Urban Renewal Can Mitigate Urban Heat Islands
Abstract
How can urban renewal effectively contribute to urban climate? Shanghai, one of the world's first metropolises to utilize effective urban heat island (UHI) mitigation strategies, is investigated by analyzing meteorological and land use observations over the past 144 years. The UHI decreased by ~0.58 °C between 2005 and 2016 due to urban renewal characterized by an increase in vegetation cover and the closure of the high-energy consumption industries in urban areas. Simulation results also indicate that future mitigation strategies should strive to increase vegetation cover, as a 10–20% increase in vegetation cover is anticipated to reduce the UHI by 0.38–0.78 °C, resulting in potential electricity savings of 3.05–5.79 × 108 kWh/year, which correspond to a carbon emissions reduction of 2.47–4.68 × 105 tCO2/year. The results will contribute to improve urban climate associated with large-scale urbanization and will provide guidance for urban renewal in other metropolises worldwide.
Plain Language Summary
China has experienced substantial urbanization since 1960, which has led to a significant increase in urban heat islands (UHIs); therefore, considerable urban renewal has been undertaken to mitigate UHI, particularly in the economically developed eastern coastal regions. The impacts of urbanization, including both urban expansion and urban renewal, on the UHI are investigated by analyzing meteorological and land use observations in Shanghai over the past 144 years. Observations indicate that the UHI decrease by ~0.58 °C between 2005 and 2016 due to urban renewal, and a simulated reduction in the UHI of 0.38–0.78 °C is obtained via a 10–20% increase in vegetation cover, resulting in potential electricity savings of 3.05–5.79 × 108 kWh/year, which correspond to a carbon emissions reduction of 2.47–4.68 × 105 tCO2/year. The results will contribute to improve urban climate associated with large-scale urbanization and will provide guidance for urban renewal in other metropolises worldwide.
1 Introduction
Urban heat islands (UHIs), which are urban areas with significantly higher air temperatures than surrounding rural areas (Oke, 1973), are receiving increasing attention due to their impacts on urban climate and energy consumption (Arnfield, 2003). Numerous studies have investigated the influence of urbanization on urban climate and have concluded that urban expansion (Kalnay & Cai, 2003) and human activities (Cao et al., 2016; Zhang et al., 2013) are the primary reasons for UHI development. Urbanization significantly influences the local climate (Oke, 1973; Santamouris, 2015; Yang et al., 2011), leading to an overall urban warming trend that has been detected in all known previous studies (Grimmond, 2007; Kalnay & Cai, 2003; Liao et al., 2017). Considerable UHI mitigation efforts, such as increasing vegetation cover and reducing anthropogenic heat (AH) emissions, have been implemented in many of the world's fastest-growing cities to reduce urban warming (Mcpherson et al., 1994; Rosenzweig et al., 2009; Zhao et al., 2014).
Shanghai, which is at the forefront of China's development, experienced the world's fastest urbanization during 1979–2016 and has devoted significant efforts toward urban renewal since 2005. Large-scale urban renewal programs in Shanghai result in improved urban climate, which therefore provide an ideal laboratory to explore lessons for other metropolises worldwide.
The potential impacts of mitigation measures in improving urban climate have been extensively assessed using numerical simulations (Georgescu et al., 2014; Liu et al., 2018; Rosenzweig et al., 2009). However, these assessments would benefit from the incorporation of long-term observations to accurately assess the practical implementation of these mitigation measures (Memon et al., 2008). To our knowledge, such a practical analysis has not been reported to date.
Here we present a real case study to investigate the UHI evolution in Shanghai using meteorological and land surface observations since 1873, with a focus on its large-scale urbanization since 1960. The aims of the study are to (1) investigate the evolution of UHIs in response to urbanization over a sufficiently long period that includes both urban expansion and urban renewal, (2) examine the factors that contribute to UHI mitigation and identify effective strategies for improving urban climate, and (3) predict future urban climate benefits from these mitigation measures.
