 DLOC* 6.295E−02 2.529 0.013 DLOC* 155.359 1.746 0.084 ACPARK −1.642E−02 −2.405 0.018 ACPARK −32.087 −1.379 0.171 EE** 1.841E−02 2.648 0.009 TYPEGR** 370.239 2.313 0.022 NPGR** 0.139 3.283 0.001 S-DPLAZA* 188.579 1.760 0.081 PRGR** −0.192 −2.780 0.006 * Indicatesstatisticalsignificanceatthe10%level. ** Indicatesstatisticalsignificanceatthe5%level. F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 245 Fig.1. Thestudyareaandgeographicdistributionofthe124sampleproperties. Tyrva¨inen and Miettinen, 2000; Morancho, 2003). The major categories(plaza,park,andsceneryforest)andweighted(with focus of this investigation was not to search for a comprehen- 1, 2, and 3, respectively) according to their expected amenity siveexplanatorymodelofhousepricedetermination,butrather effect.Thatis,morehighlyweightedgreenspaceisexpectedto todemonstratethecontributionofurbangreenspacevariables. haveapositiveandhigherimpactonhousingsaleprice.When Therefore,variablesrelatingtohousestructuralcharacteristics asizevariableforthenearesturbangreenspacewaschosen,the were dropped from the final analysis, and the real transaction question of spatial pattern (with the exception of green space property price per square meter of each housing cluster was type), was usually ignored. However, particularly uneven dis- chosentobethedependentvariableinthefinalmodel.Thereby, tribution patterns of property and green space may influence the property location in this paper was the geometrical center the analysis results. For example, as shown in Fig. 2(a), the of each housing group and was determined by GIS. A map of nearesturbangreenspacetopropertiesH andH isthesame 1 2 the study area including the location of the sample properties (S). If the other variables effecting housing price are ignored, shows that most properties are located within the contiguous the price of H would be the same as that of H . But in fact 1 2 built-upareaofJinanCity,withrelativelyfewinitsperipheral H has a higher price than H , because the effects of amenity 2 1 area(Fig.1). onthehousesalepriceareinaccordancewiththegeneraldis- tancedecaytheory,asshowninFig.2(b)(CanandMegbolugbe, 3.2. Neighborhoodenvironmentalcharacteristicsdata 1997; Donnelly, 2005). Including both the distance to and the sizeofthenearestgreenspaceasindependentvariablesinthe Thesetofexplanatoryvariablesrelatestotheneighborhood hedonic model would result in a biased estimation. Suitable environmentalattributesoftheproperties,includingurbangreen transformationofthetwovariableswasthereforeneededbefore space,educationenvironment,andenvironmentaldisamenities conductingtheregressionmodeling.Consequently,aninterac- (e.g.,industrialpollution). tionterm,the“size–distanceindex”,wasdevelopedasln(S/D), To evaluate the effect of an amenity on the selling price of where S is the size of the nearest green space and D is the a given house, previous studies generally have regressed vari- distance to the nearest green space. This index was incorpo- ablesthatdefineurbangreenspacecharacteristics,suchastype ratedinandaddedtothehedonicpricingmodeltoestimatethe of green space, the distance from the property to the nearest amenityeffectforproximitytoacertainsizegreenspace,allow- greenspace,andthesizeofthenearestgreenspace(Tyrva¨inen ing the effect of distance to vary with the size of the relevant andMiettinen,2000;Morancho,2003).Differenttypesofurban green space. The size–distance index was classified into three greenspacemayinfluencehousingprices,whichshouldbetaken types: S-DSCEN, S-DPARK, and S-DPLAZA, and they were intoconsiderationinmodels(Tyrva¨inenandMiettinen,2000). allassumedtohaveapositiveimpactonthehousingsaleprice Here,themainurbangreenspacetypesweredividedintothree (Table1). 