Investigating Variations in Anthropogenic Heat Flux along Urban–Rural Gradients in 208 Cities in China during 2000–2016

Abstract

Anthropogenic heat emissions, which are quantified as anthropogenic heat flux (AHF), have attracted significant attention due to their pronounced impacts on urban thermal environments and local climates. However, there remains a notable gap in research regarding the distinctions in the distribution of anthropogenic heat emissions (AHEs) along urban–rural gradients. To address this gap, the present study introduces a new concept—the anthropogenic urban heat island (ArUHI)—where the AHF within urban areas is higher than that in background areas. To quantitatively describe the magnitude and spatial extent of the ArUHI effect, two metrics—namely, ArUHI intensity (ArUHII) and ArUHI footprint (ArUHIFP)—are introduced. We conducted a comprehensive study across 208 cities in China to investigate the spatiotemporal patterns of AHF variations along urban–rural gradients during the period of 2000–2016. In addition, we explored how the complex interactions between land cover and building form components affect changes in the AHF along urban–rural gradients. Additionally, we analyzed how economic zones and city sizes alter the ArUHI intensity and ArUHI footprint. The results showed that 97% (201/208) of Chinese cities exhibited a significant ArUHI effect from 2000 to 2016. The modeled ArUHI intensity value exhibited a substantial increase of nearly fivefold, increasing from 5.55 ± 0.19 W/m² to 26.84 ± 0.99 W/m² over time. Regarding the spatial distribution of the ArUHI footprint, the analysis revealed that, for the majority of cities (86% or 179 out of 208), the ArUHI footprint ranged from 1.5 to 5.5 times that in urban areas. City sizes and economic zones yielded significant influences on the ArUHI intensity and ArUHI footprint values. Building forms were significantly positively correlated with AHF, with R² values higher than 0.94. This study contributes to the understanding of ArUHI effects and their driving factors in China, providing valuable insights for urban climate studies and enhancing our understanding of surface urban heat island mechanisms.

1. Introduction

Rapid global urbanization and industrialization processes have resulted in significant heat emissions from human activities, which are known as anthropogenic heat emissions [1]. AHEs disrupt the energy/water balance on the Earth’s surface and the exchange of mass and momentum [2,3,4]. Huang [5] indicated that AHEs cause an increase in surface temperatures in urban areas. Additionally, AHEs contribute to increased concentrations of urban pollutants [6,7,8] and the elevated occurrence of extreme heatwaves and heavy rainfall events [9,10,11,12,13,14], as well as having detrimental effects on the physical health of urban residents [15,16,17]. Therefore, acquiring a comprehensive understanding of the characteristics of AHEs in urban areas is crucial for comprehending urban thermal environments, mitigating the adverse impacts resulting from urbanization, and improving the living conditions of urban residents.

The distribution of AHEs in urban areas and background areas exhibits significant heterogeneity across different cities due to the diversity of human activities and the unevenness of urban development [18,19]. Urbanization has led to a substantial increase in urban populations and the transformation of land surfaces, resulting in the generation of AHEs from human metabolism and residential buildings [20,21,22]. Furthermore, the growth of urban industrialization and commercialization has contributed to increased AHEs from commercial buildings and factories within cities [21,23]. Conversely, historical buildings, which have lower AHEs, are often located in urban centers [24], while transportation-related AHEs tend to be less concentrated. Shipping ports and airports, which generate higher levels of AHEs, are often situated away from central urban areas [21,25]. As a result, it is still essential to uncover how AHEs vary along urban–rural gradients and understand the underlying mechanisms. Quantitative investigations and analyses of AHE distributions in urban areas and background areas are necessary. Such research will facilitate urban climate modeling and inform urban energy planning efforts. Earlier studies delved into the spatiotemporal characterization of AHF and explored the association between it and the structures of underlying surfaces. For instance, Chen [26] employed the Cubist model, integrating points of interest and multisource remotely sensed data to estimate AHFs in China at a spatial resolution of 1 km. Their findings revealed substantial variations in AHF values in cities of different sizes, ranging from 60 to 190 W/m². Additionally, Wang [27], combining energy consumption data, socioeconomic statistics, and multisource remotely sensed data, applied a partition model to estimate AHF values in China from 2000 to 2016. Their model demonstrated high precision, with the highest R² value reaching 0.97. Importantly, the spatial distribution of AHF exhibited notable disparities across different economic zones. Previous findings consistently highlight the substantial influence of city size and economic background on AH emissions.

According to existing research, AHF at the urban scale typically displays a radial spatial distribution pattern, diminishing as it extends toward the periphery. This variation aligns with factors such as population density, building density, and traffic volume [22,23,28,29]. Nevertheless, at present, there is a gap in quantitative studies that delineate the concentrated trend of AHF from urban to background areas and explore the variations in this trend among different cities. In the existing literature, it is indicated that the AHF distribution exhibits significant differences in urban, suburban, and rural areas [20,29]. These studies represent some of the limited research efforts delving into the variations in AHF between urban and background areas. However, the current research lacks specially designed standardized measurement standards to quantify the trend of concentration changes from cities to background areas.

