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

: The urban heat island (UHI) poses a signiﬁcant threat to urban ecosystems, 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 temperature range (DTR) and UHI intensity was evaluated and analyzed quantitatively under different meteorological conditions classiﬁed by precipitation. The average daily maximum UHI intensity ( UHII max ) 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 UHII max distribution, which was higher in autumn and winter, but lower in spring and summer. DTR has a signiﬁcant linear correlation with the daily UHII max , with a Pearson’s correlation coefﬁcient of >0.7 and statistical signiﬁcance of <0.001. Based on a quantitative evaluation of our results, we determined that 10 ◦ C could be regarded as the appropriate DTR threshold to identify the meteorological conditions conducive to UHI development; the meteorological conditions exhibited a high daily UHII max in Guangzhou. This study provides a simple method to select typical meteorological days for UHI measurement and simulation, and a method to early-warning of intense UHI events based on weather forecasts.


Introduction
Accompanying global warming and accelerated urbanization, the urban heat island (UHI) effect has become an established urban environment phenomenon. The UHI effect 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 threat to urban ecosystems, human health, and urban energy systems [2,3].
Many factors significantly affect the formation and development of UHIs. Broadly, these factors are classified into two categories based on their influence on the occurrence and development of UHIs [4]. The first category is the characteristics of the city, such as geographical location, topography, morphology, demography [5], land cover and land use [6], built-up intensity, and anthropogenic heat emissions [7]. These factors contribute significantly to the development of UHIs. The second category is meteorological conditions, They proposed that high UHII max 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][26][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 UHII max in Guangzhou. Annual hourly observation data were used to quantitatively analyze the relationship between the DTR and UHII max 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 UHII max 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.

Study Area
Guangzhou spans 7434.4 km 2 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 capital city of Guangdong Province, and the central city of the Guangdong-Hong Kong-Macao Greater Bay Area and Pan-Pearl River Delta Economic Zone. Figure 1 shows an overview of the study area. Guangzhou is characterized by hot summers and warm winters. 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 km 2 and is located around the river. As shown in Figure 1, more rural farmland, represented by large yellow areas, is situated adjacent to the estuary. The large green areas in Figure 1 are mainly evergreen trees or shrubs, where the terrain is low elevation mountains. Most buildings in the city are air-conditioned in the summer.

Site for Urban Air Temperature Measurements
The data on hourly air temperature in the suburbs and urban areas are required to explore the relationship between the DTR and UHII. The air temperature was obtained through field measurements in this study. Selecting an appropriate site for urban temperature measurement is highly important; the temperature should reflect the physical structure, surface properties, and thermal climate of the city, as well as a high UHII. LCZ 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 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 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 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 temperature field experiments was selected based on the LCZ schemes in Guangzhou.
According to the guidelines for using the LCZ classification system suggested by Oke [7], a LCZ is defined as an area with a minimum radius of 200-500 m, which has uniform features in terms of surface cover, structure, material, and human activity. Additionally, different LCZs with a radius of 500 m were selected to study the thermal behavior of LCZs in Nanjing, China [10]. A total of 10 different LCZs were selected to study the thermal behavior in Guangzhou (represented by the black spot in Figure 1). Based on the relationship between UHII and DTR, two LCZ sites (THB and TFX) with a radius of 500 m were selected for meteorological observations in this study. These two sites were selected because they exhibited a higher UHII in a previous study [30]. Figure 1 shows the location of the sites (represented by red spots). One site (THB) is located in the new town center of the city, characterized by high building density and several types of buildings (office buildings, residential buildings, and hotels). This site is classified as LCZ 2. The other site (TFX), characterized by mid-rise residential buildings, belongs to the LCZ 3 and is located in the old town center of the city. Table 1 lists the detailed site parameters and satellite images. Figure 2 shows the satellite images of the selected LCZs.