246 F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 Fig.2. (a)Descriptionofthepossibleinfluenceofdistancetothenearesturbangreenspaceonthehousingprice,causedbythespatialpatternofgreenspacesand properties;(b)distance–pricerelationshipinaccordancewiththegeneraldistancedecaytheory. Schools are an important local public service that affects survey of households in Jinan City, conducted by Zhang and house prices (Goodman and Thibodeau, 1998; Haurin and Wang(2003)andSun(2005)foundthatmostconsumerswould Brasington, 1996). It is a common belief that people consider prefer residences in the south and east rather than in the west theeducationenvironmentwhenbuyinganewhouse,andmany andnorth.Thefoursectors(north,west,east,andsouth)were researchershavefoundthatpropertyvaluesarehigherwherethe givenweightsof1,2,3,and4,respectively.Iftheseweightsare education environment is better (Hanushek, 1986; Brasington accurate,positiveregressioncoefficientscanbeexpected. and Hite, 2005). In this study, universities were chosen as a proxy to perceive the effect of the education environment. It 3.3. Spatialcharacteristicsdata wasfocusedontheamenityeffectrelatedtodistancefromthe nearestuniversitytoagivenpropertyandtheuniversitydensity 3.3.1. Accessibilityvariables (numberofuniversities)withinaspecifieddistancesurrounding Accessibility variables define the ease with which local theproperty.Thus,thereisapremiseinthispartoftheanaly- amenities can be reached from a property (Lake et al., 2000). sisthatalluniversitiesareofthesamequality.Theinteraction In this study, the measure of accessibility was the calculated proxyvariableofeducationenvironment,calculatedusingspa- accumulatedtraveltime(min)fromeachpropertytoeachofthe tialstatistics,wasadaptedtomeasuretheimpactoftheeducation public green spaces and the central business district by public environment.Educationenvironmentwasdefinedasthenumber bus or on foot along various road types. A GIS-based spatial ofuniversitieswithin500mbufferaroundeachhousinggroup, analysisprovidesapowerfulmeansofdeterminingtheaccessi- dividedbythemeandistance.Iftherewasnouniversitywithin bilityvariables. a500mbuffer,thereciprocalofthenearestdistancewaschosen Thefirststageincalculatingaccessibilitywastogeneratea (Table1). land-usemapofthestudyareafromremotesensingdata(2004 Incontrasttothepositiveeffectofamenities,pollutionsuchas SpotImages,10m,4bands),whichincludedtheroad(Fig.1) the“threewastes”(wastegas,wastewater,andwasteresidues) and walking networks; a vector topographic map (1:10000) dischargedbyindustrywouldimpingedirectlyonthesurround- created in 2000 was used as a reference. Road network land- inginhabitantsandresultindisamenityeffectstotheproximate use types were defined as arterial road, secondary truck road, properties(Adleretal.,1982;Schulzeetal.,1986;Brasington branchroad,andpath;andwalkingnetworkland-usetypeswere and Hite, 2005). Previous research has generally found a sta- definedasriver/water,mountains,andothers(Table4).Thenext tistically significant price–distance relationship regarding the stage involved transforming the land-use map to grid data of disamenity effects of these sites (Michaels and Smith, 1990). 10mresolutionandassigningdifferentimpedancevaluestothe However, usually not only the distance but also the density of gridcells(namedcost-distance)todepictthetimecosttotravel theindustrysurroundingapropertywillaffectthehousingprice. 10km along different land-use types by public bus or on foot Consequently,aproxyvariablefortheenvironmentaldisamenity (Table 4). For the road network, the cost-distance was defined ofindustrialpollutionwasdevelopedthathasananalogywith as the time (min) it would take a public bus to travel between the education environment, except that the window size was a each property and the nearest public green spaces or central 1000mradius(Table1). businessdistrict.However,thewalkingnetworkdiffersfromthe Locationsectordifferencesalsoappeartohaveanimpacton roadnetwork,becausetherearepathsalongwhichonlywalkers