In light of the preceding discussion, this study introduces a novel concept termed the anthropogenic urban heat island effect (ArUHI). This concept employs AHF as the measurement criterion, signifying anthropogenic heat fluxes in urban areas that surpass those in background regions. China, as the world’s largest developing country, has experienced swift urbanization over the last thirty years, with many cities experiencing elevated levels of anthropogenic heat emissions (AHEs) [30]. Projections suggest that, by 2050, around 70% of China’s urban population will be affected by excessive AHEs [31]. Investigating the existence of ArUHI effects in China and understanding how AHEs vary along urban–rural gradients is valuable. Given that AH emissions are some of the primary causes of the surface urban heat island (SUHI) effects [32], this research contributes to urban climate studies and enhances our understanding of the mechanisms of SUHIs.

We undertook a comprehensive study across 208 cities in China to probe the extended temporal patterns and spatial imprints of anthropogenic urban heat island (ArUHI) effects during 2000–2016. The scientific inquiries are encapsulated in the following questions:

• Are ArUHI effects discernible in China throughout the specified period, and what nuances characterize their spatiotemporal variations?
• How do the intricate interactions between land cover and building form components influence the variations in AHF along urban–rural gradients?
• To what extent can economic zones and city sizes alter the intensities and footprints of ArUHI effects?

This multifaceted investigation aims to shed light on the dynamics of the influence of anthropogenic heat in Chinese urban areas, contributing to a deeper understanding of the complexities inherent in ArUHI effects. Figure 1 illustrates the hypotheses underpinning this study. We posit that the ArUHI effect attains its maximal intensity within urban regions and progressively diminishes as it extends along the urban–rural gradient [33,34,35]. Moreover, we hypothesize that land cover components and building forms play a pivotal role in shaping the ArUHI effect, and economic zones and city size may exert influences on both the intensities and footprints of ArUHI. Studying the ArUHI effect is beneficial for understanding its extent along urban–rural gradients through anthropogenic heat flux in cities. This, in turn, provides a basis for urban energy planning and development.

2. Materials and Methods

2.1. Materials

2.1.1. Study Area

The study area for our research comprised 208 cities in China (Figure 2). China has undergone significant urbanization over the last four decades, resulting in substantial AHEs. The selection of these 208 cities was based on data availability and their representativeness in capturing the diverse urban contexts across the country. To examine the influence of economic zones on AH emissions, we classified the 208 cities into eight economic zones according to the Development Research Center of the State Council, China. For further details on the economic division scheme, please refer to Table S1. By considering the economic zones, we aimed to investigate potential variations in AHEs within different regional contexts, taking the economic characteristics, development patterns, and industrial activities specific to each zone into account. In addition to the economic division, we further defined the selected cities into five levels based on their population size by following the categorization proposed by [36].

This population-based classification scheme allowed for a finer understanding of the variations in AHEs across cities with different sizes, and helped to identify potential differences in AHE patterns, urban characteristics, and development scales among the selected cities.

2.1.2. Data Used

The data used in this study mainly included AHF data, land use data, and two-dimensional (2D) and three-dimensional (3D) building form data, as detailed in

Table 1. City Level division scheme used in the study.

Level	Abbreviation	Population Size
Super city	Level 1	>10 million
Mega city	Level 2	>3 million, and <10 million
Large city	Level 3	>1 million, and <3 million
Medium city	Level 4	>500,000, and <1 million
Small city	Level 5	<500,000

.

• AHF data

AHF data provide a quantitative means of assessing the disparities in anthropogenic heat intensities between urban and rural areas. The estimation of AHF encompasses a spectrum of methodologies, including energy inventory approaches, surface energy balance residual models, and building energy models [28]. These approaches have been harnessed to compute AHF across diverse spatiotemporal scales. Recent advancements in this field have harnessed the amalgamation of remotely sensed data, techniques for temporal downscaling, and machine learning algorithms to fabricate comprehensive AHF datasets capable of encapsulating both the spatial and temporal fluctuations in AHF [37,38,39,40]. Among these methods, the energy inventory approach draws on energy consumption data and socioeconomic statistics, allocating them to different study areas using gridded spatial information while often utilizing sources such as night light data [29]. This approach is particularly effective for estimating AHF at larger and medium scales.

In this context, we utilized an AHF dataset (Figure 3) generated by combining the inventory-based method with multiple remotely sensed data sources [27]. This gridded AHF dataset boasts a high spatial resolution of 500 m, spanning the years 2000 to 2016 with a four-year interval. Significantly, it offers not only robust accuracy but also temporal consistency. The dataset’s strength is highlighted by its average coefficient of determination (R²) value of 0.76 between AHF and the adjusted night-time light urban index. This makes the dataset well-suited for exploring the anthropogenic urban heat island effect on a national scale.

• Land use data

The land use data utilized in this study were acquired from the Resource and Environmental Science and Data Center of the Chinese Academy of Sciences (RESDC) and are accessible at https://www.resdc.cn/DOI/DOI.aspx?DOIID=54 (accessed on 1 August 2024). This dataset relies on Landsat satellite imagery data as its primary source of information. It is important to note that the creation of this dataset involved manual visual interpretation, where experts thoroughly analyzed the satellite images to classify and outline different land use categories. The selected land use data had a spatial resolution of 1 km and encompassed the years 2000, 2005, 2010, and 2015. The primary purpose of integrating this dataset into the research was twofold: to delineate urban and rural areas accurately and to investigate the correlations between changes in AHF and the evolving land use patterns along the urban–rural gradient.