2 Data and Methods
2.1 UHI Estimation
Here the UHI is defined as the difference in daily mean air temperature between urban and rural meteorological stations (Oke, 1973; Sakakibara & Owa, 2005; Stewart, 2011). Four-times-daily temperature observations at a height of 1.5 m from the ground at 11 meteorological stations across Shanghai, which are obtained from the China Meteorological Bureau, are analyzed in this study (Figure 1). These observations comply with the World Meteorological Organization's standards of the location of meteorological stations, instruments, observing program, and data process. All air temperature data have been homogenized and quality controlled due to changes in measuring locations, measurement height, and instrumentation practices according to standard methods (China Meteorological Administration, 2003; China Meteorological Administration, 2007). Xujiahui station (station ID: 58367) has 144 years of historical meteorological observations, and the other 10 stations have 57 years of observations. Careful selection of representative urban and rural stations for the UHI analysis is important, because there are significant climatic variations due to the long-term influence of urbanization and varying topography (Peterson & Owen, 2005). The urbanization influence is determined from the impervious surface (IS) fractions around the stations, which are derived from a linear spectral unmixing analysis of Landsat remote sensing data (Wang et al., 2014). Urban stations are those located in areas with a consistently higher IS fraction than those in surrounding rural stations. The reference rural stations should be selected on the basis of both avoiding the effects of urbanization as much as possible and possessing a similar climatic background to that of the urban station (Sakakibara & Owa, 2005). Since the annual temperature cycle in coastal areas lags that in inland areas (Miller, 1953), the reference stations should be selected such that there is no significant difference between the urban and rural station observations of the average monthly temperature amplitude and timing of the warmest/coldest temperatures in the harmonic analysis (Zhou & Shu, 1994). A piecewise linear fitting method is applied to detect the UHI trends during different time periods (Tomé & Miranda, 2004). Two simultaneous conditions are imposed when determining the linear fits in this study: a minimum of 10 years per linear trend and a minimum temperature change of 0.1 °C/10 years. Finally, the solution that minimizes the root mean square error of the linear fits to the entire time series is chosen.
2.2 Land Surface and AH Observations
Land surface observations, including land cover maps, IS, and green vegetation fraction (GF), in approximately 5-year intervals from 1990 to 2016 are estimated using Landsat Level-2 surface reflectance products. Land cover maps in 1873 and 1960 are obtained from historical land cover maps. Detailed methods for mapping land surface observations are provided in the supporting information Text S1. AH from buildings, industry, human metabolism, and vehicles between 1990 and 2016 are estimated by the method in the supporting information Text S2.
2.3 WRF Simulations
The Weather Research and Forecasting (WRF) model coupled with a single-layer urban canopy model (UCM) is a promising utility for urban meteorological modeling and has been widely used for studying the long-term influence of urbanization-induced land cover changes and human activities on the UHI (Chen et al., 2011; Chen et al., 2014; Ronda et al., 2017; Ryu & Baik, 2012). The WRF/UCM configuration is provided in the supporting information Text S3.
(1)
(2)
(3)The methods for analyzing the contributions between 2005 and 2016 are the same as those between 1990 and 2005. The potential contributions of IS, GF, and AH to the UHI are compared with 205 cases where the UHI is >1 °C during either the daytime or nighttime in 1990, which are generally characterized as days with clear skies, no precipitation, and gentle winds (wind speed <2 m/s).
Furthermore, the WRF model is used to predict the impacts of effective mitigation measures on urban climate. This study examines the implementation of GF as a potential urban climate mitigation strategy for the scenarios where the GF is increased by 10% and 20%. The direct repurposing of urban land for GF in high-density urban areas may be unrealistic due to limited available land, such that roof greening, where plants are grown on rooftops, is designed to increase the GF in high-density regions. It is estimated that 20 million m2 of roof space can be used for roof greening in Shanghai, which accounts for approximately 20% of the total area of built-up land (Shanghai Bureau of Statistics, 2017). Increasing the proportion of tree and grass canopy cover in medium-density urban areas can also be characterized by increasing urban park space.
