sign in sign up
<<
Home , Urban heat island

sustainability

Article A Method for Selecting the Typical Days with Full Urban Heat Island Development in Hot and Humid Area, Case Study in Guangzhou, China

Guang Chen 1,2, Minjie He 1, Nan Li 1, Hao He 1, Yunnan Cai 1 and Senlin Zheng 1,3,*

1 School of Architecture and , Guangdong University of Technology, Guangzhou 510090, China; [email protected] (G.C.); [email protected] (M.H.); fi[email protected] (N.L.); [email protected] (H.H.); [email protected] (Y.C.) 2 State Key Laboratory of Subtropical Building Science, South China University of Technology, Guangzhou 510641, China 3 Department of and Regional Planning, The Ohio State University, Columbus, OH 43210, USA * Correspondence: [email protected]

Abstract: The urban heat island (UHI) poses a significant threat to urban , human health, and urban energy systems. Hence, days with a relatively higher UHI intensity should be selected for UHI observation and analysis. However, there is still a lack in the method and criteria for selecting the typical meteorological days for UHI survey and simulation. In this study, field measurements were conducted based on Local Climate Zone (LCZ) schemes over a one-year period to assess the UHI behavior in Guangzhou, China. The relationship between the diurnal range (DTR) and UHI intensity was evaluated and analyzed quantitatively under different meteorological conditions classified by . The average daily maximum UHI intensity (UHIImax) during precipitation days was approximately 1.8 ◦C lower than that during non-precipitation days, confirming that   precipitation has a negative effect on UHI development. The monthly DTR distribution was similar to the daily UHIImax distribution, which was higher in autumn and , but lower in spring Citation: Chen, G.; He, M.; Li, N.; He, and . DTR has a significant linear correlation with the daily UHIImax, with a Pearson’s H.; Cai, Y.; Zheng, S. A Method for correlation coefficient of >0.7 and statistical significance of <0.001. Based on a quantitative evaluation Selecting the Typical Days with Full of our results, we determined that 10 ◦C could be regarded as the appropriate DTR threshold to Urban Heat Island Development in Hot and Humid Area, Case Study in identify the meteorological conditions conducive to UHI development; the meteorological conditions Guangzhou, China. Sustainability exhibited a high daily UHIImax in Guangzhou. This study provides a simple method to select typical 2021, 13, 320. https://doi.org/ meteorological days for UHI measurement and simulation, and a method to early-warning of intense 10.3390/su13010320 UHI events based on weather forecasts.

Received: 7 December 2020 Keywords: urban heat island; precipitation; diurnal temperature ranges; threshold Accepted: 27 December 2020 Published: 31 December 2020

Publisher’s Note: MDPI stays neu- 1. Introduction tral with regard to jurisdictional clai- Accompanying global warming and accelerated , the urban heat island ms in published maps and institutio- (UHI) effect has become an established urban environment phenomenon. The UHI effect nal affiliations. occurs when the temperature in urban areas is significantly higher than in suburban areas. The temperature difference between urban and suburban areas, which occurs due to an energy imbalance [1], is referred to as the UHI intensity (UHII). UHI poses a significant Copyright: © 2020 by the authors. Li- threat to urban ecosystems, human health, and urban energy systems [2,3]. censee MDPI, Basel, Switzerland. Many factors significantly affect the formation and development of UHIs. Broadly, This article is an open access article these factors are classified into two categories based on their influence on the occurrence distributed under the terms and con- and development of UHIs [4]. The first category is the characteristics of the city, such as ditions of the Creative Commons At- geographical location, topography, morphology, demography [5], cover and land tribution (CC BY) license (https:// use [6], built-up intensity, and anthropogenic heat emissions [7]. These factors contribute creativecommons.org/licenses/by/ significantly to the development of UHIs. The second category is meteorological conditions, 4.0/).

Sustainability 2021, 13, 320. https://doi.org/10.3390/su13010320 https://www.mdpi.com/journal/sustainability Sustainability 2021, 13, 320 2 of 15

such as precipitation, and temperature, cover, speed and direction, and solar radiation that affect the UHI intensity [8–11]. The daily maximum UHI intensity (UHIImax) can be regarded as a measure of UHI development. For a specific city, identifying the days with full UHII development over an extended period is required to study the mechanism of UHIs. Many studies have focused on iden- tifying the days when the UHI is fully developed [12,13], e.g., the influence of different meteorological conditions on UHI development and the methods for identifying the meteo- rological conditions conducive to the UHI development. Extensive research has established the significant effect of precipitation on UHIs; most show a negative correlation between these phenomena. Arifwidodo and Tanaka [14] reported that the weakest UHI (2 ◦C) in , Thailand, occurred in August, which was also the month with maximum precipitation. Unal [15] demonstrated that the UHII during the daytime on rainy summer days is 1 ◦C higher than on dry days, regardless of the precipitation amount in , Turkey. The cloud coverage, as well as the evaporation of rainwater, increases before and after rainfall; further, these factors are considered to be the reason for precipitation leading to UHI variations as they remove urban sensible and latent heat [4]. Previous studies have identified strong UHI when meteorological conditions were clear and calm, or during cloudless days with weak . However, the quantitative descriptions of these conditions vary according to the study. Both and wind speed factors are used to identify clear and calm days, which are considered conducive to UHI development. These factors differ according to the region and must be adjusted accordingly. Oke and Maxwell [16] selected variable thresholds with winds under 1.3 m/s and cloud cover of less than two-tenths as the appropriate meteorological conditions to determine UHI development in Montreal and Vancouver, Canada. Based on a study by Leconte [17] in Nancy, France, high daily UHIImax prefers meteorological conditions that have less than two-tenths cloud cover and wind speeds less than 9 m/s. Yang [18] suggested wind speeds of less than 2 m/s and under three-tenths cloud cover for the examination of UHIs in Nanjing, China. Yang [18] suggested that the criteria of wind speeds and cloud cover must be re-calibrated because it varies according to the characteristics of the city. Although wind factors are easily measured, cloud cover data are difficult to obtain, implying that this method cannot be applied simply and rapidly. Oke [19] proposed a new method, i.e., the weather factor, that combines the effects of wind speed and , including the cover and type. In previous studies, these meteorological variables were adopted to identify clear and calm conditions, or cloudless with weak wind conditions for UHI development [20,21]. Despite the efficiency of the proposed method, obtaining cloud data for several regions remains challenging, primarily because cloud data consisting of cover and type used to determine coefficient k are required to calculate weather factors. Yang [18] derived an indicator from the daily diagnostic equation to identify the meteorological conditions conducive to UHI development in Nanjing, China. However, further investigations are required to determine the effectiveness of this proposed indicator. Recently, to overcome this challenge, studies have begun to explore the possibility of using diurnal temperature ranges (DTRs) in suburban or rural areas as an indicator of the meteorological conditions and identify days with high daily UHIImax because this is related to the difference in cooling rates in urban and suburban areas. DTR is the difference between the highest and the lowest daily in a suburban area. Holmer [22] stated that DTR states can be adopted for weather classification. A positive correlation between the DTR and daily UHIImax was found in northwestern Europe by Theeuwes [23]. Yao et al. [24] measured the UHII of different local climate zones (LCZs) in Nanjing, finding that the suburban DTR had a strong positive correlation with UHIImax. Moreover, they determined that the DTR as an index can be employed to identify days with high daily UHIImax. Yang [18] investigated the performance of cloud cover and wind speed, DTR, and the weather factor on identifying days with full UHII development. The quantitative relationship between the DTR and UHIImax were analyzed and established, showing that the correlation coefficient between them is greater than 0.67. Sustainability 2021, 13, 320 3 of 15

◦ They proposed that high UHIImax was observed when DTR ≥ 10 C in Nanjing based on their observations. This method can be used effectively to identify the days with full UHI development. The geographical location and climate of other cities are different from those of Nanjing, resulting in significant differences in the DTR; thus, the method requires evaluation in other cities. UHI research has generally focused on the characteristics of DTR in the process of urbanization [25–27]. The quantitative analysis of DTR and UHII development has gradually increased [28]. We emphasize that reliable assessments of UHIs depend, to a large degree, on the days selected for observation and analysis. The selection of the sample days with full UHI development is crucial for the field measurement of the UHI. Methods to identify days with full UHI development have been developed; however, the methods for measuring cloud cover and wind speed, weather factor, and the daily diagnostic equation are not straightforward and not easy to use for field measurements due to the several parameters involved and the difficulty in obtaining them. DTR is easy to use because only routine observations are required. The methods to identify days with full UHI development have not been previously studied in Guangzhou, China, which has a typical subtropical climate different from Nanjing. This study aims to establish a method for selecting the typical days with full urban heat island development by investigating the correlation between the DTR and UHIImax in Guangzhou. Annual hourly observation data were used to quantitatively analyze the relationship between the DTR and UHIImax under different meteorological conditions classified according to the precipitation. The appropriate threshold for the DTR, as an index, was determined to identify days conducive to the development of UHIs in Guangzhou. The quantitative relationship between the DTR and UHIImax can help establish a method for selecting days with full UHII development for the measurement and simulation of UHI phenomena. This study and the results are also beneficial for the prediction of intense UHI events.