residentialprices.Thestudyareawasdividedintofoursectors can travel or roads unsuitable for pedestrians. Thus, the walk- based on their different development histories, land-use com- ingnetworkcost-distancewasestimatedbasedonthepossible positions,andurbanfunctions,asdesignatedintheJinanCity walkingspeed.Thedistance/costweightedsamplingtoolinthe masterplanning2004–2020(JinanPlanningBureau,2004).A GIS spatial analysis module was then utilized to calculate the F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 247 Table4 state,theintegrationofeconomicsandecologyishamperedby Definitionsoftravelspeedandcost-distance(min/10km)alongdifferentland- thefailuretoincludespaceineconomictheoriesandmodels.The usetypes joiningoflandscapeecologyandeconomicswouldcontributeto Land-usetype Travelspeed Cost-distance the promotion of improved modeling approaches and relevant (km/h) (min) techniques in the field of ecological economics. In this study, Roadnetwork Arterialroad 30 20 in an attempt to capture the impact of green space and land- Secondarytruckroad 25 24 usespatialpatternonthesurroundingproperties,sixlandscape Branchroad 20 30 spatial indices (percentage of urban green space-PLGR, patch Path 15 40 richness of urban green space-PRGR, number of green space Walkingnetwork River/water 0.75 800 patches-NPGR,aggregationofurbangreenspace-AIGR,patch Mountains 1.2 500 richness of land-use-PRLA, and number of patches of land- Others 4 150 use-NPLA)wereaddedtothehedonicpricingmodel.Allwere selectedaccordingtotheirpresumedeconomicrelevance.Itwas timeitwouldtakeapublicbusorawalkertogettothenearest hopedthatthespatiallandscapemetrics,whichwerecharacter- amenities(greenspacesorthecentralbusinessdistrict). ized by differing richness, density, or aggregation, might add explanatory power to the hedonic function, and consequently 3.3.2. Landscapemetrics enrichthemodel.Tocapturetheselandscapemetricsinalocal Beyond the preceding traditional variables used to explain area,a“movingwindow”analysiswasconductedsupportedby propertyvalue,itwasassumedthatthespatialpattern(composi- FRAGSTATS(version3.3),withwindowsizesof300and500m tionandconfiguration)ofsurroundinggreenspaceandland-uses radius(McGarigaletal.,2002).Thus,thecompositionindices wouldalsoaffectthehousingsaleprice.Inthepasttwodecades, (percentage of green spaces, patch richness of green spaces, landscape ecology has become one of the most rapidly devel- numberofpatchesofgreenspaces,patchrichnessofland-use, opingecologicalfieldsworldwide.Onereasonisthatitenables andnumberofpatchesofland-use;seeTable1fordescriptions) spatial pattern analysis through its extensive use of landscape werespatiallyreferencedwhichallowedtheirspatialvariability metrics(O’Neilletal.,1988;Gustafson,1998;LiandWu,2004). tobevisualized(Fig.3)(KongandNakagoshi,inpress).There- The spatial pattern quantified by landscape indices is related after, a sampling tool in the GIS spatial analysis module was not only to ecological functions but can also reflect the socio- usedtocapturequantitativeinformationatawindow-sizescale economicfunctionsofthelandscape.AsCostanzaetal.(1997) aroundtheproperty. Fig.3. Greenspacerichnessaroundthepropertywith500mradiuswindowsizewasquantifiedbypatchrichnessmetricsandobtainedbythe“movingwindow” method. 248 F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 Landscapepatternsarescale-dependent(Wuetal.,2002),as of green space patches (NPGR) are both positive. This result are the amenity impacts of urban green spaces or other land- is somewhat surprising because aggregation of green spaces use types on housing prices (Geoghegan et al., 1997; Mitsch (AIGR) presents the degree of aggregation, whereas number and Gosselink, 2000). However, most landscape indices are of green space patches (NPGR) indicates the fragmentation sensitive to changes in scale (resolution or extent) (Turner et of public green spaces. This result can be