2.4 Electricity Savings
The electricity effects of urban cooling due to effective urban climate mitigation measures are assessed based on electricity consumption. Urban cooling can reduce the summertime electricity demands and increase the wintertime electricity demands. The relationship between electricity consumption and urban thermal environments is analyzed using the degree-day metric (see supporting information Text S4 for more details), which can reflect the electricity demand to heat or cool houses and businesses (Hou et al., 2014; Zhu et al., 1982).
Furthermore, the impacts of electricity consumption on CO2 emissions are assessed by the carbon intensity factor for electricity, which is set to 0.8086 tCO2/MWh, as defined in East China's Regional Grid Baseline Emission Factors in 2015 (National Development and Reform Commission, 2016).
3 Results
3.1 UHI Variations
Representative urban and rural stations for the UHI analysis are carefully evaluated based on our selection criteria (see section 2.1). Xujiahui station, which is located in the central business district with a high percentage of IS cover (Figures 1 and S1), is selected as the urban station. Since Shanghai is a coastal city, the advection of warm or cool air from the sea can influence the temperature records at local meteorological stations based on their proximity to the sea (Sakakibara & Owa, 2005). Our results from a harmonic analysis of the average monthly temperature (see supporting information Text S5) indicate that there are no significant differences in the amplitudes and phase angles of the average monthly temperature cycles between Fengxian station (station ID: 58463) and the selected urban station (Xujiahui station), suggesting that they experience the same offshore effect (Table S2). Therefore, Fengxian station, which is located in an area with low IS cover (Figure S1), is selected as a representative rural station in Shanghai.
The mean annual temperature patterns from the representative stations are shown in Figure 2a, which are important parameters for UHI calculations. The change in mean annual temperature during the 1873–1990 period is 0.03–0.11 °C/10 years (Table S3). The general temperature pattern at both stations shows a high rate of warming during the 1990–2005 period, followed by a slowdown in warming or even slight cooling during the 2005–2016 period. Xujiahui station exhibits a higher warming (or cooling) rate than that at Fengxian station for all the available data during the 1960–2016 period (Table S3).
The UHI first increases and then decreases, as shown in Figure 2b. A piecewise linear fitting method is applied to detect abrupt changes in the UHI trend (see section 2.1), with two significant breakpoints determined in 1990 and 2005. The UHI exhibits a slow increasing trend between 1960 and 1990, with an average rate of only 0.17 °C/10 years (p < 0.01), followed by a significant increasing trend between 1990 and 2005, with an average rate of 0.63 °C/10 years (p < 0.01), and then a decreasing trend between 2005 and 2016, with an average rate of −0.53 °C/10 years (p < 0.01). This recent reduction in UHI is indicated by the higher cooling rate at the urban station than that at the rural station, which is not attributed to changes in meteorological factor characterized by 10-m wind speed, incoming shortwave radiation, and diurnal temperature range based on the UHI diagnostic equation proposed by (Theeuwes et al., 2017) at the representative meteorological station of Shanghai (Figure S2) (see supporting information Text S6 for more details).