2. Methods 2.1. Study Area Guangzhou spans 7434.4 km2 on both sides of the Pearl River from 112◦57” to 114◦03” E and 22◦26” to 23◦56” N in south-central Guangdong. The Pearl River flows through the city. Guangzhou is the of Guangdong Province, and the central city of the Guangdong-Hong Kong-Macao Greater Bay Area and Pan-Pearl River Delta Economic Zone. Figure1 shows an overview of the study area. Guangzhou is characterized by hot and warm . The months of June to September are the hottest months of the year, with daily average temperatures ranging from 27.7 to 29.2 ◦C, daily maximum temperatures ranging from 34.5 to 36 ◦C, and a daily average relative humidity of approximately 80%. The annual number of rainy days is approximately 152, and the annual precipitation is 1696.5 mm. The UHII is higher in the summer and autumn in Guangzhou. The average UHII is 1.2 ◦C in the summer and 1.5 ◦C in the autumn [29]. The built-up area is approximately 1249.11 km2 and is located around the river. As shown in Figure1, more rural farmland, represented by large yellow areas, is situated adjacent to the estuary. The large green areas in Figure1 are mainly evergreen trees or shrubs, where the terrain is low elevation mountains. Most buildings in the city are air-conditioned in the summer. Sustainability 2021, 13, x FOR PEER REVIEW 4 of 15 Sustainability 2021, 13, 320 4 of 15

FigureFigure 1.1. StudyStudy areaarea ofof Guangzhou,Guangzhou, China.China. SatelliteSatellite imagesimages of the meteorological observation sites (Tian(Tian He He Bei Bei abbreviated abbreviated as as THB THB and and Tong Tong Fu Fu Xi Xi abbreviated abbreviated as TFX)as TFX) and and two two temperature temperature measuring meas- pointsuring points at the observationat the observation sites; location sites; location of the Internationalof the International Meteorological Meteorological Station. Station.

2.2.2.2. FieldField MeasurementMeasurement 2.2.1.2.2.1. SiteSite forfor UrbanUrban AirAir TemperatureTemperature MeasurementsMeasurements TheThe datadata onon hourlyhourly airair temperaturetemperature inin thethe suburbssuburbs andand urbanurban areasareas areare requiredrequired toto exploreexplore thethe relationshiprelationship betweenbetween thethe DTRDTR andand UHII.UHII. TheThe airair temperaturetemperature waswas obtainedobtained throughthrough fieldfield measurements in in this this study. study. Se Selectinglecting an an appropriate appropriate site site for for urban urban temper- tem- peratureature measurement measurement is highly is highly important; important; the te themperature temperature should should reflect reflect the physical the physical struc- structure,ture, surface surface properties, properties, and and thermal thermal climate climate of ofthe the city, city, as as well well as as a a high UHII.UHII. LCZLCZ schemes proposed by Stewart and Oke [7] have been widely adopted to globally study schemes proposed by Stewart and Oke [7] have been widely adopted to globally study UHIs; these schemes aim to provide an objective and standardized classification criterion UHIs; these schemes aim to provide an objective and standardized classification criterion for UHI studies. The LCZ schemes enhance the description of surface conditions in urban for UHI studies. The LCZ schemes enhance the description of surface conditions in urban and rural areas, thereby easing the process of site selection. An area spanning hundreds of and rural areas, thereby easing the process of site selection. An area spanning hundreds meters to several kilometers with uniform features in terms of the structure, land cover, of meters to several kilometers with uniform features in terms of the structure, land cover, material, and human activity can be described as an LCZ [7]. There are 10 LCZ built types material, and human activity can be described as an LCZ [7]. There are 10 LCZ built types and seven LCZ land cover types in this study area. Each LCZ is expected to present a and seven LCZ land cover types in this study area. Each LCZ is expected to present a characteristic screen-height (1–2 m above ground) temperature. The site for the urban air characteristic screen-height (1–2 m above ground) temperature. The site for the urban air temperature field experiments was selected based on the LCZ schemes in Guangzhou. temperature field experiments was selected based on the LCZ schemes in Guangzhou. According to the guidelines for using the LCZ classification system suggested by According to the guidelines for using the LCZ classification system suggested by Oke Oke [7], a LCZ is defined as an area with a minimum radius of 200–500 m, which has uniform[7], a LCZ features is defined in terms as an of area surface with cover,a minimum structure, radius material, of 200–500 and humanm, which activity. has uniform Addi- tionally,features differentin terms LCZsof surface with acover, radius structure, of 500 m material, were selected and human to study activity. the thermal Additionally, behavior ofdifferent LCZs in LCZs Nanjing, with a China radius [10 of]. 500 A m total were of se 10lected different to study LCZs the were thermal selected behavior to study of LCZs the thermalin Nanjing, behavior China in [10]. Guangzhou A total of (represented 10 different by LCZs the black were spot selected in Figure to study1). Based the thermal on the relationshipbehavior in betweenGuangzhou UHII (represented and DTR, two by the LCZ blac sitesk spot (THB in andFigure TFX) 1). with Based a radiuson the ofrelation- 500 m wereship selectedbetween forUHII meteorological and DTR, two observations LCZ sites (THB in this and study. TFX) These with twoa radius sites of were 500 selected m were becauseselected theyfor meteorological exhibited a higher observations UHII in a previousin this study. study These [30]. Figuretwo sites1 shows were the selected location be- ofcause the sitesthey (representedexhibited a higher by red UHII spots). in a One previous site (THB) study is [30]. located Figure inthe 1 shows new town the location center ofof thethe city,sites characterized(represented by by red high spots). building One densitysite (THB) and is severallocated typesin theof new buildings town center (office of buildings,the city, characterized residential buildings, by high andbuilding hotels). density This site and is classifiedseveral types as LCZ of buildings 2. The other (office site (TFX),buildings, characterized residential by buildings, mid-rise residentialand hotels). buildings, This site is belongs classified to the as LCZ 32. andThe isother located site in(TFX), the old characterized town center by of mid-rise the city. residential Table1 lists bu theildings, detailed belongs site to parameters the LCZ 3 and is located images.in the old Figure town2 showscenter theof the . imagesTable 1 oflists the the selected detailed LCZs. site parameters and satellite images. Figure 2 shows the satellite images of the selected LCZs. Sustainability 2021, 13, x FOR PEER REVIEW 5 of 15

Sustainability 2021, 13, x FOR PEER REVIEW 5 of 15

Sustainability 2021, 13, x FOR PEER REVIEW 5 of 15 Sustainability 2021, 13, x FOR PEER REVIEW 5 of 15

(a) (b)