• Two-dimensional (2D) and three-dimensional (3D) building forms

Two-dimensional and three-dimensional building form parameters (BFPs), including building coverage (BC), mean building height (MBH), and mean building volume (MBV), hold a pivotal role in shaping anthropogenic heat emissions. These parameters exert a substantial influence on the variability of AHF. In our pursuit to comprehend how building forms impact the intensity of the anthropogenic urban heat island (ArUHI) effect across urban–rural gradients, we employed the BFP open dataset developed by the authors of [41]. This dataset provides invaluable BC, MBH, and MBV information with a spatial resolution of 1 km, and it is specifically for mainland China around the year 2015. It was curated through the fusion of a diverse array of spatial and reference data sources, including Landsat 8 spectral bands, Sentinel Synthetic Aperture Radar images, 3D building data, and street-view maps procured from commercial map vendors. This fusion process utilized a random forest algorithm to facilitate inversion. The quality and accuracy of the derived building density, building height, and building volume parameters were rigorously assessed by comparing them with observed data, yielding highly satisfactory outcomes. The coefficient of determination (R²) values for building coverage, height, and volume stood at 0.90, 0.81, and 0.88, respectively [41]. This attests to the robustness and reliability of the dataset, affirming its utility in investigating the intricate relationship between building forms and their implications for AHF and ArUHI effects (Figure 4).

2.2. Methods

2.2.1. Delineation of Urban and Background Areas

To accurately quantify the intensities of the ArUHI effect and determine its spatial extent, an accurate demarcation of urban and rural regions is crucial. In this study, we employed built-up areas as the defining criterion for urban regions. To delineate rural (or background) regions, a well-established method was adopted, which involved the creation of buffer zones around urban areas [42,43]. Urban boundaries for the specified years of 2000, 2004, 2008, 2012, and 2016 were meticulously defined by integrating land use data from corresponding periods: 2000, 2005, 2010, and 2015, respectively [44]. For each city, 14 adjacent buffer zones were generated, with each buffer zone’s area set at half of the urban area (Figure 5). The procedure for generating buffer zones included the following steps:

• Utilizing the Geopandas Python library, a loop structure was implemented to iteratively generate buffer zones, which varied in size from 1.5 to 8 times the urban area. The difference in area between adjacent buffer zones was set at 0.5 times the urban area.
• The intersection of adjacent buffer zones was performed, followed by the removal of overlapping sections, resulting in 14 non-overlapping buffer zones, each with an area equivalent to half of the urban area.

In accordance with the methodology proposed by Zhou [45], the designation of the natural background encompassed the three buffer zones farthest from the urban area. Additionally, in alignment with the approach outlined by Imhoff [34], pixels with elevations surpassing ± 50 m from the median elevation value of urban regions were excluded. This step was taken to mitigate the potential influence of varying elevations on the analysis. Furthermore, any built-up areas within rural regions were excluded to ensure the precision of the ArUHI assessment.

2.2.2. Definition of Anthropogenic Urban Heat Islands

Anthropogenic heat emissions can be quantified using the AHF. As such, the concept of the ArUHI can be characterized by discrepancies in AHF between urban and rural areas, denoted as ΔAHF. In this research, we operationalized ΔAHF_i as the variance in AHF between the i-th buffer zone (AHF_i) and the natural background area (AHF_b). This can be computed using the following formula:

$${\mathsf{\Delta }\mathit{AHF}}_i={\mathit{AHF}}_i-{\mathit{AHF}}_{\mathrm{b}},i=\left\{0\dots 14\right\},$$

(1)

where AHF₀ represents the AHF in urban areas and AHF_b denotes the AHF value of the natural background area. To portray the intensity of anthropogenic heat emissions in the natural background, the median AHF value from the three buffer zones positioned farthest from the city was selected. The rationale behind choosing the median value was to mitigate the influence of outlier data points present in the three buffer zones situated at the greatest distance from the city:

$${\mathit{AHF}}_{\mathrm{b}}=\mathrm{median}\left({\mathit{AHF}}_{12},{\mathit{AHF}}_{13},{\mathit{AHF}}_{14}\right),$$

(2)

where AHF₁₂, AHF₁₃, and AHF₁₄ correspond to AHF values within the 12th, 13th, and 14th buffer zones, respectively. Furthermore, we introduced an additional index to verify the presence of the ArUHI effect in a city:

$$C_{AUHI}={\displaystyle \frac{{\mathit{AHF}}_0-{AHF}_{\mathrm{b}}}{{AHF}_{\mathrm{b}}}},$$

(3)

where C_ArUHI stands for the city-level ArUHI effect, which is a measure of the rate of growth of the AHF in urban areas compared with that in rural regions within a city. This metric is used to assess the extent of the urban heat island effect at the city level. As indicated by Huang [7], a city is deemed to exhibit an ArUHI effect when the C_ArUHI value exceeds 0.05.

To delve into the evolving trends of ΔAHF_i as one moves across the urban–rural spectrum within each city, an exponential decay model was formulated:

$${\mathsf{\Delta }\mathit{AHF}}_i=A\times {\mathrm{e}}^{-S\times i}+{\mathit{AHF}}_{\mathrm{a}},$$

(4)

where A signifies the maximum difference in AHF, S represents the decay rate of the exponential trend, i denotes the distance from the urban area, which is also the buffer zone index, and AHF_a is the asymptotic value of the exponential trend. If the calculated ΔAHF_i values for a particular city align well with this exponential decay model—that is, if the coefficient of determination (R²) value is greater than 0.9—this suggests that the city is experiencing an ArUHI effect. This model helps in understanding the gradual decrease in AHF differences as one moves away from the urban core towards rural areas, providing insights into the influence of urbanization on heat emissions.