attributed to the al., 1989; Wu et al., 2000; Li and Wu, 2004). Consequently, scale chosen for this study. The 500m radius window around in this paper all indices (with one exception) were calculated the property is sufficiently small such that all public green fromafixed500mradiuswindowtoavoidscaleeffectsinthe spaces,whetheraggregatedorfragmented,servetoincreasethe analysis. The sole exception is the percentage of urban green valueofneighboringproperties.Thusitcanbeconcludedthat, spaces (PLGR), which was captured in a 300m radius win- given the shortage of and resultant pressure on land in urban dow. The percentage of urban green spaces includes not only areas, the insertion of green spaces (even if fragmented) and thethreetypesofpublicgreenspacebutalsotheattachedgreen anincreaseinthepatchnumberwillproducepositiveamenity spaces,suchasresidentialgreenspace,roadsidegreenspaces, valuesandbefavoredbypeople,indicatingthatthisisaprefer- andripariangreenspace(KongandNakagoshi,inpress).With able approach for increasing green space in urban settings. In a 300m radius window surrounding a property, percentage of additiontoaggregationofgreenspaces(AIGR),percentageof urbangreenspaces(PLGR)couldindicatethegreeningratioin greenspaces(PLGR)exhibitedapowerfulpositiveinfluenceon thehousinggroup.Severaltrialsweremadetotesttheimpactof housingprices,andthevarioussize–distanceindicesalsohada changesinwindowsizeontheestimatedresults,anditproved positiveimpactonhousesaleprices.Ingeneral,thedecreased that the 500m radius window (300m for percentage of urban accessibilityofpublicgreenspacesandcentralbusinessdistrict greenspaces)wasappropriatetocapturetheamenityeffect,as all have a relatively weak, but significant negative impact on expected. the property values. The positive price impact of green space Theland-usepatternisamajorcontributortoqualityoflife types (TYPEGR) indicates that various types of public green andthusinfluencestheneighborhoodhousingprice(Geoghegan space with varying attributes differ in their impact on housing etal.,1997;AcharyaandBennett,2001).Thenumberofpatches price. The spatial landscape metrics for land-use, patch rich- and richness, indicative of the fragmentation and diversity of ness of land-use (PRLA) and number of patches of land-use land-useatacertainscale(500mradiuswindow),wereadded (NPLA), both demonstrate a negative impact on neighboring tothehedonicpricingmodel.Theland-usedatasetusedinthe houseprices.Theenvironmentaldisamenity(EDM)appearsto studyareawasreclassifiedintoeighttypes(residential,industry, haveaweakbutstatisticallysignificantrelationshipwithhouse publicfacility,trafficandroad,greenspace,agriculture,water, prices,whichconfirmsthatnearbyindustrialpollutantsdepress and others) from the standard for classification of urban land- houseprices. useinChina(MinistryofConstruction,PRChina,1990),based Beforemakingadetailinterpretationoftheestimatedcoef- onthefunctions,land-usecover,andownership.Theimpactof ficients in Table 2, the model was improved by eliminat- land-usepatternsonhousingpriceisdeterminedbyhowpeople ing collinear variables and non-significant variables through perceiveaparticularlocation,aswellasbythescalechosenfor a stepwise approach. The results of this estimation show that theanalysis(Geogheganetal.,1997).Higherdiversityandfrag- size–distance index of scenery forest (S-DSCEN), patch rich- mentationoflandproximatetoresidentialhousingmayproduce nessofland-use(PRLA),percentageofgreenspaces(PLGR), negativeeffectsonpropertyvaluesbyintroducingnoiseorlight greenspacetypes(TYPEGR),andeducationenvironment(EE) pollutionandnegativevisualimpacts.Ontheotherhand,these are all statistically significant at the 5% level (Table 3). The mayhaveapositiveimpactbyprovidingconvenience,suchas natural logarithm of the selling price yielded a slightly better closeproximitytoshopping,work,andrecreation.Conversely, functionalformforthedependentvariable,whichwasslightly