Site for Urban Air Temperature Measurements
The data on hourly air temperature in the suburbs and urban areas are required to explore the relationship between the DTR and UHII. The air temperature was obtained through field measurements in this study. Selecting an appropriate site for urban temperature measurement is highly important; the temperature should reflect the physical structure, surface properties, and thermal climate of the city, as well as a high UHII. LCZ 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 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 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 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 temperature field experiments was selected based on the LCZ schemes in Guangzhou.
According to the guidelines for using the LCZ classification system suggested by Oke [7], a LCZ is defined as an area with a minimum radius of 200-500 m, which has uniform features in terms of surface cover, structure, material, and human activity. Additionally, different LCZs with a radius of 500 m were selected to study the thermal behavior of LCZs in Nanjing, China [10]. A total of 10 different LCZs were selected to study the thermal behavior in Guangzhou (represented by the black spot in Figure 1). Based on the relationship between UHII and DTR, two LCZ sites (THB and TFX) with a radius of 500 m were selected for meteorological observations in this study. These two sites were selected because they exhibited a higher UHII in a previous study [30]. Figure 1 shows the location of the sites (represented by red spots). One site (THB) is located in the new town center of the city, characterized by high building density and several types of buildings (office buildings, residential buildings, and hotels). This site is classified as LCZ 2. The other site (TFX), characterized by mid-rise residential buildings, belongs to the LCZ 3 and is located in the old town center of the city. Table 1 lists the detailed site parameters and satellite images. Figure 2 shows the satellite images of the selected LCZs. (a) (b)

Experiment Design
The air temperatures measured at the THB and TFX sites were used to represent the urban temperature of Guangzhou. The hourly temperatures of the two sites were derived from the temperature observation experiment of the Guangzhou LCZ, which was conducted in July 2019 and is still ongoing. At each site, two fixed points located in the core 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 installed 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.

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).

Experiment Design
The air temperatures measured at the THB and TFX sites were used to represent the urban temperature of Guangzhou. The hourly temperatures of the two sites were derived from the temperature observation experiment of the Guangzhou LCZ, which was conducted in July 2019 and is still ongoing. At each site, two fixed points located in the core 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 installed 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.

Air Temperature in Suburban Area
We obtained the hourly suburban air temperatures in Guangzhou from the National Meteorological Station (   The air temperatures measured at the THB and TFX sites were used to represent the urban temperature of Guangzhou. The hourly temperatures of the two sites were derived from the temperature observation experiment of the Guangzhou LCZ, which was conducted in July 2019 and is still ongoing. At each site, two fixed points located in the core 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 installed 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.

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).

Experiment Design
The air temperatures measured at the THB and TFX sites were used to represent the urban temperature of Guangzhou. The hourly temperatures of the two sites were derived from the temperature observation experiment of the Guangzhou LCZ, which was conducted in July 2019 and is still ongoing. At each site, two fixed points located in the core 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 installed 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.

Air Temperature in Suburban Area
We obtained the hourly suburban air temperatures in Guangzhou from the National Meteorological Station (

Experiment Design
The air temperatures measured at the THB and TFX sites were used to represent the urban temperature of Guangzhou. The hourly temperatures of the two sites were derived from the temperature observation experiment of the Guangzhou LCZ, which was conducted in July 2019 and is still ongoing. At each site, two fixed points located in the core 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 installed on street lampposts or poles at a height of 2.1-2.5 m from the ground. To avoid the

Experiment Design
The air temperatures measured at the THB and TFX sites were used to represent the urban temperature of Guangzhou. The hourly temperatures of the two sites were derived from the temperature observation experiment of the Guangzhou LCZ, which was conducted in July 2019 and is still ongoing. At each site, two fixed points located in the core 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 installed 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.

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). These data were used to classify the meteorological conditions in Guangzhou and calculate the UHII and suburban DTR. The station is a standard meteorological station that sends and receives data internationally. The station, represented by black pentagram in Figure 1, is located on the northeast of the city. These data include meteorological variables, such Sustainability 2021, 13, 320 6 of 15 as solar radiation, air temperature, humidity, wind speed and direction, precipitation, and atmospheric pressure. 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.

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 = T urban − T suburban ). The maximum UHII in a day is UHII max , 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 UHII max and DTR. The average daily UHII max at each site for the study duration was defined as UHII max . A day was regarded as a full UHI development day when UHII max > UHII max .