2.2.3. Metrics of ArUHI: ArUHII and ArUHIFP

To quantitatively assess ArUHI effects, we introduced two metrics: the anthropogenic urban heat island intensity (ArUHII) and anthropogenic urban heat island footprint (ArUHIFP). These metrics were specifically developed to aid in the quantitative analysis and characterization of the ArUHI phenomenon, enabling a better understanding of its intensity and spatial influence.

The ArUHII index is defined as the maximum difference in AHF between the urban region and natural background areas in a city. Mathematically, it is calculated as follows:

$$\mathit{ArUHII}={\mathit{AHF}}_0-{AHF}_{\mathrm{b}},$$

(5)

where AHF₀ denotes the AHF in urban areas and AHF_b denotes the AHF value of the natural background area.

The ArUHIFP index serves as a metric that quantitatively assesses the span of persistent influence exerted by the ArUHI effect, extending from urban areas into background regions. This concept parallels the definition of the surface urban heat island (SUHI) effect [35]. Our hypothesis posits that the ArUHI effect exhibits its most substantial magnitude within urban regions and gradually attenuates as it extends along the urban–rural gradient [33,34,35]. This makes it vital to precisely gauge the spatial extent of influence of ArUHI effects, giving rise to the ArUHIFP metric. By establishing the ArUHIFP, we gain insights into the spatial reach and intensity of the ArUHI effect, offering a comprehensive understanding of its influence across different urbanization levels.

Quantifying the footprint of the ArUHI effect can indeed be challenging due to the significant spatial and temporal variations in ΔAHF (i.e., urban–rural changes in AHF) across different buffer zones or spatial scales [23,26]. Additionally, the ArUHI effect can vary widely not only from city to city but also within a single city, making it a complex phenomenon to characterize [46]. Zhang [35] introduced a method for determining the footprint of urban climates. This method relies on measuring the distance at which each of the exponential models reaches 95% of its asymptotic values (i.e., AHF_a). It is important to note that this method may not effectively capture the footprint of ArUHI for cities where there is no significant delay trend in ∆AHF_i along the urban–rural gradient [45]. However, in our study, we specifically calculated this footprint for cities with ArUHI effects. Therefore, the method proved to be feasible and applicable in our research context.

2.2.4. Statistical Analysis

• Selection of the influencing factors

Table 2. Data used in the study.

Data	Year	Resolution	Source
AHF data	2000–2016	500 m × 500 m	https://doi.org/10.1016/j.scitotenv.2020.139457 (accessed on 1 August 2022)
Land use data	2000–2015	1 km × 1 km	https://www.resdc.cn/DOI/DOI.aspx?DOIID=54 (accessed on 1 August 2022)
2D and 3D building form data	2015	1 km × 1 km	https://doi.org/10.1016/j.rse.2020.111859 (accessed on 1 August 2022)

shows the influencing factors of ArUHIs. In this study, we focused on the influence of land use compositions and building forms on ArUHI effects. Land use compositions encompass parameters such as the rural residential area proportion (RA_P) and the proportion of other types of construction (OC_P). Concerning building forms, we considered building coverage (BC), mean building height (MBH), and mean building volume (MBV) (

Table 3. Influencing factors of ArUHIs.

Category	Parameter	Abbreviation	Definition	Source
Land use patterns	Rural residential area proportion	RA_P	The percentage of rural residential area in a 1 × 1 km grid	RESDC
	Proportion of other types of construction	OC_P	The percentage of other types of construction in a 1 × 1 km grid	RESDC
Building forms	Building coverage	BC	The proportion of building area to total pixel area	Li et al. (2020) [41]
	Mean building height	MBH	The average height of buildings	Li et al. (2020) [41]
	Mean building volume	MBV	The average volume of buildings	Li et al. (2020) [41]

).

• Statistical analyses

The ordinary least squares (OLS) model was employed to investigate the impacts of the selected factors on the ArUHI effect. The OLS model, a classic statistical analysis method, aims to minimize the sum of squared residuals [47]. In previous studies, the OLS model has been frequently utilized to explore relationships between variables [48]. Its formula is as follows:

$$y={\beta }_0+{\beta }_{1}x+{\epsilon }_k,$$

(6)

where y is the dependent variable (i.e., AHF along the urban–rural gradient); x refers to the independent variable (i.e., influencing factors in urban areas and different buffer zones); β₁ and β₀ are the regression coefficient and intercept, respectively; and ε_k is the residual error. Additionally, two measurement metrics were used to evaluate the model performance: the coefficient of determination (R²) value and the pseudo-t-test values of AHF and influencing factors.