Figure 2. Temp/RH(a )data logger at the urban sites: (a) THB;(b) (b) TFX. Sustainability 2021 13 , , 320 5 of 15 (a) (b) FigureTable 2. Temp/RH 1. Characteristics data(a) logger of the at thelocal sites: zone( b(a) ) (LCZ)THB; ( sites.b) TFX. Figure 2. Temp/RH data logger at the urban sites: (a) THB; (b) TFX. Figure 2. Temp/RH data logger at the urban sites: (a) THB; (b) TFX. TableTable 1. 1. CharacteristicsCharacteristics of of the the local local climate climate zone zone (LCZ) (LCZ)Properties sites. sites. Building Type SatelliteTable Images 1. Characteristics of the local climate zone (LCZ) sites. Table 1. CharacteristicsZone of Parameter the local climate Value zone (LCZ) sites. Indicator Range for the LCZ Type [7] PropertiesProperties Properties Building TypeBuilding Type Satellite Satellite Images Properties Building TypeBuilding Type Satellite Satellite Images Images Zone ParameterH/W 1: 2.24Value Indicator Range for the LCZH/W: Type [7]>2 ZoneZone ParameterZone Value ParameterValue Indicator Value Indicator Range for Indicator theRange LCZ Type Rangefor the[7] for LCZ the Type LCZ Type[7] [7] H/WSVF: 1: 2.24 0.48 * 1 H/W: >2 SVF: 0.2–0.4 H/W 1: 2.241 H/W : 2.24 H/W: >2 H/W: >2 BSF:H/WSVF: 0.48 30.55% : 2.24* * SVF: 0.2–0.4 BSF:H/W: 40–60% >2 SVF: 0.48 * SVF: 0.48 *SVF: 0.2–0.4 SVF: 0.2–0.4 BSF: 30.55% * BSF: 40–60% BSF:SVF:ISF: 30.55% 49.28%0.48 * BSF:* 30.55% *BSF: 40–60%SVF: ISF: 0.2–0.440–60% BSF: 40–60% ISF: 49.28% ISF: 40–60% LCZ 1: Compact high-rise BSF:ISF:HRE: 49.28% 30.55% 35 mISF: * 49.28%ISF: 40–60%BSF: HRE: 40–60% >25 ISF: m 40–60% LCZ 1: Compact high-rise HRE: 35 m HRE: >25 m LCZ 1: CompactLCZ high-rise1: Compact high-rise ISF:HRE: 49.28%35 m HRE: 35 mHRE: >25 m ISF: 40–60% HRE: >25 m LCZ 1: Compact high-rise HRE:H/W: 0.84 35 m H/W: 0.84H/W: 0.75–1.5HRE: H/W:>25 m 0.75–1.5 H/W:H/W: 0.84 0.84 H/W: 0.75–1.5H/W: 0.75–1.5 SVF: 0.57 SVF: 0.57SVF: 0.2–0.6 SVF: 0.2–0.6 SVF: 0.57 SVF: 0.2–0.6 BSF:SVF: 44.86% 0.57 BSF: 44.86%BSF: 40–70% SVF: 0.2–0.6 BSF: 40–70% BSF: 44.86% BSF: 40–70% ISF:BSF:H/W: 51.61% 44.86% 0.84 * ISF: 51.61% *ISF: 20–50%H/W: BSF: 0.75–1.540–70% ISF: 20–50% LCZ 3: Compact low-rise ISF: 51.61% * ISF: 20–50% LCZ 3: CompactLCZ low-rise3: Compact low-rise HRE:ISF:SVF: 1151.61% m0.57 * HRE: * 11 m *HRE: 3–10 mSVF: ISF: 0.2–0.620–50% HRE: 3–10 m Sustainability 2021, 13, x FOR PEER REVIEW HRE: 11 m * HRE: 3–10 m 5 of 15 LCZ1 H/W:3: Compact aspect1 H/W: ratio; low-rise aspect SVF: ratio; sky SVF: view sky factor; view factor BSF:; buildingBSF: building surfaceBSF: surface 44.86% fraction; fraction; ISF: ISF: impe imperviousrvious surface surface fraction; fraction; HRE:BSF: height HRE:40–70% ofheight of roughness 1 HRE: 11 m * HRE: 3–10 m elements; *: valueroughness H/W: deviates aspect elements; ratio; from SVF: *: the value sky parameter deviatesview factor from value; BSF: the rangebuildingparameter specified surface value fraction;range by the specif ISF: correspondingied impe byrvious the corresponding surface LCZ types.fraction; LCZ HRE: types. height of roughness elements; *: value deviates from the parameterISF: value 51.61% range specif* ied by the corresponding LCZ types.ISF: 20–50% LCZ1 3: Compact low-rise H/W: aspect ratio; SVF: sky view2.2.2. factor Experiment; BSF: building DesignHRE: surface 11 m fraction; * ISF: impervious surfaceHRE: fraction; 3–10 m HRE: height of 2.2.2. Experiment Design roughness elements; *: value deviatesThe from air thetemperatures parameter measured value atrange the THB specif andied TFX by sites the were corresponding used to represent LCZ the types. 1 H/W: aspect ratio; SVF: sky viewurban factorThe temperature; airBSF: temperatures building of Guangzhou. surfacemeasured The fraction; at hourlythe THB ISF:temperatures and impe TFX sitesrvious of werethe surfacetwo used sites to fraction;wererepresent derived theHRE: height of roughness elements; *: value2.2.2. deviates urbanfromExperiment from the temperature temperature the parameter Design of Guangzhou. observation value The experimerange hourly specifnt temperatures of theied Guangzhouby the of thecorresponding twoLCZ, sites which were was derivedLCZ con- types. fromducted the in temperatureJuly 2019 and observation is still ongoing. experime At eantch of site, the two Guangzhou fixed points LCZ, located which in was the con-core Theductedarea ofair thein temperaturesJuly site within2019 and a radius is still measured of ongoing. 100 m were At at eaequipped thech site, THB two with andfixed temp/RH TFXpoints data sites located loggers were in the(HOBO used core to represent the 2.2.2.urban ExperimentareaU23X-001) temperature of the siteinside within Design a ofmatching a Guangzhou. radius radiation of 100 m wereshield The equipped hourlyto collect with thetemperatures hourly temp/RH air datatemperature. ofloggers the (HOBOtwo Figure sites were derived U23X-001) inside a matching radiation shield to collect the hourly air temperature. Figure fromThe 1the shows air temperature temperaturesthe locations of observation the measured data loggers. experimeat The the temp/RH THBnt anddataof the logger TFX Guangzhou usedsites was were the HOBOusedLCZ, towhich represent was con-the 1U23X-001, shows the manufactured locations of the by dataOnset loggers. (Bourne, The MA, temp/RH USA), withdata alogger measuring used wasrange the of HOBO0 to 50 urbanducted U23X-001,°Ctemperature inand July uncertainty manufactured 2019 of andof Guangzhou. ±0.2 byis °C. Onsetstill For ongoing.(Bourne, instrume The MA,hourlynt Atsafety, USA), ea temperatures chthe with site,temp/RH a measuring two data fixed of rangelogger the points oftwo was 0 to in-sites50located were in derived the core fromarea ofthe°Cstalled theand temperature onsiteuncertainty street within lampposts of a observation±0.2 radius or°C. poles For of at instrume100 a experimeheight m werent of safety, 2.1–2.5 ntequipped theof m fromtemp/RHthe theGuangzhou with ground. data temp/RH logger To avoid LCZ,was data thein- which loggers was (HOBO con- stalledinfluence on ofstreet any lampposts artificial heator poles sources at a heightand ensure of 2.1–2.5 adequate m from ventilation, the ground. the To point avoid was the ductedU23X-001)influence in July inside of 2019 any a artificial andmatching is heat still sourcesradiation ongoing. and ensureshield At ea adequate chto site,collect ventilation,two the fixed hourly the points point air was temperature.located in the Figure core placed far from(a vehicles) and air conditioners, and at a distance(b) of over 3 m from walls. area1 shows ofplacedThe the loggersthe site far locations fromwithin were vehicles installed a ofradius and the in airthe data of conditioners, open 100 loggers. area m were which and The equipped isat takena temp/RH distance as the with of area overdata temp/RH never 3 mlogger fromcovered walls. dataused by loggers was the (HOBO HOBO U23X-001)U23X-001,Thetree shadowloggers inside manufactured wereall a day matchinginstalled long. Figure byin the radiationOnset 2open shows area(Bourne, one shield which data loggeris MA,to taken collect installedUSA), as the the area withat hourlyTHB never a and measuring covered airTFX. temperature. The by range of Figure 0 to 50 Figuretree 2. shadowTemp/RH all day data long. logger Figure at2 shows the urban one data sites: logger (a) installed THB; (b at) TFX.THB and TFX. The 1Figure°C shows andaverage 2. theuncertaintyTemp/RH locationsvalue of data the of of temperaturelogger ±0.2 the data °C.at the readingsFor loggers. urban instrume obtained sites: The (nt a temp/RHat) THB; safety,the two (b )measuring the dataTFX. temp/RH logger points used wasdata was logger the HOBOwas in- averagetaken as valuethe air oftemperature the temperature value ofreadings the site. obtained at the two measuring points was U23X-001,stalled2.2.2.taken Experiment on as manufacturedstreet the air lampposts temperature Design by value orOnset poles of the (Bourne, atsite. a height MA, of USA), 2.1–2.5 with m from a measuring the ground. range To ofavoid 0 to 50the Table 1. Characteristics of the local climate zone (LCZ) sites. °C and2.2.3. uncertainty Air Temperature of ±0.2in Suburban °C. For Area instrume nt safety, the temp/RH data logger was in- influence2.2.3.The Air airof Temperatureany temperatures artificial in Suburban heat measured sourcesArea at and the ensure THB and adequate TFX sites ventilation, were used the to point represent was stalled onWe street obtained lampposts the hourly or suburban poles airat temperaturesa height of in 2.1–2.5 Guangzhou m from from the the National ground. To avoid the placedthe urban farWe from temperatureobtained vehicles the hourly ofand suburban Guangzhou. air conditioners, air temperatures The hourly andinProperties Guangzhou at temperaturesa distance from the ofNational ofover the 3 twom from sites walls. were Building Type influence SatelliteMeteorological Imagesof any artificialStation (NMS) heat No. sources59287. Hourly and temperature ensure adequatedata from the ventilation, NMS can be the point was Thederived loggersMeteorologicalaccessed from viawere the the Station temperatureChinainstalledZone (NMS)Meteorological Parameter in No. the observation 59287. open DataValue Hourly Service area temperature experiment whichCenter Indicator (CMDC, is data taken of from theRangehttp://data.cma.cn). theas Guangzhou NMSthe for area can the be neverLCZ LCZ, Type covered which [7] was by placedtreeconducted shadowaccessed far from invia all July the vehiclesday China 2019 long. Meteorological and Figure isair still conditioners, 2 Datashows ongoing. Service one Center At anddata each (CMDC,at logger a site, distance http://data.cma.cn). twoinstalled fixed of over at points THB 3 m and located from TFX. walls. in The the 1 Theaveragecore loggers area value of were the of site installedthe within temperatureH/W in a the radius: 2.24 open readings of area 100 mwhich obtained were is equipped taken at the as withtwoH/W:the areameasuring temp/RH >2 never datacoveredpoints loggers wasby treetaken(HOBO shadow as U23X-001)the allair daytemperature insidelong.SVF: Figure a matching value 0.48 2 shows *of the radiation site.one data shield logger to collect installedSVF: the 0.2–0.4 hourlyat THB airand temperature. TFX. The averageFigure 1value shows of the the locations temperatureBSF: 30.55% of the readings data* loggers. obtained The temp/RHat theBSF: two data40–60% measuring logger used points was was the taken2.2.3.HOBO asAir the U23X-001, Temperature air temperature manufactured inISF: Suburban value 49.28% of by Areathe Onset site. (Bourne, MA, USA),ISF: 40–60% with a measuring range ◦ ± ◦ LCZ 1: Compact high-rise of 0 toWe 50 obtainedC and uncertaintythe hourlyHRE: suburban of35 m0.2 C.air For temperatures instrument in safety, GuangzhouHRE: the >25 temp/RH m from the data National logger 2.2.3.Meteorologicalwas Air installed Temperature on Station street in (NMS) lampposts Suburban No. 59287. orArea poles Hourly at a height temperature of 2.1–2.5 data m from from the the NMS ground. can be To accessedavoidWe the obtained via influence the Chinathe of hourly any MeteorologicalH/W: artificial suburban 0.84 heat air Data sources temperatures Service and ensureCenter in GuangzhouH/W: adequate (CMDC, 0.75–1.5 ventilation,http://data.cma.cn). from the National the point Meteorologicalwas placed far Station from vehicles (NMS)SVF: andNo. 0.57 air59287. conditioners, Hourly temperature and at a distanceSVF: data 0.2–0.6 from of over the 3 NMS m from can walls. be The loggers were installed in the open area which is taken as the area never covered by accessed via the China MeteorologicalBSF: 44.86% Data Service Center BSF:(CMDC, 40–70% http://data.cma.cn). tree shadow all day long. Figure2 shows one data logger installed at THB and TFX. The ISF: 51.61% * ISF: 20–50% LCZ 3: Compact low-rise average value of the temperature readings obtained at the two measuring points was taken HRE: 11 m * HRE: 3–10 m as the air temperature value of the site. 1 H/W: aspect ratio; SVF: sky view factor; BSF: building surface fraction; ISF: fraction; HRE: height of roughness elements; *: value2.2.3. deviates Air from Temperature the parameter in Suburban value range Area specified by the corresponding LCZ types. We obtained the hourly suburban air temperatures in Guangzhou from the National 2.2.2.Meteorological Experiment Station Design (NMS) No. 59287. Hourly temperature data from the NMS can be accessedThe air via temperatures the China Meteorological measured at Datathe THB Service and CenterTFX sites (CMDC, were used http://data.cma.cn to represent the). urbanThese temperature data were used of Guangzhou. to classify the The meteorological hourly temperatures conditions of in the Guangzhou two sites were and calculatederived fromthe UHII the temperature and suburban observation DTR. The stationexperime is standard of the Guangzhou meteorological LCZ, station which thatwas sends con- and receives data internationally. The station, represented by black pentagram in Figure1, ducted in July 2019 and is still ongoing. At each site, two fixed points located in the core is located on the northeast of the city. These data include meteorological variables, such area of the site within a radius of 100 m were equipped with temp/RH data loggers (HOBO U23X-001) inside a matching radiation shield to collect the hourly air temperature. Figure 1 shows the locations of the data loggers. The temp/RH data logger used was the HOBO U23X-001, manufactured by Onset (Bourne, MA, USA), with a measuring range of 0 to 50 °C and uncertainty of ±0.2 °C. For instrument safety, the temp/RH data logger was in- stalled on street lampposts or poles at a height of 2.1–2.5 m from the ground. To avoid the influence of any artificial heat sources and ensure adequate ventilation, the point was placed far from vehicles and air conditioners, and at a distance of over 3 m from walls. The loggers were installed in the open area which is taken as the area never covered by tree shadow all day long. Figure 2 shows one data logger installed at THB and TFX. The average value of the temperature readings obtained at the two measuring points was taken as the air temperature value of the site.