DTR in Suburban Areas
The suburban DTR is considered a measure of the cooling potential at night in suburban areas. This can be quantified according to the maximum hourly temperature minus the hourly minimum temperature in a day (DTR = T max − T min ). 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 radiative cooling, 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 urban area, especially after sunset [1].

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 meteorological 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 dailyaccumulated (≥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. Table 2 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.

Characteristics of Daily UH I I max
Research on the seasonal characteristics of the UHI is necessary to alleviate the UHI effect. Figure 3 depicts the monthly statistics and analysis of the daily UHII max for the study duration at the THB and TFX sites. The UHII max was 4.19 • C for 337 days at the THB site, as indicated by the red line in Figure 2a. At the THB site, the maximum of the monthly averages of the daily UHII max was 6.0 • C in December while the minimum was 2.3 • C in May; the monthly average daily UHII max values for September, October, November, and December were significantly higher than the UHII max . Furthermore, more than 20 days and over 75% of every month had a daily characteristic of UHII max > UHII max during these months. However, during the hotter months of May, June, July, and August, the monthly average daily UHII max values were lower than the UHII max , during which no more than 25% of the days exhibited UHII max > UHII max . The characteristics of the daily UHII max distribution at the TFX site were consistent with the THB site.

Characteristics of Daily UHII max
Research on the seasonal characteristics of the UHI is necessary to alleviate the UHI effect. Figure 3 depicts the monthly statistics and analysis of the daily UHII max for the study duration at the THB and TFX sites. The UHII max was 4.19 °C for 337 days at the THB site, as indicated by the red line in Figure 2a. At the THB site, the maximum of the monthly averages of the daily UHII max was 6.0 °C in December while the minimum was 2.3 °C in May; the monthly average daily UHII max values for September, October, November, and December were significantly higher than the UHII max . Furthermore, more than 20 days and over 75% of every month had a daily characteristic of UHII max ＞ UHII max during these months. However, during the hotter months of May, June, July, and August, the monthly average daily UHII max values were lower than the UHII max , during which no more than 25% of the days exhibited UHII max ＞ UHII max . The characteristics of the daily UHII max distribution at the TFX site were consistent with the THB site.
The UHII max of TFX was 4.62 °C for 337 days; Figure 2b shows the monthly distribution of UHII max . At the TFX site, the maximum of the monthly averages of the daily UHII max was 6.45 °C in December while the minimum was 2.57 in May. The average daily UHIIs of the two urban areas were lower in the spring and summer than in the autumn and 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 UHII max is necessary to reveal the characteristics of the UHI. Figure 4 shows the statistics of the occurrence time of the daily UHII max 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 The UHII max of TFX was 4.62 • C for 337 days; Figure 2b shows the monthly distribution of UHII max . At the TFX site, the maximum of the monthly averages of the daily UHII max was 6.45 • C in December while the minimum was 2.57 in May. The average daily UHIIs of the two urban areas were lower in the spring and summer than in the autumn and 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 UHII max is necessary to reveal the characteristics of the UHI. Figure 4 shows the statistics of the occurrence time of the daily UHII max 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 UHII max 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, 320 8 of 15 night due to the differences in cooling rates caused by the thermal characteristics between urban and suburban areas [14]. This observation is consistent with the findings of several previous studies [18]. After sunset, the air layer structure is stable, and the cooling rate is faster because of the increased long-wave radiation in the suburban area due to large open spaces. The urban area retains significant heat during the day due to slow heat dissipation, resulting 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.
NPDs occurred at 23:00 at TFX and 20:00 at THB. PDs occurred at 23:00 at both TFX and THB. The daily UHII max 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 night due to the differences in cooling rates caused by the thermal characteristics between urban and suburban areas [14]. This observation is consistent with the findings of several previous studies [18]. After sunset, the air layer structure is stable, and the cooling rate is faster because of the increased long-wave radiation in the suburban area due to large open spaces. The urban area retains significant heat during the day due to slow heat dissipation, resulting 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.