3. Results

3.1. Exponential Decay of ArUHIs along Urban-Rural Gradients

Following the methods outlined earlier, we performed calculations to ascertain both the ArUHII and ArUHIFP for each city in different time periods. Figure 6 serves as an illustrative example, demonstrating the computation of the ArUHII and ArUHIFP for the city of Beijing over the period spanning from 2000 to 2016. The results depicted in Figure 6 reveal a notable increase in the ArUHII value over this timeframe, rising from 6.82 W/m² in 2000 to 30.21 W/m² in 2016. This substantial rise in ArUHII signifies a significant intensification of the ArUHI effect within Beijing during these years. Furthermore, it is important to note that the value of the ArUHIFP exhibited relative stability throughout these years, with fluctuations typically ranging from approximately two to three times that of the corresponding urban area. This suggests that, while the intensity of the ArUHI effect experienced significant growth in Beijing from 2000 to 2016, the spatial extent of its influence, as depicted by the ArUHIFP, remained relatively constant. It was noted that the decay rate, represented by the parameter S, experienced a significant increase, climbing from 0.64 to 0.91 (Figure 6). A higher value of S signifies a swifter decline in AHF along the urban–rural gradient. In simpler terms, this suggests that the decreased rate of heat emissions decreased more rapidly along the urban–rural gradient in Beijing over time. These parameter values offer valuable insights into both the magnitude and characteristics of the ArUHI effect within the city of Beijing over the specified time period.

Upon examination, it was found that out of the 208 cities studied, all but seven cities (97% of the cities selected)-namely, Foshan, Shuangyashan, Karamay, Baotou, Yulin, Tianshui, and Zhangjiajie—exhibited an exponential decay trend in ΔAHF_i from urban areas to background areas and a high C_ArUHI value (i.e., the value of C_ArUHI was >0.05, and the R² value of the decay trend model was >0.90). To evaluate the multi-year average status of the ArUHI effect in China, we calculated the average ΔAHF_i value from 2000 to 2016. As shown in Figure 7A, a significantly exponential decay trend of ΔAHF_i was observed (R² value = 0.99 and p-value < 0.01) during the period of 2000–2016. The average ArUHII value was 12.44 ± 0.43 W/m². The findings mentioned above provide confirmation of the existence of the ArUHI effect in Chinese cities, underscoring that the AHF within urban areas was notably higher than that in the background areas, and that this heat diminished swiftly as it extended towards the natural background areas, following an exponential model of decay (

Table 4. Parameters of the exponential model of AHF from 2000 to 2016.

Year	A	S	AHFa	R2
2000	5.55 ± 0.19	0.99 ± 0.07	1.00 ± 0.08	0.99
2004	8.20 ± 0.26	0.96 ± 0.07	1.42 ± 0.10	0.99
2008	9.91 ± 0.31	1.01 ± 0.07	1.97 ± 0.12	0.99
2012	11.84 ± 0.37	1.20 ± 0.08	2.99 ± 0.17	0.99
2016	26.84 ± 0.99	1.01 ± 0.08	6.08 ± 0.39	0.98
Mean	12.44 ± 0.43	1.04 ± 0.08	0.32 ± 0.17	0.99

).

Figure 7B–F depicts the temporal variations in ΔAHF_i across the 201 cities where ArUHII effects were observed, spanning the years from 2000 to 2016. Notably, the modeled maximum AHF difference between urban and background regions, represented by parameter A (which was closely related to the ArUHII), exhibited a substantial increase of nearly fivefold. It increased from 5.55 ± 0.19 W/m² to 26.84 ± 0.99 W/m². This significant increase underscores the intensification of the ArUHI effect during this period. In addition, the increase in AHF in background regions, as reflected by the change from 1.00 ± 0.08 W/m² to 6.08 ± 0.39 W/m², suggests a rising trend in AHF within rural regions. This finding implies that rural areas also experienced an escalation in anthropogenic heat emissions.

3.2. The Influence of City Sizes and Economic Zones on the Exponential Decay of ΔAHF

Figure 8 illustrates the exponential trends of ArUHI effects along urban–rural gradients across different city sizes. Notably, the highest value for the maximum difference in AHF between urban and rural regions was observed in small cities, with an A value of 19.71 ± 0.37 W/m². The lowest value of A was found to be in large cities (A value = 12.21 ± 0.42 W/m²). Interestingly, the population of large cities (i.e., those with a population between 1 and 3 million) emerged as a critical point. In cities with populations exceeding 3 million (i.e., mega and super cities), the A value increased with the increase in population. For instance, the A value in super cities (13.83 ± 0.45 W/m²) was found to be higher than that in mega cities (13.39 ± 0.14 W/m²). In contrast, when the population was below 1 million, the A value decreased as the population increased. For example, the A value in small cities (19.71 ± 0.37 W/m²) was found to be higher than that in medium cities (14.45 ± 0.65 W/m²). These findings highlight the nuanced relationship between the population-based city size and the intensity of anthropogenic heat emissions. In general, the value of parameter S increased with the growth of city sizes. The highest decay rate of the exponential trend of AHF was observed in super cities (S value = 1.33), and the ranking of the S value among different city sizes was as follows: super (1.33), mega (1.11), large (0.96), small (0.49), and medium (0.40) cities. This implies that larger cities underwent a more rapid decrease in heat emissions along the urban–rural gradient. In terms of the background AHF, specifically AHF_a, it was noted that larger cities showcased elevated AHF_a values. For example, super cities recorded the highest AHF_a value at 3.11 ± 0.15 W/m², which was 1.47 times greater than that of small cities (2.12 ± 0.29 W/m²) (

Table 5. Parameters of the exponential model of AHF in cities of different sizes.