a simple district marked mainly by residential uses may result improved for two variables, that is, number of green space inaquietandcomfortableenvironment.Thus,duetothecon- patches (NPGR) and patch richness of green spaces (PRGR), flictingeffects,thesign(+or−)ofpatchrichnessofland-use enteringthemodel,butontheotherhand,thenaturallogarithm (PRLA) and number of patches of land-use (NPLA) remains of the selling price prevented the variables: green space types ambiguous (indicated with the sign “?” in Table 1) before the (TYPEGR)andsize–distanceindexofplaza(S-DPLAZA),from hedonicestimationisperformed. beingincludedinthemodel.Thesemi-logformisexpressedas relativeratherthanabsoluteimplicitprices;consequently,itcan 4. Results beusedtogiveaflexibleinterpretationoftheestimatedcoeffi- cients(DesRosiersetal.,2002).Thefindingscanbesummarized The results of the parameter estimates are presented in asfollows: Table 2. The estimated coefficients generated using both the linearandsemi-logmodelsexhibitedsignsinaccordancewith (1) Theconstantispositiveduetothebasevalueoftheproper- expectations. They provide a clear indication that the features ties,regardlessofadditionalfeatures.However,comparison ofurbangreenspacedoexertasignificantimpactonproperty ofthevalueoftheconstant(transformedtoChinesemone- prices. taryunit-RMB3539.4)withtheaveragehousingpriceofthe A particularly noteworthy result of the analysis is that the 124samples(RMB3465.3),indicatesthatthechosenvari- impact of aggregation of green spaces (AIGR) and number ablesinthehedonicpricingmodelyieldedgoodexplanatory F.Kongetal./LandscapeandUrbanPlanning79(2007)240–252 249 power and also reflect the preferences of homeowners in (8) Finally, the positive price impact of the education envi- JinanCity; ronment suggests people tend to purchase houses in good (2) The size–distance index of scenery forest (S-DSCEN) university districts. The good education environment is a revealed that proximity to scenery forest had a pos- type of substitute for urban green space amenities, imply- itive amenity impact, raising the house value by 5% ing that homeowners may trade off amenity to get a good for each percentage increase in the index. However, the educationenvironment.Onaverage,a1%increaseinedu- othertwosize–distanceindices(size–distanceofparkand cationenvironmentisassociatedwithaconsistent1.9%rise size–distanceofplaza,S-DPARKandS-DPLAZA,respec- inhouseprice(persquaremeter). tively) did not show significance even at the 10% level. The higher attractiveness of scenery forest is not without 5. Discussionandconclusions its drawbacks, as 90.5% of scenery forests are located in thesouthandsoutheastofJinanCity,whichisthesourceof This paper reports a first attempt to apply hedonic pricing thespringssuchasBaotuQuanspring,andnownegatively models to evaluation of the amenity of urban green spaces in impactedbyencroachmentanddevelopment; Jinan City, China, and highlights the fact that hedonic pricing (3) Thepatchrichnessofland-usevariable(PRLA)ina500m modelswouldbeimprovedbyembodyingspatialcharacteristic radiuswasfoundtohaveapowerfulandnegativeinfluence variables captured by GIS and landscape metrics; it thus con- on housing prices, indicating that people prefer to live in tributes to the further joining of economics and ecology, and a quiet and comfortable environment, with homogeneous promotesdevelopmentofecologicaleconomics. land-useintheimmediatevicinityoftheirhouses; (4) Properties with a higher percentage of green space area 5.1. Specificationandcaptureofcharacteristicvariablesin (PLGR) within a 300m radius have higher house values, thehedonicpricingmodel with each percentage point of green space adding about 2.1%tothepricepersquaremeter; Thisstudyinvestigatesonlythemeanvaluepersquaremeter (5) Accessibility to scenery forest (ACSCEN) was not