Influence of Precipitation on UHII max
Guangzhou experiences approximately 114 days of precipitation per year. Therefore, examining the influence of precipitation on the development of the UHI is necessary. Figure 5 depicts the effect of precipitation on the UHII in Guangzhou. At the THB site, the average daily UHII max was 3.1 °C during the PDs, which was approximately 1.8 °C lower than the value during the NPDs. At the TFX site, the average daily UHII max decreased by 1.8 °C due to precipitation. Both sites recorded an approximately 1.8 °C lower average daily UHII max during PDs than during NPDs. This indicates that precipitation has a negative effect on UHI development. For the FPDs, the average daily UHII max was higher than that for the PDs; however, a lower average daily UHII max than that during NPDs proved that FPDs are not conducive to UHI development.

Influence of Precipitation on UH I I max
Guangzhou experiences approximately 114 days of precipitation per year. Therefore, examining the influence of precipitation on the development of the UHI is necessary. Figure 5 depicts the effect of precipitation on the UHII in Guangzhou. At the THB site, the average daily UHII max was 3.1 • C during the PDs, which was approximately 1.8 • C lower than the value during the NPDs. At the TFX site, the average daily UHII max decreased by 1.8 • C due to precipitation. Both sites recorded an approximately 1.8 • C lower average daily UHII max during PDs than during NPDs. This indicates that precipitation has a negative effect on UHI development. For the FPDs, the average daily UHII max was higher than that for the PDs; however, a lower average daily UHII max than that during NPDs proved that FPDs are not conducive to UHI development.  Figure 6 depicts the monthly precipitation and box plots of the daily UHII max for three categories of meteorological conditions classified by precipitation. There is a negative correlation between the daily UHII max and precipitation. The monthly average daily UHII max 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 UHII max because precipitation weakens UHI development.  Figure 6 depicts the monthly precipitation and box plots of the daily UHII max for three categories of meteorological conditions classified by precipitation. There is a negative correlation between the daily UHII max and precipitation. The monthly average daily UHII max was lower during PDs than NPDs, except when more than half of the months Sustainability 2021, 13, 320 9 of 15 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 UHII max because precipitation weakens UHI development. Figure 5. Box plots of the daily UHII max during PDs and NPDs at the study sites. Figure 6 depicts the monthly precipitation and box plots of the daily UHII max for three categories of meteorological conditions classified by precipitation. There is a negative correlation between the daily UHII max and precipitation. The monthly average daily UHII max 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 UHII max because precipitation weakens UHI development.

Relationship between DTR and UHI
To reveal the effect of the DTR on UHI development, the seasonal features and distribution 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 suburbs 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 precipitation 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 negative effect of precipitation on UHI development.

Relationship between DTR and UHI
To reveal the effect of the DTR on UHI development, the seasonal features and distribution 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 suburbs 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 precipitation 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 negative effect of precipitation on UHI development.  Figure 8 depicts the distribution of the UHII max during NPDs and the relation to the DTR at THB and TFX. In the figure, r and P, which represent Pearson's correlation coefficient and statistical significance, respectively, can be used to measure the intensity of the linear association between the UHII max 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 statistically significant when p < 0.05. The daily UHII max 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  Figure 8 depicts the distribution of the UHII max during NPDs and the relation to the DTR at THB and TFX. In the figure, r and P, which represent Pearson's correlation coefficient and statistical significance, respectively, can be used to measure the intensity of the linear association between the 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 statistically significant when p < 0.05. The daily UHII max 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 UHII max in Guangzhou. Previous study in Nanjing showed that the daily UHII max is positively correlated with DTR, where the correlation coefficient is 0.67 and p < 0.005 [31]. The correlation coefficient in Guangzhou is higher than that in Nanjing. Therefore, the DTR can be used as an index to identify days with high daily UHII max in Guangzhou.  Figure 8 depicts the distribution of the UHII max during NPDs and the relation to the DTR at THB and TFX. In the figure, r and P, which represent Pearson's correlation coefficient and statistical significance, respectively, can be used to measure the intensity of the linear association between the UHII max 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 statistically significant when p < 0.05. The daily UHII max 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 UHII max in Guangzhou. Previous study in Nanjing showed that the daily UHII max is positively correlated with DTR, where the correlation coefficient is 0.67 and p < 0.005 [31]. The correlation coefficient in Guangzhou is higher than that in Nanjing. Therefore, the DTR can be used as an index to identify days with high daily UHII max in Guangzhou.