2.2.3. Air Temperature in Suburban Area We obtained the hourly suburban air temperatures in Guangzhou from the National Meteorological Station (NMS) No. 59287. Hourly temperature data from the NMS can be accessed via the China Meteorological Data Service Center (CMDC, http://data.cma.cn). Sustainability 2021, 13, 320 6 of 15

as solar radiation, air temperature, humidity, wind speed and direction, precipitation, and . The minimum and maximum temperature, humidity and wind direction observed in this station appear in daily weather forecast and represent the weather conditions in Guangzhou.

2.3. Calculation of UHII and DTR 2.3.1. UHII The difference in the hourly air temperature between that measured in urban areas and that observed in suburban areas was defined as the UHII (UHII = Turban − Tsuburban). The maximum UHII in a day is UHIImax, which represents the development level of the UHI. To capture the complete diurnal cycle of the climatic processes (nocturnal cooling after daytime warming), a 24 h period from 08:00 a.m. to 07:00 a.m. was defined as a day to calculate the UHIImax and DTR. The average daily UHIImax at each site for the study duration was defined as UHIImax. A day was regarded as a full UHI development day when UHIImax > UHIImax.

2.3.2. DTR in Suburban Areas The suburban DTR is considered a measure of the cooling potential at night in subur- ban areas. This can be quantified according to the maximum hourly temperature minus the hourly minimum temperature in a day (DTR = Tmax − Tmin). DTR represents the comprehensive effect of multiple meteorological variables in suburban areas and can be used to estimate the change in the regional thermal environment. During sunny and cloudless conditions, the open space in a suburban area and strong solar radiation result in a relatively high daily maximum temperature. At night, due to the large sky view factor and enhanced long-wave , the daily minimum temperature reaches a relatively low minimum value. The UHII is generally high under such meteorological conditions because the cooling rate in a suburban area is faster than that in an , especially after sunset [1].

2.4. Meteorological Conditions during Experiments The hourly data for a year from 1 August 2019 to 31 July 2020, at the sites observed by the NMS in the suburban area and those measured by temp/RH data loggers in the urban area were used in this study. Excluding missing data for 26 days, data from a total of 337 days were used in this study. To analyze the influence of different meteoro- logical conditions on the UHI in Guangzhou, the influence of rainfall on the UHI must be examined. Three categories of meteorological conditions were classified by the NMS (59,287) precipitation data. Days with precipitation events recording a minimum daily- accumulated (≥0.1 mm) precipitation were defined as precipitation days (PDs); 114 days were analyzed. Days without rainfall following the PDs were defined as the days following precipitation (FPDs). Our study period included 29 FPDs. Other days without precipitation were defined as non-precipitation days (NPDs); 194 NPDs were analyzed. Table2 lists the number of PDs, FPDs, and NPDs used for analysis and the monthly meteorological conditions, including air temperature and precipitation from the NMS in the suburban area. The monthly average temperature and precipitation were consistent with the climatic characteristics of Guangzhou, whereby higher temperatures occur in the summer and autumn and precipitation is more likely to occur in the spring and summer. During the study period, the mean temperatures from May to September all exceeded 27.0 ◦C. The most precipitation occurred in May, and the least in November. Sustainability 2021, 13, x FOR PEER REVIEW 7 of 15

Table 2. Number days in three meteorological categories and monthly mean air temperature and precipitation from Na- tional Meteorological Station (NMS) in Guangzhou.

Month Jan Feb Mar Apr May Jun July Aug Sep Oct Nov Dec Total PDs 1 4 10 15 11 16 16 9 17 9 6 0 1 114 NPDs 24 16 10 10 7 5 17 6 17 23 30 29 194 FPDs 3 3 5 2 1 3 2 3 4 2 0 1 29 precipitation 22.4 124.9 150.2 83.5 511.1 250.8 65.5 452.5 104.6 39.5 0 5.6 1811 Sustainability 2021, 13, 320 7 of 15 Mean T (°C) 16 16.8 20.2 20.7 27.3 28.5 30.3 28.5 27.0 24.5 19.8 16.5 - 1 PDs: Precipitation days; FPDs: days following a precipitation day; NPDs: non-precipitation days. Table 2. Number days in three meteorological categories and monthly mean air temperature and precipitation from National Meteorological Station (NMS)3. Results in Guangzhou. 3.1. Characteristics of Daily UHIImax Month Jan Feb Mar Apr May Jun July Aug Sep Oct Nov Dec Total Research on the seasonal characteristics of the UHI is necessary to alleviate the UHI PDs 1 4 10 15 11 16 16 9 17 9 6 0 1 114 effect. Figure 3 depicts the monthly statistics and analysis of the daily UHIImax for the NPDs 24 16study 10duration 10 at the 7THB and 5 TFX 17sites. The 6 UHII 17 was 23 4.19 °C 30 for 337 29 days at 194 the FPDs 3 3 5 2 1 3 2 3 4max 2 0 1 29 precipitation 22.4 124.9THB 150.2site, as indicated 83.5 511.1 by the250.8 red line65.5 in Figure452.5 2a.104.6 At the THB 39.5 site, 0the maximum 5.6 1811of the Mean T (◦C) 16 16.8monthly 20.2 averages 20.7 of 27.3the daily 28.5 UHII 30.3max was 28.5 6.0 °C 27.0in December 24.5 while 19.8 the minimum 16.5 -was 1 2.3 °C in May; the monthly average daily UHIImax values for September, October, No- PDs: Precipitation days; FPDs: days following a precipitation day; NPDs: non-precipitation days. vember, and December were significantly higher than the UHIImax. Furthermore, more ＞ than3. Results 20 days and over 75% of every month had a daily characteristic of UHIImax UHII during these months. However, during the hotter months of May, June, July, 3.1. Characteristicsmax of Daily UHIImax and August, the monthly average daily UHII values were lower than the UHII , Research on the seasonal characteristics ofmax the UHI is necessary to alleviate the UHImax during which no more than 25% of the days exhibited UHII ＞ UHII . The charac- effect. Figure3 depicts the monthly statistics and analysis ofmax the daily maxUHII for the teristics of the daily UHII distribution at the TFX site were consistent withmax the THB study duration at the THBmax and TFX sites. The UHII was 4.19 ◦C for 337 days at the THB site. max site, as indicated by the red line in Figure2a. At the THB site, the maximum of the monthly The UHIImax of TFX was 4.62 °C for◦ 337 days; Figure 2b shows the monthly distri-◦ averages of the daily UHIImax was 6.0 C in December while the minimum was 2.3 C in bution of UHIImax. At the TFX site, the maximum of the monthly averages of the daily May; the monthly average daily UHIImax values for September, October, November, and UHIImax was 6.45 °C in December while the minimum was 2.57 in May. The average daily December were significantly higher than the UHIImax. Furthermore, more than 20 days UHIIs of the two urban areas were lower in the spring and summer than in the autumn and over 75% of every month had a daily characteristic of UHIImax > UHIImax during andthese winter. months. However, However, due during to the thesmall hotter number months of research of May, samples June, July, and and the August, number theof years analyzed, this seasonal UHI characteristic trend in Guangzhou requires further re- monthly average daily UHIImax values were lower than the UHIImax, during which no search. more than 25% of the days exhibited UHIImax > UHIImax. The characteristics of the daily UHIImax distribution at the TFX site were consistent with the THB site.