3.2 Changes in Land Cover and AH
The UHI changes can be attributed to several factors, including changes in IS, AH, and GF. We estimate these factors during the key stages of urban development (see supporting information Texts S1 and S2). Long-term urbanization is evident in land cover maps that span the 1873–2016 period (Figure S3). There are only small IS cover changes before 1990, whereas significant changes are observed between 1990 and 2005 that are primarily characterized by the conversion of croplands to IS cover at an average annual rate of 4.8%. This large-scale urbanization means that much heat can be stored in the walls of high-rise buildings during the daytime, such that this energy can directly heat the atmospheric boundary layer via turbulent mixing of sensible heat at night. Urban expansion continued at an average annual rate of 2.0% during the 2005–2016 period, which is much lower than that the 1990–2005 rate (4.8%), and the number of high-rise buildings (taller than eight stories) increased at an annual rate of 11%, which is also lower than the 1990–2005 rate (19%) (Shanghai Bureau of Statistics, 2017), suggesting a potential change in the urbanization pattern from urban expansion to urban renewal. Furthermore, high-resolution GF maps have been created to illustrate the urbanization process in more detail (Figure 3). A clear trend of vegetation reduction during the 1990–2005 period is observed within the outer ring (green ring in Figure 3) that encompasses the urban center of Shanghai, with this trend attributed to urban expansion. However, due to effective urban renewal projects, more than half of the reduced GF reappeared as scattered green patches embedded among the high-rise buildings during the 2005–2016 period (Figures 3 and S4), which can play a key role in reducing heat storage and increasing heat release in urban environments.
The spatiotemporal AH distribution, which is sourced from urban metabolism, buildings, vehicles, and industry across the study area during the 1990–2016 period, is shown in Figure S5. The total heat distribution is highly heterogeneous, with higher values along the major road networks and in the industrial and high-building density areas and lower values in the surrounding suburbs. Approximately 90% of the high-energy consumption industries were closed by the end of 2016, leading to a significant decrease in industrial AH in the urban areas. The change in AH sourced from buildings and urban metabolism in urban areas is smaller than that in rural areas during the 2005–2016 period, which may be due to stabilizing building and population densities, as well as the urban renewal programs in the city center. However, the total vehicle AH increases by a factor of ~17 during the 1990–2016 period (Figure S6).
3.3 Impacts of Urbanization on UHI
We further evaluate the impacts of urbanization on the UHI by conducting sensitivity experiments that are designed using the 1990, 2005, and 2016 urbanization-related factors (IS, AH, and GF) under the 1990 meteorological conditions (Table S1). Our evaluation of these simulations highlights that the WRF model produces accurate estimations under real land-surface conditions (Table S4) (see supporting information). The modeled UHI changes for different urbanization scenarios are compared using frequency histograms (Figure 4), which are in agreement with the statistical analysis of the historical meteorological data in Shanghai. The histograms indicate that more days possess an increasing UHI trend (both daytime and nighttime) during the 1990–2005 period (Figures 4a and 4b), with a distinct deviation in this trend observed during the 2005–2016 period. The frequency distribution of daytime changes in UHI during the 2005–2016 period exhibits an approximately normal distribution with zero mean (Figure 4c), whereas a reduction in nighttime UHI is observed during the 2005–2016 period, as evidenced by the large negative tail (decrease in UHI) and small positive tail (increase in UHI) in Figure 4d. Furthermore, the contributions of three urbanization-related factors (IS, AH, and GF) are calculated using equations 1–3 in section 2.3 to better understand how they contribute to the UHI changes during the 1990–2016 period. These trends, with the IS and GF frequency distributions shifted to the right and left, respectively, suggest that IS contributes positively to the increase in UHI intensity, whereas GF exhibits a negative contribution (Figure 5). The positive IS contribution is greater than the negative GF contribution during the 1990–2005 period, which indicates that IS is the primary factor driving this increase in UHI. However, these three urbanization-related factors make a net negative contribution to the UHI changes during the 2005–2016 period, with the GF contribution being the most significant factor in nighttime UHI reduction. The positive AH contribution to the increase in UHI is very small during the 1990–2005 period, with a further reduction in this small positive contribution between 2005 and 2016.