City Size	A	S	AHFa	R2
L1	13.83 ± 0.45	1.33 ± 0.02	3.11 ± 0.15	0.99
L2	13.39 ± 0.14	1.11 ± 0.06	2.34 ± 0.14	0.99
L3	12.21 ± 0.42	0.96 ± 0.08	2.60 ± 0.17	0.98
L4	14.45 ± 0.65	0.40 ± 0.02	2.23 ± 0.64	0.98
L5	19.71 ± 0.37	0.49 ± 0.10	2.12 ± 0.29	0.99

).

Figure 9 shows the exponential trends of ΔAHF along urban–rural gradients across distinct economic zones. Remarkably, the Northeast Economic Zone (NEEZ) displayed the highest modeled ArUHII, with an A value of 21.01 ± 0.97 W/m². This indicates that AH emissions in the NEEZ exhibited a more pronounced contrast between urban and rural areas than in other economic zones. This distinction could be attributed to the NEEZ being a significant hub for heavy equipment manufacturing [49], causing a substantial amount of industrial waste heat in urban areas. Additionally, the Northwest Economic Zone (NWEZ) also showed a higher ArUHII value (21.01 ± 0.97 W/m²), which was possibly due to the increased energy consumption for heating in northern cities during winter, resulting in significant differences in AH emissions between urban and rural regions [50]. The Southern Coastal Economic Zone (SCEZ) and Middle Yangtze River Economic Zone (MAYEZ) showed lower ArUHII values (i.e., 8.35 ± 0.22 and 8.31 ± 0.27 W/m², respectively). In contrast, the Southern Coastal Economic Zone (SCEZ) and the Middle Yangtze River Economic Zone (MAYEZ) exhibited lower ArUHII values, specifically 8.35 ± 0.22 W/m² and 8.31 ± 0.27 W/m², respectively. This could be attributed to these regions serving as hubs for assimilating advanced foreign technologies (Table S1), potentially resulting in reduced urban energy consumption and lower intensities of the ArUHI effect. The appropriate climate characteristics in these regions may also reduce AH emissions in their urban areas. Figure 9 also underscores the substantial influence of economic zones on the variations in parameter S. As can be seen, the Eastern Coastal Economic Zone (ECEZ) stood out with the highest S value (1.27 ± 0.44), indicating a more accelerated decline in heat emissions along urban–rural gradients within this zone. In contrast, the Northwest Economic Zone (NWEZ) exhibited the lowest S value (0.58 ± 0.11), suggesting a more gradual decrease in heat emissions from urban to background regions in the NWEZ. This observation highlights the nuanced and diverse nature of heat emission patterns across different economic zones. In terms of the rural AHF (AHF_a), the ranking from high to low was as follows: the NCEZ (4.67 ± 0.23 W/m²), NEEZ (3.23 ± 0.42 W/m²), SCEZ (3.20 ± 0.10 W/m²), NWEZ (3.00 ± 0.27 W/m²), MYEEZ (2.90 ± 0.19 W/m²), ECEZ (2.24 ± 0.11 W/m²), MYAEZ (1.61 ± 0.12 W/m²), and MYAEZ (1.17 ± 0.10 W/m²). This ranking offers valuable insights into the diverse levels of AHF_a in rural areas across distinct economic zones, providing a nuanced understanding of anthropogenic heat contributions and regional developments (

Table 6. Parameters of the exponential model of AHF in cities of different economic zones.

Economic Zone	A	S	AHFa	R2
NEEZ	21.01 ± 0.97	0.88 ± 0.12	3.23 ± 0.42	0.97
ECEZ	12.94 ± 0.31	1.27 ± 0.04	2.24 ± 0.11	0.99
NCEZ	13.26 ± 0.55	0.92 ± 0.10	4.67 ± 0.23	0.98
SCEZ	8.35 ± 0.22	0.83 ± 0.07	3.20 ± 0.10	0.99
MYEEZ	11.66 ± 0.42	0.82 ± 0.10	2.90 ± 0.19	0.99
MYAEZ	8.31 ± 0.27	1.02 ± 0.07	7.17 ± 0.10	0.99
SWEZ	10.37 ± 0.33	1.06 ± 0.06	1.61 ± 0.12	0.99
NWEZ	16.00 ± 0.41	0.58 ± 0.11	3.00 ± 0.27	0.99

).

3.3. Spatiotemporal Variability in the ArUHII and ArUHIFP

In this section, we delve into the spatiotemporal dynamics of the ArUHII and ArUHIFP. The investigation shed light on how these metrics evolved over both space and time, offering insights into the changing landscape of the urban heat island phenomenon across the studied cities from 2000 to 2016. Figure 10 delineates the spatial heterogeneities in the average ArUHII and ArUHIFP from 2000 to 2016. Cities exhibiting higher ArUHII values were predominantly situated north of the Huai River, with a notable concentration in the northeastern provinces (Heilongjiang, Jilin, and Liaoning). Notably, among the 208 cities scrutinized, the 18 cities with the highest ArUHII values (e.g., >20 W/m²) were exclusively located in the northeast. This spatial pattern can likely be attributed to the lower winter temperatures in the northern regions, leading to increased heating demand in terms of both extent and intensity, a phenomenon that is consistent with the observations in [37] regarding heightened anthropogenic heat emissions during winter in the northern regions. Regarding the spatial distribution of the ArUHIFP, the analysis revealed that, for the majority of cities (86% or 179 out of 208), the ArUHIFP ranged from 1.5 to 5.5 times the footprint of the urban area. Cities with varying ArUHIFP values were widely dispersed across the country and did not demonstrate significant clustering.