(a) (b)

FigureFigure 3. 3. BoxBox plots plots for for the the monthly monthly distribution distribution of of daily daily UHIImax in Guangzhou: ( a) THB; ( b) TFX.

◦ AnalyzingThe UHIImax the ofoccurrence TFX was time 4.62 frequencyC for 337 of days; the daily Figure UHII2b showsmax is thenecessary monthly to reveal distri- thebution characteristics of UHIImax of. Atthe the UHI. TFX Figure site, the4 shows maximum the statistics of the monthly of the occurrence averages oftime the of daily the ◦ dailyUHII maxUHIIwasmax 6.45 at THBC in and December TFX in whilea 24-hthe period. minimum We divided was 2.57 the in study May. Theperiod average into three daily categoriesUHIIs of the to twointerpret urban the areas occurrence were lower time in frequency the spring results. and summer The maximum than in the frequency autumn and of winter. However, due to the small number of research samples and the number of years analyzed, this seasonal UHI characteristic trend in Guangzhou requires further research. Analyzing the occurrence time frequency of the daily UHIImax is necessary to reveal the characteristics of the UHI. Figure4 shows the statistics of the occurrence time of the daily UHIImax at THB and TFX in a 24-h period. We divided the study period into three categories to interpret the occurrence time frequency results. The maximum frequency of NPDs occurred at 23:00 at TFX and 20:00 at THB. PDs occurred at 23:00 at both TFX and THB. The daily UHIImax generally occurred from 20:00 to 2:00 during the night for both stations, indicating that the UHI prefers to develop at night and the intensity reaches its maximum at Sustainability 2021, 13, x FOR PEER REVIEW 8 of 15

NPDs occurred at 23:00 at TFX and 20:00 at THB. PDs occurred at 23:00 at both TFX and THB. The daily UHIImax generally occurred from 20:00 to 2:00 during the night for both stations, indicating that the UHI prefers to develop at night and the intensity reaches its Sustainability 2021, 13, 320 8 of 15 maximum at night due to the differences in cooling rates caused by the thermal character- istics between urban and suburban areas [14]. This observation is consistent with the find- ingsnight of dueseveral to the previous differences studies incooling [18]. After rates sunset, caused the by air the layer thermal structure characteristics is stable, and between the coolingurban andrate suburbanis faster because areas [14 of]. the This increa observationsed long-wave is consistent radiation with thein the findings suburban of several area dueprevious to large studies open [spaces.18]. After The sunset, urban the area air retains layer structure significant is stable,heat during and the the cooling day due rate to is slowfaster heat because dissipation, of the increased resulting long-wavefrom the geom radiationetric shape in the of suburban the urban area block. due toThis large results open inspaces. a high Thetemperature urban area in retainsurban areas significant and aheat low duringtemperature the day in duesuburban to slow areas, heat increasing dissipation, theresulting UHII. from the geometric shape of the urban block. This results in a high temperature in urban areas and a low temperature in suburban areas, increasing the UHII.

(a) (b)

FigureFigure 4. 4. FrequencyFrequency of of daily daily UHII UHIImaxmax observedobserved in in a a 24-h 24-h period period for for 337 337 days: days: ( (aa)) THB; THB; ( (bb)) TFX. TFX.

3.2.3.2. Influence Influence of of Precipitation Precipitation on on UHII UHIImaxmax GuangzhouGuangzhou experiences experiences approximately approximately 114 114 days days of of precipitation precipitation per per year. year. Therefore, Therefore, examiningexamining the the influence influence of ofprecipitation precipitation on the on thedevelopment development of the of UHI the is UHI necessary. is necessary. Fig- ureFigure 5 depicts5 depicts the theeffect effect of ofprecipitation precipitation on on the the UHII UHII in in Guangzhou. Guangzhou. At At the the THB THB site, site, the the ◦ ◦ averageaverage daily daily UHIIUHIImaxmax waswas 3.1 3.1 °CC during during the the PDs, PDs, which which was was approximately approximately 1.8 1.8 °CC lower lower thanthan the the value value during during the the NPDs. NPDs. At At the the TFX TFX site, site, the the average average daily daily UHIIUHIImaxmax decreaseddecreased by by 1.81.8 °C◦C due due to precipitation.precipitation. BothBoth sites sites recorded recorded an an approximately approximately 1.8 ◦1.8C lower°C lower average average daily dailyUHII UHIImax duringmax during PDs thanPDs than during during NPDs. NPDs. This This indicates indicates that that precipitation precipitation has has a negative a neg- Sustainability 2021, 13, x FOR PEER REVIEWativeeffect effect on UHI on development.UHI development. For the For FPDs, the FPDs, the average the average daily UHII dailymax UHIIwas higher was than higher9 of that15 max thanfor thethat PDs; for however,the PDs; however, a lower average a lower daily averageUHII dailymax than UHII thatmax during than that NPDs during proved NPDs that provedFPDs arethat not FPDs conducive are not toconducive UHI development. to UHI development.

Figure 5. Box plots of the daily UHIImax during PDs and NPDs at the study sites. Figure 5. Box plots of the daily UHIImax during PDs and NPDs at the study sites. Figure6 depicts the monthly precipitation and box plots of the daily UHIImax for threeFigure categories 6 depicts of meteorological the monthly precipitation conditions classified and box by plots precipitation. of the daily There UHII is amax negative for threecorrelation categories between of meteorological the daily UHII conditionsmax and classified precipitation. by precipitation. The monthly There average is a nega- daily tiveUHII correlationmax was between lower during the daily PDs UHII thanmax NPDs, and exceptprecipitation. when more The monthly than half average of the monthsdaily UHIImax was lower during PDs than NPDs, except when more than half of the months were PDs and precipitation was heavy. Precipitation had the least influence on the UHI in the autumn and winter, and a significant impact in the spring and summer. A negative

correlation exists between precipitation and daily UHIImax because precipitation weak- ens UHI development.

(a) (b)

Figure 6. Box plots of the daily UHIImax for PDs and NPDs for different months at (a) THB and (b) TFX.

3.3. Relationship between DTR and UHI To reveal the effect of the DTR on UHI development, the seasonal features and dis- tribution of the DTR should be examined first, especially on the PDs and NPDs. Figure 7a shows the DTR statistics distribution for the 337-day study period. During this time, the minimum recorded DTR value in the was 1.3 °C, the maximum value was 16.0 °C, and the average value was 8.38 °C. A significant difference existed in the distribution of the monthly average DTR ranges from 5.75 to 11.25 °C. The DTR in autumn and winter was higher than that in the spring and summer, which may be due to increased precipi- tation during the spring and summer. Figure 7b shows the DTR statistics distribution of the PDs and NPDs. The difference in the monthly average DTR between the PDs and NPDs was significant where the DTR showed a negative correlation with the temperature. NPDs were used to analyze the relationship between the DTR and UHI, owing to the neg- ative effect of precipitation on UHI development. Sustainability 2021, 13, x FOR PEER REVIEW 9 of 15

Figure 5. Box plots of the daily UHIImax during PDs and NPDs at the study sites.

Figure 6 depicts the monthly precipitation and box plots of the daily UHIImax for three categories of meteorological conditions classified by precipitation. There is a nega- Sustainability 2021, 13, 320 9 of 15 tive correlation between the daily UHIImax and precipitation. The monthly average daily UHIImax was lower during PDs than NPDs, except when more than half of the months were PDs and precipitation was heavy. Precipitation had the least influence on the UHI were PDs and precipitation was heavy. Precipitation had the least influence on the UHI in in the autumn and winter, and a significant impact in the spring and summer. A negative the autumn and winter, and a significant impact in the spring and summer. A negative correlation exists between precipitation and daily UHII because precipitation weak- correlation exists between precipitation and daily UHII maxbecause precipitation weakens ens UHI development. max UHI development.