3.4 Future Urban Climate Mitigation
We also conduct WRF simulations for future urban climate mitigation scenarios (10% and 20% increases in GF) to predict the impact of GF on UHI intensity in Shanghai (see section 2.3 for details). The simulation results show that the increases in GF are estimated to reduce the UHI by 0.38–0.78 °C, which is associated with 47–94 additional cooling degree days and 39–82 fewer heating degree days compared to the average of the last 5 years (2012–2016) (see supporting information Text S4). The linear relationship between electricity use E (108 kWh) and the number of degree days, where E = 0.13 × CDDs+77.4 and E = 0.083 × HDDs+78.4, yields changes in the net electricity demand (ΔEnet) that are equivalent to 3.05–5.79 × 108 kWh/year in electricity savings for the 10–20% increase in GF. Furthermore, a marginal carbon intensity factor of 0.8086 tCO2/MWh yields a CO2 emissions reduction of 2.47–4.68 × 105 tCO2/year, which is 0.13–0.25% of Shanghai's total CO2 emissions in 2015 (Shan et al., 2018).
4 Discussion
The effects of urbanization on urban climate in Shanghai are not entirely consistent with previous studies, which have generally indicated an increasing UHI (Kalnay & Cai, 2003; Liao et al., 2017; Yang et al., 2011). This is because few studies have focused on describing the UHI changes during a sufficiently long period with urban renewal. The urban renewal effects in the city center during the 2005–2016 period are obvious in our results (Figures S3 and 3). The selection of the urban and rural stations may bring biases into the UHI calculation. The average of several rural stations is usually used to determine the UHI in Shanghai (Cui et al., 2007; Zhao et al., 2006); however, they do not consider the effects of urbanization on rural stations and the background climate due to their different distances to the sea. Furthermore, previous studies have concentrated mostly on UHIs at global or regional scales, such that the effects of urban climate mitigation, especially minor urban center improvements, on urban climate are not captured at these coarse scales (Li et al., 2004; Peng et al., 2012). The increase in GF, which is characterized by the appearance of small patches embedded in urban areas, releases more latent heat via transpiration and stores less heat via high surface albedo, thereby playing an important role in decreasing UHIs (Zhao et al., 2014). While the net AH contribution to UHI changes is relatively small, the reduction in industrial AH also plays a significant role in offsetting the extreme increase in vehicle AH in urban areas. However, the migration of industrial areas away from urban areas and its impact on urban climate are largely ignored when high spatiotemporal resolution AH data are unavailable.
Urbanization led to a ~1.0 °C increase in the UHI between 1990 and 2005 in Shanghai. This UHI increase would have likely reached ~2.1 °C in 2016 if it had continued to increase at the 1990–2005 rate (0.63 °C/10 years; Table S3). Fortunately, urban climate mitigation has yielded a decrease in UHI of ~0.58 °C due to the slowdown in urbanization and an increase in GF between 2005 and 2016. A 10–20% increase in GF and increased restrictions on the development of high-rise buildings in coastal areas are anticipated to reduce UHI in Shanghai by 0.38–0.78 °C, which will reduce electricity consumption and CO2 emissions, as well as contribute to meeting the global warming limit of 1.5 °C recommended by the International Panel on Climate Change (Shan et al., 2018; Zhao et al., 2006). The role of cities in minimizing global warming cannot be neglected, and further mitigation strategies must be considered for continued improvements to urban climate. This detailed investigation of UHI mitigation in Shanghai provides real examples and lessons for effective urban climate mitigation in cities worldwide.
Acknowledgments
All satellite data are acquired from the United States Geological Survey National Center for Earth Resources Observation and Science (http://earthexplorer.usgs.gov). The energy consumption data are from the Shanghai Statistical Yearbook (http://tjj.sh.gov.cn/html/sjfb/tjnj/). The pollution data are derived from the Shanghai Environmental Monitoring Center and Ministry of Ecology and Environment (http://datacenter.mee.gov.cn/). The meteorological data are downloaded from China Meteorological Data Service Center (http://data.cma.cn). The computation was supported by the East China Normal University Public Platform for Innovation (001). This research was supported by the National Natural Science Foundation of China (Grants 41801014 and 41271055) and the China Postdoctoral Science Foundation (Grant 2017M610237).