Figure 11 shows the temporal variations in the ArUHII and ArUHIFP in the cities selected. As shown, in terms of the temporal evolution of the ArUHII, a significant increase in ArUHII was noted from 2012 to 2016. This upward trajectory could be attributed to two primary factors. Firstly, the ongoing process of urbanization had a notable impact on building forms, influencing the spatial distribution of AH emissions [46]. The rapid proliferation of urban structures may contribute to the intensification of the ArUHI phenomenon, especially considering the role of building envelopes and air conditioning systems, which have been shown to exacerbate the AHF in urban areas [51]. Secondly, urbanization has implications for AHEs linked to human metabolism [52]. The progress of urbanization has led to a substantial rural-to-urban migration, resulting in a significant increase in the proportion of the urban population. The larger urban population generates increased AH emissions associated with human metabolism, thereby further amplifying the intensity of the ArUHI effect. Concerning the temporal dynamics of the ArUHIFP, there were varying degrees of growth observed from 2000 to 2012, reaching a peak in 2012. However, from 2012 to 2016, the ArUHIFP displayed a declining trend. One plausible explanation for this trend could be attributed to the period from 2000 to 2012, during which many cities underwent rapid expansion. There was a significant influx of rural populations into both urban and background areas, leading to a continuous increase in the ArUHIFP. However, by 2012, urban development and infrastructure construction had largely taken shape, resulting in a downward trend in the ArUHIFP.

4. Discussion

4.1. The Existence of ArUHIs and Correlations with Land Uses and Building Form Parameters

Previous research has highlighted the influence of alterations in urban surface spatial morphology on the distribution of AH emissions [23,53]. This body of work suggests that modifications in the layout and structure of urban surfaces play a significant role in shaping patterns of AH emissions. Understanding the relationship between urban surface spatial morphology and AH emissions is crucial for effective urban planning and mitigation strategies. By considering the impact of urban design on heat emissions, cities can implement measures to enhance sustainability, mitigate heat-related challenges, and improve overall urban climate resilience.

In this study, our investigation focused on assessing the alterations in land use compositions and building forms along the urban-rural gradient and understanding their interactions with AHF. Figure 12 illustrates the changes in land use composition and building forms in correlation with the variations in AHF along urban–rural gradients in 2016. In addition, we calculated the coefficient of determination (R²) and pseudo-t-test values of AHF and its influencing factors using an ordinary least squares model (Figure S1A).

The proportion of rural residential areas (RA_P) and those with other types of construction (OC_P) displayed no discernible change patterns along urban–rural gradients, as depicted in Figure 12C. Notably, no statistical significance was found concerning the relationship between RA_P and AHF (Figure S1D, R² value = 0.03). This implies that rural areas contributed minimally to AH emissions. However, OC_P was positively correlated with AHF (R² value = 0.21 and p-value < 0.05), indicating that other construction areas increased AH emissions (Figure S1E). It should be noted that other construction land encompassed zones allocated for factories, mines, extensive industrial areas, oilfields, salt flats, and quarries, as well as transportation infrastructure, airports, and specialized land [54]. These regions feature high AH emissions due to their high energy consumption. Figure 12 confirms that building form parameters, including building coverage (BC), mean building height (MBH), and mean building volume (MBV), showed similar decreasing trends in AHF along urban–rural gradients. Additionally, BC, MBH, and MBV were significantly positively correlated with AHF, with R² values of 0.95, 0.99, and 0.94, respectively. Buildings typically house human activities, which involve energy consumption for heating, cooling, lighting, and various other purposes. The larger and more numerous the buildings in an area, the more energy that is consumed, leading to higher AHF.

4.2. The Influence of City Sizes and Economic Zones on the ArUHIFP

An intriguing inquiry pertains to how city size and economic zones might influence variations in the ArUHIFP. Figure 13A offers compelling insights into this matter, revealing a noteworthy connection between city size and the ArUHIFP. Specifically, it indicates that larger cities tended to have lower ArUHIFP values. Small cities (labeled as level 5) stood out with the highest average ArUHIFP values. These values were approximately 6.25 times those of their corresponding urban areas. In contrast, super cities (Level 1) featured the lowest average ArUHIFP values, which were approximately 3.42 times those of their corresponding urban areas. This phenomenon can be attributed to two primary reasons. Firstly, larger cities exhibit higher population numbers and densities in urban areas. Research has shown a positive correlation between population density and AHEs [22]. In larger cities, AHEs tend to concentrate toward the urban center due to the higher human activity and energy consumption. This concentration leads to a smaller impact range of the ArUHI effect. Figure 13B illustrates a notable disparity in the ArUHIFP among cities situated in various economic zones. Cities located in the NWEZ exhibited the highest ArUHIFP values (5.08 times), while those in the NCEZ (3.59 times) and the NEEZ (3.71 times) had relatively lower ArUHIFP values. The observed phenomenon may be attributed to industrial development. The NWEZ, which is positioned in inland areas with constrained transportation access and lower industrial output [55], displayed a relatively dispersed distribution of AHF, contributing to its higher ArUHIFP. In contrast, the NEEZ and the NCEZ, encountering lower winter temperatures, featured densely populated city centers with increased heating demands [56]. Consequently, these cities produced a noteworthy amount of anthropogenic heat emissions from residential buildings. These observations underscore the intricate dynamics of urban climatology. The interplay of city size, economic zones, and the ArUHIFP is a multifaceted phenomenon with crucial implications for urban planning and strategies aimed at mitigating the urban heat island effect.