(a) (b)

FigureFigure 6 6.. BoxBox plots plots of of the the daily daily UHIImax forfor PDsPDs andand NPDsNPDs forfor differentdifferent monthsmonths atat ((aa)) THBTHB andand ((bb)) TFX.TFX.

3.3. Relationship between DTR and UHI To revealreveal thethe effecteffect ofof thethe DTR DTR on on UHI UHI development, development, the the seasonal seasonal features features and and distri- dis- tributionbution of of the the DTR DTR should should be be examined examined first, first, especially especially on onthe the PDsPDs andand NPDs.NPDs. Figure7 7aa shows the DTR statistics distribution for the 337-day337-day study period.period. During this time, the minimum recordedrecorded DTRDTR valuevalue in in the the suburbs suburbs was was 1.3 1.3◦C, °C, the the maximum maximum value value was was 16.0 16.0◦C, °C,and and the averagethe average value value was was 8.38 ◦8.38C. A °C. significant A significant difference difference existed existed in the in distribution the distribution of the ofmonthly the monthly average average DTR rangesDTR ranges from 5.75from to 5.75 11.25 to 11.25◦C. The °C. DTR The DTR in autumn in autumn and winterand winter was washigher higher than than that inthat the in spring the spring and summer,and summer, which which may may be due be todue increased to increased precipitation precipi- tationduring during the spring the spring and summer. and summer. Figure7 Figureb shows 7b the shows DTR the statistics DTR statistics distribution distribution of the PDs of theand PDs NPDs. and The NPDs. difference The difference in the monthly in the averagemonthly DTR average between DTR the between PDs and the NPDs PDs was and Sustainability 2021, 13, x FOR PEER REVIEWNPDssignificant was wheresignificant the DTR where showed the DTR a negative showed correlation a negative withcorrelation the temperature. with the temperature. NPDs10 wereof 15

NPDsused to were analyze used the to relationshipanalyze the relationship between the between DTR and the UHI, DTR owing and toUHI, the owing negative to the effect neg- of precipitationative effect of onprecipitation UHI development. on UHI development.

(a) (b)

Figure 7. Box plots by month for DTR: ( a) 337 days; ( b) PDs and NPDs.

FigureFigure 8 depicts the the distribution distribution of of the the UHIIUHIImaxmax duringduring NPDs NPDs and and the therelation relation to the to DTRthe DTR at THB at THBand TFX. and TFX.In the In figure, the figure, r and rP, and which P, whichrepresent represent Pearson’s Pearson’s correlation correlation coeffi- cientcoefficient and statistical and statistical significance, significance, respectively, respectively, can be can used be used to measure to measure the intensity the intensity of the of linearthe linear association association between between the the UHII andmax DTR. and The DTR. correlation The correlation criteria ofcriteria the absolute of the absolute value of value of r were very weak (0.00–0.19), weak (0.20–0.39), moderate (0.40–0.59), strong (0.60– 0.79), and very strong (0.80–1.0) [31]. In general, the correlation analysis was considered statistically significant when p < 0.05. The daily UHIImax was positively correlated with the DTR at both the THB and TFX sites, which is consistent with previous findings [18,23,32]. The absolute value of r represents correlation coefficients that exceeded 0.7 at both sites, indicating strong correlations between the DTR and daily UHIImax in Guang- zhou. Previous study in Nanjing showed that the daily UHIImax is positively correlated with DTR, where the correlation coefficient is 0.67 and p < 0.005 [31]. The correlation coef- ficient in Guangzhou is higher than that in Nanjing. Therefore, the DTR can be used as an index to identify days with high daily UHIImax in Guangzhou.

(a) (b)

Figure 8. Daily UHIImax against DTR: (a) THB; (b) TFX.

3.4. Thresholds for Identifying Days with High Daily UHIImax

A significant correlation between the DTR and UHIImax indicates that the DTR can be used to identify days with high daily UHIImax in Guangzhou. Figure 9 shows the rela- tionship between the DTR and UHIImax for 194 NPDs. The UHIImax for 194 NPDs was Sustainability 2021, 13, x FOR PEER REVIEW 10 of 15

(a) (b) Figure 7. Box plots by month for DTR: (a) 337 days; (b) PDs and NPDs.

Figure 8 depicts the distribution of the UHIImax during NPDs and the relation to the DTR at THB and TFX. In the figure, r and P, which represent Pearson’s correlation coeffi- cient and statistical significance, respectively, can be used to measure the intensity of the Sustainability 2021, 13, 320 10 of 15 linear association between the UHIImax and DTR. The correlation criteria of the absolute value of r were very weak (0.00–0.19), weak (0.20–0.39), moderate (0.40–0.59), strong (0.60– 0.79), and very strong (0.80–1.0) [31]. In general, the correlation analysis was considered r were very weak (0.00–0.19), weak (0.20–0.39), moderate (0.40–0.59), strong (0.60–0.79), and statistically significant when p < 0.05. The daily UHII was positively correlated with very strong (0.80–1.0) [31]. In general, the correlation analysismax was considered statistically the DTR at both the THB and TFX sites, which is consistent with previous findings significant when p < 0.05. The daily UHIImax was positively correlated with the DTR at both [18,23,32]. The absolute value of r represents correlation coefficients that exceeded 0.7 at the THB and TFX sites, which is consistent with previous findings [18,23,32]. The absolute both sites, indicating strong correlations between the DTR and daily UHII in Guang- value of r represents correlation coefficients that exceeded 0.7 at both sites,max indicating zhou. Previous study in Nanjing showed that the daily UHIImax is positively correlated strong correlations between the DTR and daily UHIImax in Guangzhou. Previous study with DTR, where the correlation coefficient is 0.67 and p < 0.005 [31]. The correlation coef- in Nanjing showed that the daily UHIImax is positively correlated with DTR, where the correlationficient in Guangzhou coefficient isis 0.67higher and thanp < 0.005that in [31 Nanjing.]. The correlation Therefore, coefficient the DTR can in Guangzhou be used as an is higherindex to than identify that in days Nanjing. with high Therefore, daily UHII the DTRmax in can Guangzhou. be used as an index to identify days with high daily UHIImax in Guangzhou.

(a) (b)

FigureFigure 8.8. DailyDaily UHIIUHIImaxmaxagainst against DTR: DTR: ( a()a) THB; THB; ( b(b)) TFX. TFX.

3.4.3.4. ThresholdsThresholds forfor IdentifyingIdentifying DaysDays withwith HighHigh DailyDailyUHII UHIImax

AA significantsignificant correlationcorrelation betweenbetween thethe DTRDTR andand UHIIUHIImaxmax indicatesindicates that that the the DTR DTR can can bebe usedused to identify identify days days with with high high daily daily UHIIUHIImaxmax in inGuangzhou. Guangzhou. Figure Figure 9 shows9 shows the therela- ̅̅̅̅̅̅̅ relationshiptionship between between the the DTR DTR and and UHIIUHIImaxmax forfor 194 194 NPDs. NPDs. The The UHIImax for 194194 NPDsNPDs waswas 4.78 ◦C and 5.22 ◦C for the THB and TFX sites, respectively. The same trend can be observed at both sites; the average daily UHIImax increased, and the number of days with a daily UHIImax > UHIImax decreased with an increasing DTR. At the THB site, 183 days were ◦ ◦ selected when the DTR was ≥5 C, and the average daily UHIImax was 4.93 C. Only 48.6% of the days were characterized by UHIImax > UHIImax. At the TFX site, the average daily ◦ UHIImax was 5.37 C and 48.6% of the days recorded UHIImax > UHIImax when the DTR was ≥5 ◦C. When the DTR increased to 12 ◦C, only 44 days were recorded. However, the ratios of the number of days with a daily UHIImax > UHIImax to the number of the days when the DTR was ≥12 ◦C were 84.1% and 81.8% for the THB and TFX sites, respectively. With the increase in the DTR, fewer sample days were selected, resulting in a decrease in the sample size and an increase in the identification rate. The meteorological conditions conducive to UHI development can be identified by selecting an appropriate DTR to ensure a sufficient sample size and identification rate. In a previous study [18], a sample size higher than 30% and an identification rate of more than 70% were defined as the criteria to determine the DTR threshold value in Nanjing. In this study, when the DTR was 10 ◦C, the sample size and identification rate at the THB and TFX sites were 48% and 77%, and 33.5% and 74%, respectively. A DTR of 10 ◦C can, thus, be regarded as the threshold to identify days with high daily UHIImax because both sites meet the criteria proposed by Yang [18]. Days with DTR ≥10 ◦C can be considered conducive to UHI development in Guangzhou. However, the regional climate and geographical features of a city influence the DTR in a significant manner. Hence, a more extended period of data should be collected Sustainability 2021, 13, x FOR PEER REVIEW 11 of 15

4.78 °C and 5.22 °C for the THB and TFX sites, respectively. The same trend can be ob- served at both sites; the average daily UHIImax increased, and the number of days with a daily UHIImax ＞ UHIImax decreased with an increasing DTR. At the THB site, 183 days were selected when the DTR was ≥5 °C, and the average daily UHIImax was 4.93 °C. Only 48.6% of the days were characterized by UHIImax ＞ UHIImax. At the TFX site, the average daily UHIImax was 5.37 °C and 48.6% of the days recorded UHIImax ＞ UHIImax when the DTR was ≥5 °C. When the DTR increased to 12 °C, only 44 days were recorded. However, Sustainability 2021, 13the, 320 ratios of the number of days with a daily UHIImax ＞ UHIImax to the number of the 11 of 15 days when the DTR was ≥12 °C were 84.1% and 81.8% for the THB and TFX sites, respec- tively. and analyzed to verify UHI development in Guangzhou; further, more cities should be examined to fully evaluate the effectiveness and threshold.