sig- of a housing cluster and focuses solely on the neighborhood nificant at the 10% level, whereas accessibility to plazas environment,especiallyspatialconcerns,whereashousestruc- (ACPLAZA)andparks(ACPARK)weresignificantat5% turalcharacteristicsandtimeseriesproblemswerebothavoided. level.Accessibilitytoplazasdisplayedslightlyhighersig- The omission of variables relating to structural characteristics nificance than accessibility to parks. A 1% improvement of housing was rarely found in previous research that applied inaccessibilitytoplazasandparkswillleadtoanincrease hedonic models. Thereby, it is possible that the results of this inpricepersquaremeterof1.8and1.6%,respectively.In analysismightsufferfromsampleselectionbias.However,the Jinan City, 93.5% of plazas and 70.2% of parks, but only resultswereconsistentwithexpectations,suggestingthatsam- 0.9%ofsceneryforestareasarelocatedwithinthebuilt-up pleselectionbiasisnotanissue. area;mostofthelatterarelocatedinthesouthandsoutheast Thecreationoftheinteractiontermseducationenvironment urban peripheral area (Fig. 1). The convenient transporta- and environmental disamenity and the size–distance indices tionwithinthebuilt-upareamakesiteasiertoaccessparks allowed refined specification variables to be used in the hedo- andplazasthantoaccesssceneryforests.But,comparison nicpricingmodel,whichdifferentiatesthisfromsomeprevious ofparksandplazasrevealedtheydifferedmainlybecauseof studies.Asisusualwithhedonicpricingmodels,manyfactors thesurroundingland-usetype.Theplazadesignationdefines werelistedthatmayinfluencepropertyvalues,anditisconse- green spaces of an open type, most of which are located quentlyoftendifficulttosortouttheeffectsofspecificvariables nearmainroads(Fig.1),whereasgreenspacesdesignated withgreataccuracy. asparksarealwayssurroundedbywallsorwater,whichwill Theintegrationofeconomyandlandscapeecologyisawor- form a serious barrier in terms of accessibility and reduce thygoalthatwouldbenefitgreatlyfromsystematicanalysisthat theiramenityeffect; includesspatialscale,ratherthanusingonlyaggregatedstatis- (6) The significant and positive impact of the location sector tical data (Vermaat et al., 2005). In this paper, six landscape (DLOC)provedthatpeoplewouldprefertoliveinthesouth metrics(percentageofurbangreenspace-PLGR,patchrichness andeastratherthanthewestandnorth; of urban green space-PRGR, number of green space patches- (7) The positive significant impact of number of green space NPGR,aggregationofurbangreenspace-AIGR,patchrichness patches(NPGR)furtherconfirmedthatincreasingthepatch ofland-use-PRLA,andnumberofpatchesofland-use-NPLA) numberofpublicgreenspacesraisesthehousevalue.How- wereaddedtothehedonicpricingmodel,whichallowedamuch ever, the negative sign of patch richness of green space more complete explanation of urban green spaces and land- (PRGR)indicatesthatincreasingthistypeofpublicgreen use. Moreover, the inclusion of the “moving window” method spacewithina500mradiusaroundapropertywilldepress (supported by FRAGSTATS) allowed the composition indices thehouseprice.Greaterrichnessofurbangreenspacetypes (percentage of green spaces, patch richness of green spaces, willbringmoreopportunityforrecreation,butatthesame numberofpatchesofgreenspaces,patchrichnessofland-use, timewillhavenegativeeffectsontheresidentslivingnearby, and number of patches of land-use) to be spatially referenced, suchasnoise,andneonlightsatnight.Therichnesseffect and allowed their spatial variability to be visualized. Clearly, mayalsoberelatedtotheresearchscale; thedevelopmentofGIShasbeenamajordrivingforcehere,for

Description:
non-market price benefits from urban green space is urgently required. Properties bought and . have been frequently applied in America after being devel- oped by immigrants and protection of scenery forest as source areas for.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.