Thresholds for Identifying Days with High Daily UHII max
A significant correlation between the DTR and UHII max indicates that the DTR can be used to identify days with high daily UHII max in Guangzhou. Figure 9 shows the relationship between the DTR and UHII max for 194 NPDs. The UHII ̅̅̅̅̅̅̅ max for 194 NPDs was

Thresholds for Identifying Days with High Daily UH I I max
A significant correlation between the DTR and UHII max indicates that the DTR can be used to identify days with high daily UHII max in Guangzhou. Figure 9 shows the relationship between the DTR and UHII max for 194 NPDs. The UHII max for 194 NPDs was 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 UHII max increased, and the number of days with a daily UHII max > UHII max decreased with an increasing DTR. At the THB site, 183 days were selected when the DTR was ≥5 • C, and the average daily UHII max was 4.93 • C. Only 48.6% of the days were characterized by UHII max > UHII max . At the TFX site, the average daily UHII max was 5.37 • C and 48.6% of the days recorded UHII max > UHII max 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 UHII max > UHII max 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 UHII max 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 and analyzed to verify UHI development in Guangzhou; further, more cities should be examined to fully evaluate the effectiveness and threshold.
were selected when the DTR was ≥5 °C, and the average daily UHII max was 4.93 °C. Only 48.6% of the days were characterized by UHII max ＞ UHII max . At the TFX site, the average daily UHII max was 5.37 °C and 48.6% of the days recorded UHII max ＞ UHII max 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 UHII max ＞ UHII max 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.

Discussion
Meteorological conditions impact the UHI in a significant manner. Therefore, selecting the days with a high daily UHII max 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 UHII max 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 and winter seasons in northern west Siberian cities [33] and in the winter in Beijing [34]. The monthly distribution of the daily UHII max 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 UHII max 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 UHII max 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 UHII max . In other words, more precipitation days will be eliminated with an increasing DTR. Based on these observations, the relationships between the DTR and daily UHII max 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 UHII max in Guangzhou. Figure 10 shows the performance of a DTR threshold of ≥10 • C for identifying days with a high UHII max during PDs and FPDs. Here, a stricter UHII max 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. 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 identify days with full UHI development when focusing only on days with high UHII max 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. 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 UHII max 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.

Conclusions
Selecting days with a high daily UHII max from different meteorological conditions is crucial to assess and understand UHIs. In this study, DTR was applied as an index to identify days with a high daily UHII max ; a year of observation data was collected from Guangzhou. The seasonal characteristics of the daily UHII max were consistent with the findings of previous studies. Notably, precipitation negatively affects UHI development, resulting in a daily UHII max occurring more frequently during NPDs. The correlation between the daily UHII max and DTR was significant during NPDs. Further, more days were selected when UHII max ＞ UHII max . A DTR of 10 °C can be used as a threshold to identify the meteorological conditions conducive to UHI development in Guangzhou. It is im- Figure 10. Performance of the DTR threshold of ≥10 • C for identifying days with a high UHII max during PDs and FPDs.

Conclusions
Selecting days with a high daily UHII max from different meteorological conditions is crucial to assess and understand UHIs. In this study, DTR was applied as an index to identify days with a high daily UHII max ; a year of observation data was collected from Guangzhou. The seasonal characteristics of the daily UHII max were consistent with the findings of previous studies. Notably, precipitation negatively affects UHI development, resulting in a daily UHII max occurring more frequently during NPDs. The correlation between the daily UHII max and DTR was significant during NPDs. Further, more days were selected when UHII max > UHII max . A DTR of 10 • C can be used as a threshold to identify the meteorological conditions conducive to UHI development in Guangzhou. It is important to choose the typical days with full UHI Development in UHI research. We anticipate that the findings of this study will be useful for UHI studies, not only in Guangzhou but also in other cities. We suggest a greater focus on evaluations of this method and the DTR threshold in other cities with different climate conditions and properties.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.