(a)

(b)

Figure 9. SelectionFigure of appropriate 9. Selection DTR of appropriate threshold values DTR threshold to identify values the days to identify with full the UHI days devel- with full UHI develop- opment in Guangzhou:ment in ( Guangzhou:a) THB; (b) TFX. (a) THB; Sample (b) TFX.size (%), Sample represented size (%), representedby the black by solid the line, black is solid line, is defined defined as the number of days selected by the DTR threshold to the total number of NPDs (194 as the number of days selected by the DTR threshold to the total number of NPDs (194 days in this days in this study). Identification rate (%), represented by the red solid line, is defined as the num- study). Identification rate (%), represented by the red solid line, is defined as the number of days ber of days with a daily UHII ＞ UHII (4.78 °C at THB and 5.22 °C at TFX; represented by max > max ◦ ◦ the red dotted line)with to a the daily numberUHIImax of daysUHII selectedmax (4.78 by theC DTR at THB threshold. and 5.22 C at TFX; represented by the red dotted line) to the number of days selected by the DTR threshold.

4. Discussion Meteorological conditions impact the UHI in a significant manner. Therefore, selecting the days with a high daily UHIImax from different meteorological conditions is crucial to assess and understand the UHI. In this study, the DTR for 337 days was used as an index to select days with a high daily UHIImax in Guangzhou. The geography, topography, anthropogenic heat, and regional climate of a city have complex effects on urban and suburban thermal behavior. In addition, the air temperature is also affected by the soil type, vegetation type, and coverage. Previous studies have discussed the various characteristics of the UHI. For example, the largest UHII values were found in summer in Istanbul [15], while a strong UHI was observed in both the summer Sustainability 2021, 13, 320 12 of 15

and winter seasons in northern west Siberian cities [33] and in the winter in Beijing [34]. The monthly distribution of the daily UHIImax in the two urban areas of Guangzhou was studied for a one-year study duration; the results revealed that the UHII in the autumn and winter is more pronounced than that in the spring and summer, which is consistent with Jiang [35]. The characteristics of the UHI during autumn in Guangzhou may be related to the meteorological conditions. The meteorological conditions are stable, with more sunny days and less cloud cover, which are conducive to UHI development. However, further research with an extended period is required to establish the seasonal characteristics of the UHI in Guangzhou. Guangzhou experiences heavy precipitation, such that the average daily UHIImax increased by 1.8 ◦C during NPDs as compared with PDs at both sites. This indicates that precipitation negatively affects UHI development, which is consistent with the researches in Bangkok [14] and Istanbul [15]. During FPDs, the study sites recorded a lower average daily UHIImax than that during NPDs, indicating that FPDs are not conducive to UHI development. Future research should focus on the analysis of more data including over a more extended period to verify these results. For practical applications, this method not only can be used for selecting the typical meteorological days for UHI survey and simulation, but also can be used to forecast an early-warning of intense UHI events based on weather forecasts, which can provide the diurnal maximum and minimum temperature. For the study period, the difference in the monthly average distribution of the suburban DTR was moderate in Guangzhou, ranging from 5.75 to 11.25 ◦C. The DTR during NPDs was higher than that during PDs and FPDs, which was similar to the daily UHIImax. In other words, more precipitation days will be eliminated with an increasing DTR. Based on these observations, the relationships between ◦ the DTR and daily UHIImax during NPDs were quantitatively evaluated. Here, 10 C can be regarded as the appropriate DTR threshold for NPDs to identify typical meteorological days with high daily UHIImax in Guangzhou. Figure 10 shows the performance of a DTR threshold of ≥10 ◦C for identifying days with a high UHIImax during PDs and FPDs. Here, a stricter UHIImax for 194 NPDs, 4.78 and 5.22 ◦C for the THB and TFX sites, respectively, was used to calculate the identification rate during PDs and FPDs. When DTR was ≥10 ◦C, only 23 days were selected as sample days, yielding an identification rate of 50% and 69% for the THB and TFX sites, respectively. Al- though the identification rate was less than 70% at THB, only 23 sample days were selected, indicating that most of the PDs and FPDs were excluded. For practical applications, we suggest that a threshold of 10 ◦C be used directly for all samples to identify days with full UHI development when focusing only on days with high UHIImax and the precipitation data is difficult to obtain. Otherwise, a higher accuracy will be observed when identifying days with full UHI development using DTR ≥ 10 ◦C for NPDs in Guangzhou. The method by DTR to select the typical days with full UHI Development is very straightforward and easy to use, which only observation or forecasting of screen level temperature are required. However, there are limitations for this method. The observations of this study are limited by the amount of data analyzed. The DTR threshold in Guangzhou used to select the typical days with full UHI Development need to be examined over a more extended period data in future research. It is better to determine an appropriate DTR thresholds by season if a city with clear seasonal variability in the UHI effect. Moreover, the climate and geographical features of a city influence the DTR in a significant manner and the factors lead to UHI development and its intensity are not equal in all regions. Hence, the DTR threshold may not be suitable for other cities, and more cities with different climate conditions and properties are encouraged to examine and evaluate the effectiveness and threshold to adjust the method. Sustainability 2021, 13, x FOR PEER REVIEW 13 of 15

and 5.22 °C for the THB and TFX sites, respectively, was used to calculate the identifica- tion rate during PDs and FPDs. When DTR was ≥10 °C, only 23 days were selected as sample days, yielding an identification rate of 50% and 69% for the THB and TFX sites, respectively. Although the identification rate was less than 70% at THB, only 23 sample days were selected, indicating that most of the PDs and FPDs were excluded. For practical applications, we suggest that a threshold of 10 °C be used directly for all samples to iden- tify days with full UHI development when focusing only on days with high UHIImax and the precipitation data is difficult to obtain. Otherwise, a higher accuracy will be observed when identifying days with full UHI development using DTR ≥ 10 °C for NPDs in Guang- Sustainability 2021, 13, 320 zhou. 13 of 15

◦ Figure 10. Performance of the DTR threshold of ≥10 C for identifying days with a high UHIImax Figureduring 10. Performance PDs and FPDs. of the DTR threshold of ≥10 °C for identifying days with a high UHIImax during PDs and FPDs. 5. Conclusions

TheSelecting method daysby DTR with to aselect high dailythe typicalUHIImax daysfrom with different full UHI meteorological Development conditionsis very straightforwardis crucial to assess and easy and understandto use, which UHIs. only Inobservation this study, or DTR forecasting was applied of screen as an indexlevel to temperatureidentify days are required. with a high However, daily UHII theremax ar; ae yearlimitations of observation for this method. data was The collected observa- from tionsGuangzhou. of this study The are seasonal limited characteristics by the amount of of the data daily analyzed.UHIImax Thewere DTR consistent threshold with in the Guangzhoufindings ofused previous to select studies. the typical Notably, days precipitation with full UHI negatively Development affects need UHI to development, be exam- inedresulting over a more in a dailyextendedUHII periodmax occurring data in futu morere research. frequently It duringis better NPDs. to determine The correlation an ap- propriatebetween DTR the thresholds daily UHII bymax seasonand DTR if a wascity significantwith clear duringseasonal NPDs. variability Further, in the more UHI days ◦ effect.were Moreover, selected whenthe climateUHII maxand >geographicalUHIImax. Afeatures DTR of of 10 a cityC can influence be used the as DTR a threshold in a significantto identify manner the meteorological and the factors conditions lead to UHI conducive development to UHI and development its intensity are in Guangzhou.not equal in allIt isregions. important Hence, to choose the DTR the threshold typical days may with not be full suitable UHI Development for other cities, in UHI and research.more citiesWe with anticipate different that climate the findings conditions of this an studyd properties will be usefulare encouraged for UHI studies,to examine not onlyand in evaluateGuangzhou the effectiveness but also in otherand threshold cities. We to suggest adjust a the greater method. focus on evaluations of this method and the DTR threshold in other cities with different climate conditions and properties. 5. Conclusions AuthorSelecting Contributions: days withAll a high three daily authors UHII significantly from contributed differentto meteorological the scientific study conditions and writing. Conceptualization, S.Z.; Methodology, S.Z., Y.C.max and G.C.; Supervision, Y.C.; Validation, S.Z.; data is crucial to assess and understand UHIs. In this study, DTR was applied as an index to curation, N.L. and H.H.; Visualization, N.L. and H.H.; Writing—original draft, G.C. and M.H.; identifyWriting—review days with & a editing, high daily G.C. andUHII M.H.max All; a authorsyear ofhave observation read and agreeddata was to the collected published from version Guangzhou.of the manuscript. The seasonal characteristics of the daily UHIImax were consistent with the findings of previous studies. Notably, precipitation negatively affects UHI development, Funding: This research was funded by the National Natural Science Foundation of China (Project resulting in a daily UHIImax occurring more frequently during NPDs. The correlation be- No. 51708129, Project No. 51978173, Project No. 52008115) and the State Key Laboratory of Sub- tween the daily UHII and DTR was significant during NPDs. Further, more days were tropical Building Science,max South China University of Technology, China (Project No. 2018B19, selected when UHII ＞ UHII . A DTR of 10 °C can be used as a threshold to identify Project No. 2021ZB08).max max the meteorological conditions conducive to UHI development in Guangzhou. It is im- Data Availability Statement: The data presented in this study are available on request from the corresponding author. Acknowledgments: Helpful comments from Lihua Zhao and Associate Xiaoshan Yang were greatly appreciated. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. Sustainability 2021, 13, 320 14 of 15