4.3. The Significance of the ArUHII and ArUHIFP

In this study, we proposed a new conception of ArUHIs and quantified the impact of anthropogenic heat in urban areas using two crucial metrics, namely the ArUHII and ArUHIFP. This provides a clear understanding of how human activities contribute to local temperature variations. Undoubtedly, this study contributes to UHI research by focusing on the anthropogenic component, allowing for a more nuanced understanding of the urban warming phenomenon. It distinguishes between natural and human-induced factors. Wu [57] provided a quantitative depiction of the contributions of AHEs to the SUHI through the urban anthropogenic heat index (UAHI), which was calculated based on distinctions in the land surface temperature and albedo between urban and rural areas. In comparison with the UAHI, the distinctive advantage of the ArUHII and ArUHIFP lies in their capacity to independently characterize AHEs without inherently reflecting their contributions to the SUHI. This unique feature allows for an exploration of the influence of the ArUHII and its extent on the SUHI, as calculated using various methodologies. Future investigations could delve into unraveling the specific contributions of the ArUHII and ArUHIFP to the intensity and footprint of the SUHI.

In addition, through the analysis of the ArUHII and ArUHIFP, this research unveiled the intricate spatial and temporal patterns of anthropogenic heat, offering crucial insights into how urban development and human activities impact climate dynamics. The exploration of the influence of city size and economic zones on the ArUHIFP provides valuable perspectives on how diverse urban structures and economic activities contribute to heat variations. This understanding of the spatial distribution and intensity of anthropogenic heat serves as a valuable tool for urban planners. For example, this research has the potential to guide strategies for sustainable urban development, aiding in the mitigation of challenges related to heat. In particular, the insights gleaned from this study contribute to the development of climate-resilient cities. Urban planners and policymakers can leverage this information to design cities that harmonize economic activities with environmental considerations.

This study has a limitation in that it exclusively focused on yearly ArUHI analyses. Alternative methods, such as that proposed by Dong [52], involve estimating the monthly AHF using a temperature-based weighted function method. Incorporating monthly AHF data facilitates the examination of distinctive ArUHI characteristics in various seasons, including both heating and non-heating periods. This approach enables more nuanced insights into the impact of human activities on the SUHI and supports the strategic planning of energy sources to mitigate these effects in different seasons. Future studies could, thus, consider seasonal and diurnal ArUHI explorations for a comprehensive understanding.

5. Conclusions

In this study, an examination was conducted on the existence of the ArUHI effect in 208 cities across China from 2000 to 2016. We delved into the spatiotemporal distribution patterns of the ArUHI effect in cities of different levels and those situated in various economic zones. Additionally, we investigated how building forms and land cover components interact to influence the AHF value along urban–rural gradients. From 2000 to 2016, China experienced a pronounced ArUHI effect. A significant exponential decay pattern in ΔAHF_i was observed. The mean ArUHII was 12.44 ± 0.43 W/m². For most cities, the ArUHIFP was 1.5 to 5.5 times greater than the urban area. A marked increase in the ArUHII occurred from 2012 to 2016, while the ArUHIFP peaked in 2012, declining thereafter until 2016. Studying urban–rural gradients in terms of AHF aids in understanding its distribution and tracking annual changes, supporting energy planning and layout measures. Economic zones and city sizes had an impact on the spatial distribution of the ArUHIFP. The highest value for the modeled ArUHII was observed in small cities, which also stood out with the highest average ArUHIFP values. These values were approximately 6.25 times those of their corresponding urban areas. In contrast, super cities featured the lowest average ArUHIFP values, which were approximately 3.42 times those of their corresponding urban areas. The NEEZ displayed the highest modeled ArUHII. Cities located in the NWEZ exhibited the highest ArUHIFP values, while those in the NCEZ and the NEEZ had relatively lower ArUHIFP values. Cities located in the NWEZ exhibited the highest ArUHIFP values, while those in the NCEZ and the NEEZ had relatively lower ArUHIFP values. Studying the differences in ArUHIs between cities of different sizes and economic zones is beneficial for revealing the impacts of demographic and economic factors on ArUHIs. This will aid in designing a reasonable energy layout and reducing the differences of AHF along urban–rural gradients. The building form had an impact on the distribution of artificial heat in suburban areas: BC, MBH, and MBV were significantly positively correlated with AHF, with R² values of 0.95, 0.99, and 0.94, respectively. This suggests that the larger and more numerous the buildings in an area are, the more energy that is consumed, leading to higher AHF. Therefore, a reasonable layout of urban buildings should be adopted to reduce the urban–rural differences in AHF.

Although this study explored changes in the ArUHI effect according to the intensity and footprint through establishing an exponential model of anthropogenic heat along urban–rural gradients, there are still some limitations and unresolved issues that provide potential directions for future research. Firstly, by combining the daily and hourly estimation results of AHF, we can explore the finer variations in the ArUHI effect. Secondly, further exploring the contribution of heating and industrial development to the ArUHII is beneficial for deepening the understanding of why the ArUHII is higher in the north than in the south. Thirdly, using Gaussian surfaces to simulate the variations in artificial heat along urban–rural gradients is beneficial for enhancing the certainty of ArUHII and ArUHIFP calculations.