References 1. Oke, T.R.; Mills, G.; Christen, A.; Voogt, J.A. Urban Climate; Cambridge University Press: , UK, 2017. 2. Santamouris, M.; Ding, L.; Fiorito, F.; Oldfield, P.; Osmond, P.; Paolini, R.; Prasad, D.; Synnefa, A.J.S.E. Passive and active cooling for the outdoor built environment—Analysis and assessment of the cooling potential of mitigation technologies using performance data from 220 large scale projects. Sol. Energy 2016, 154, 14–33. [CrossRef] 3. He, B.; Wang, J.; Liu, H.; Ulpiani, G. Localized synergies between heat waves and urban heat islands: Implications on human and urban heat management. Environ. Res. 2020, 193, 110584. [CrossRef] 4. Li, L.; Zha, Y.; Wang, R. Relationship of surface urban heat island with air temperature and precipitation in global large cities. Ecol. Indic. 2020, 117, 106683. [CrossRef] 5. Oke, T.R. City Size and the Urban Heat Island. Atmos. Environ. 1973, 7, 769–779. [CrossRef] 6. Chen, G.; Zhao, L.; Mochida, A. Urban Heat Island Simulations in Guangzhou, China, Using the Coupled WRF/UCM Model with a Map Extracted from Remote Sensing Data. Sustainability 2016, 8, 628. [CrossRef] 7. Stewart, I.D.; Oke, T.R. Local climate zones for urban temperature studies. Bull. Am. Meteorol. Soc. 2012, 93, 1879–1900. [CrossRef] 8. Zhou, D.C.; Zhao, S.Q.; Liu, S.G.; Zhang, L.X.; Zhu, C. Surface urban heat island in China’s 32 major cities: Spatial patterns and drivers. Remote Sens. Environ. 2014, 152, 51–61. [CrossRef] 9. Rajagopalan, P.; Lim, K.C.; Jamei, E. Urban heat island and wind flow characteristics of a tropical city. Sol. Energy 2014, 107, 159–170. [CrossRef] 10. Yang, X.; Yao, L.; Jin, T.; Peng, L.L.H.; Jiang, Z.; Hu, Z.; Ye, Y. Assessing the thermal behavior of different local climate zones in the Nanjing , China. Build. Environ. 2018, 137, 171–184. [CrossRef] 11. He, B.; Ding, L.; Prasad, D. Enhancing urban ventilation performance through the development of precinct ventilation zones: A case study based on the Greater , Australia—ScienceDirect. Sustain. Cities Soc. 2019, 47, 101472. [CrossRef] 12. He, B. Potentials of meteorological characteristics and synoptic conditions to mitigate urban heat island effects. Urban Clim. 2018, 24, 26–33. [CrossRef] 13. Stewart, I.D. A systematic review and scientific critique of methodology in modern urban heat island literature. Int. J. Climatol. 2011, 31, 200–217. [CrossRef] 14. Arifwidodo, S.D.; Tanaka, T. The Characteristics of Urban Heat Island in Bangkok, Thailand. Procedia Soc. Behav. Sci. 2015, 195, 423–428. [CrossRef] 15. Unal, Y.S.; Sonuc, C.Y.; Incecik, S.; Topcu, H.S.; Diren-Ustun, D.H.; Temizoz, H.P. Investigating urban heat island intensity in Istanbul. Theor. Appl. Climatol. 2020, 139, 175–190. [CrossRef] 16. Oke, T.R.; Maxwell, G.B. Urban heat island dynamics in Montreal and Vancouver. Atmos. Environ. 1975, 9, 191–200. [CrossRef] 17. Leconte, F.; Bouyer, J.; Claverie, R.; Petrissans, M. Using Local Climate Zone scheme for UHI assessment: Evaluation of the method using mobile measurements. Build. Environ. Pollut. 2015, 83, 39–49. [CrossRef] 18. Yang, X.; Chen, Y.; Peng, L.L.H.; Wang, Q. Quantitative methods for identifying meteorological conditions conducive to the development of urban heat islands. Build. Environ. 2020, 178, 106953. [CrossRef] 19. Oke, T.R. An Algorithmic Scheme to Estimate Hourly Heat Island Magnitude. In Proceedings of the 2nd Urban Environment Symposium, Albuquerque, NM, USA, 2–6 November 1998. 20. Skarbit, N.; Stewart, I.D.; Unger, J.; Gál, T. Employing an urban meteorological network to monitor air temperature conditions in the ‘local climate zones’ of Szeged, Hungary. Int. J. Climatol. 2017, 37 (Suppl. 1), 582–596. [CrossRef] 21. Yang, X.S.; Yao, L.Y.; Peng, L.L.H.; Jiang, Z.D.; Jin, T.; Zhao, L.H. Evaluation of a diagnostic equation for the daily maximum urban heat island intensity and its application to building energy simulations. Energy Build. 2019, 193, 160–173. [CrossRef] 22. Holmer, B.; Thorsson, S.; Lindén, J. Evening evapotranspirative cooling in relation to vegetation and urban geometry in the city of Ouagadougou, Burkina Faso. Int. J. Climatol. 2013, 33, 3089–3105. [CrossRef] 23. Theeuwes, N.E.; Steeneveld, G.J.; Ronda, R.J.; Holtslag, A.A.M. A diagnostic equation for the daily maximum urban heat island effect for cities in northwestern Europe. Int. J. Climatol. 2017, 37, 443–454. [CrossRef] 24. Yao, L.; Yang, X.; Zhu, C.; Jin, T.; Peng, L.; Ye, Y. Evaluation of a Diagnostic equation for the daily maximum urban heat island effect. Procedia Eng. 2017, 205, 2863–2870. [CrossRef] 25. Kan, H.; London, S.J.; Chen, H.; Song, G.; Chen, G.; Jiang, L.; Zhao, N.; Zhang, Y.; Chen, B. Diurnal temperature range and daily mortality in Shanghai, China. Environ. Res. 2007, 103, 424–431. [CrossRef] 26. Wang, K.; Ye, H.; Chen, F.; Xiong, Y.; Wang, C. Urbanization Effect on the Diurnal Temperature Range: Different Roles under Solar Dimming and Brightening. J. Clim. 2012, 25, 1022–1027. [CrossRef] 27. Yang, J.; Liu, H.Z.; Ou, C.Q.; Lin, G.Z.; Zhou, Q.; Shen, G.C.; Chen, P.Y.; Guo, Y.J.E.P. Global : Impact of diurnal temperature range on mortality in Guangzhou, China. Environ. Pollut. 2013, 175, 131–136. [CrossRef] 28. Liu, W.; Yang, P.; You, H.; Zhang, B. Heat island effect and diurnal temperature range in Beijing area. Clim. Environ. Res. 2013, 18, 171–177. 29. China Meteorological Bureau. China Standard Weather Data for Analyzing Building Thermal Conditions; China Building Industry Publishing House: Beijing, China, 2005; pp. 90–105. 30. Shi, Y.; Xiang, Y.; Zhang, Y. Factors Influencing Surface Urban Heat Island in the High-Density City of Guangzhou Based on the Local Climate Zone. Sensors 2019, 19, 3459. [CrossRef] 31. Evans, J.D. Straightforward Statistics for the Behavioral Sciences; Brooks/Cole Publishing Company: Pacific Grove, CA, USA, 1996. Sustainability 2021, 13, 320 15 of 15

32. Zhang, X.; Steeneveld, G.J.; Zhou, D.; Duan, C.; Holtslag, A.A.M. A diagnostic equation for the maximum urban heat island effect of a typical Chinese city: A case study for Xi’an. Build. Environ. 2019, 158, 39–50. [CrossRef] 33. Miles, V.; Esau, I. Seasonal and Spatial Characteristics of Urban Heat Islands (UHIs) in Northern West Siberian Cities. Remote Sens. 2017, 9, 989. [CrossRef] 34. Liu, W.; Ji, C.; Zhong, J.; Jiang, X.; Zheng, Z. Temporal characteristics of the Beijing urban heat island. Theor. Appl. Climatol. 2007, 87, 213–221. [CrossRef] 35. Jiang, X.D.; Xia, B.C.; Guo, L.; Nan, L.I. Characteristics of multi-scale temporal-spatial distribution of urban heat island in Guangzhou. Chin. J. Appl. Ecol. 2007, 18, 133–139.