Cities are the major drivers of global greenhouse gas (GHG) emissions. The rapid growth of urban GHG emissions is the main reason for the increase of global GHG emissions [1
]. Cities account for less than 1% of the Earth’s landmass but concentrate more than half of the world’s population, consume approximately 75% of the world’s energy, and contribute 70%–80% of global GHGs [2
]. Global carbon emissions have been significantly affected by the increase in urban population with the high-speed development of global urbanization [4
]. Therefore, low-carbon economy development and city construction become inevitable actions for all countries in order to face the climate challenge and promote sustainable urban development. In June 2007, the Chinese government formally released “China’s National Climate Change Programme”, which clearly defined the principles, objectives, and measures to combat climate change. At the Copenhagen Climate Change Conference in December 2009, the Chinese government promised to reduce the carbon intensity (carbon emissions per unit gross domestic product [GDP]) to 40%–50% below 2005 levels by 2020.
The foundation of low carbon policy making is the accurate estimation of carbon emissions. The World Resources Institute divides carbon emissions into three categories: Scope 1 emission is the direct carbon emission within the city; Scope 2 refers to the indirect carbon emission caused by the purchase and usage of energy, steam, and heating/cooling; and Scope 3 emission indicates the carbon emission that occurs outside the city due to the activities within the city (Scope 3) [5
]. Under the leadership of the Intergovernmental Panel on Climate Change (IPCC), Center for Global Engagement (CGE), C40 cities, and other agencies and organizations, numerous cities in the world have established their own carbon inventories [6
]. Some major cities, such as New York, Tokyo, Chicago, and London, have evaluated carbon emissions on the basis of the carbon inventories [5
]. Because Scope 3 emission is difficult to define, most studies concentrate on Scope 1 and Scope 2 emissions. This study focuses on the Scope 1 emission.
It is also crucial to benchmark carbon emission estimates and spatial-temporal carbon emission analysis from different sources at multiple administrative levels, for example, the national, regional, metropolitan, and precinct scales. Precincts are a microcosm of cities. Carbon assessment at the precinct level can lead to mapping and reducing the carbon signature of the entire urban system, as well as support government policy-making. Precinct-scale assessment is expected to lead urban design and planning with carbon assessment and decision support at the early stage [13
]. Different areas play different roles in city development and make different contributions to carbon emissions. Urban areas tend to offer easier access to more energy-intensive fuels, as well as electricity, and also have higher population density and limits on space, concentrating emission sources [5
]. Low-density rural and suburban areas can be less susceptible to urban heat islands and traffic congestion, but also tend to suffer from greater efficiency losses associated with the longer transmission of electricity, more vehicle miles travelled, and the inability to rely on district heating and combined heat and power systems. Moreover, suburban and sprawling communities are more likely to have single-family detached houses and to live in bigger houses, both leading to higher energy use and grater carbon emissions [5
The mostly commonly used spatial carbon emission analysis can be generally categorized into top-down and bottom-up approaches. Population density, the Defense Meteorological Satellite Program’s Operational Linescan System (DMSP/OLS) nighttime light image, point emission sources and other data are widely used to estimate carbon emissions at different scales [17
]. Several carbon emission datasets are also available at different resolutions. The Emission Database for Global Atmospheric Research (EDGAR V4.3) [21
] and the Fossil Fuel Data Assimilation System (FFDAS) [22
] have CO2 emissions data at 0.1° × 0.1° (roughly 10 km × 10 km) spatial resolution. Based on the first China pollution source census, China high resolution emission gridded data (CHRED) provided 10 km × 10 km spatial resolution of CO2 emissions estimation of China with data from 1.58 million industrial enterprises [24
Although much work has been done in this field, there are still several barriers hindering the uptake of multiple level evaluation, and these barriers include system boundary definition, data availability, dynamic inter-building effects contributed by urban morphology, uncertain preferences for occupants’ life-style, integrated modelling and the less standardized production process of precinct objects caused by their unique characteristics [25
]. Moreover, the literature research and datasets are more suitable for large scale carbon emission evaluation, such as at the regional and provincial scale, and have difficulties in reflecting the spatial pattern of carbon emission in cities.
In this paper, we first introduced the study area and data, especially the geographical conditions census data. Then, we introduced the carbon emissions disaggregation models for different sectors, including industry, agriculture, residential living, waste disposal and transportation. Land coverage, geo-graphical units, thematic data and statistical data from the census of national geographical condition data were adopted to disaggregate the carbon emissions at street scale, which is the finest administrative level in China. We analyzed characteristics of carbon emissions in different areas and sectors in Wuhan. We also discussed the spatial correlation coefficients between economical-social factors (gross domestic product, population density, road density, and land-use structure) and carbon emissions in the metropolitan development area. Finally, we compared our data and methods with other studies.
4. Results and Discussions
4.1. Carbon Emission Characteristics in Wuhan
According to our results, in 2015, the total carbon emission in Wuhan is 120 million tons, the carbon emission per capita is 11.3 tons, and the average intensity of carbon emission is 14,000 tons/km2
. Industrial, transportation, residential living, and agriculture emissions are the main sources of carbon emissions, with proportions of 87.58%, 5.75%, 5.63%, and 0.18%, respectively (Figure 3
). The study from the World Bank shows that the main carbon emission source in China’s big cities is industrial emission, whereas the main carbon emission sources in developed countries are energy supply, transportation, and architecture. Wuhan’s carbon emission has typical Chinese characteristics.
and Figure 5
shows the spatial distribution of carbon emissions in Wuhan. The total carbon emission, carbon emission intensity, and carbon emission per capita at district and street scale are illustrated from top to bottom and from left to right.
At the district scale, the total carbon emission, carbon emission intensity, and carbon emission per capita show different patterns. The gross domestic product (GDP) of the secondary industry in Hanyang District is 64.19 billion yuan in 2015, accounting for 18.90% of Wuhan’s GDP. The secondary industry GDP in Dongxihu District is 47.694 billion yuan, accounting for 14.04% of the city’s GDP. The concentration of large enterprise, large energy consumptions of these two districts result in enormous carbon emissions. ELESZ has the least carbon emissions, followed by WLHDZ and Jianghan District, in which residential living is the dominant carbon emission type. Carbon emission intensities in Hanyang, Qiaokou, and Jianghan Districts are considerably larger than those in other districts. Although the carbon emission of Jianghan and Qiaokou Districts ranked 15th and 11th among 17 districts, their carbon emissions intensity are the largest, with more than 100,000 tons/km2. The carbon emission intensities of ELHDZ and Huangpi District are the lowest, less than 5000 tons/km2. The emission per capita in WCIP, which is 138.68 tons/km2, is significantly higher than those in other districts. The emissions per capita of Hanyang and Dongxihu Districts are approximately 30 tons/person, which ranks second. Other districts’ carbon emissions per capita are less than 25 tons/person, in which the least is ELHDZ with carbon emissions of only 2.52 tons/person.
At the street scale, the spatial distribution of carbon emission in the city can be reflected more intuitively. Generally speaking, streets of Qingshan District and WCIP have significantly higher emissions than those of other districts. Carbon emissions in Bajifu Street and Wuhan Iron and Steel Company are significantly larger than those of other streets. Most of the carbon emissions are concentrated in the functional areas between the second and third belt highways. The emission intensity shows a decreasing trend from the central to the rural area, while the carbon emission per capita presents a low–high–low trend from the city center to the rural area. Many industry concentrated streets often with lower population, making their emissions per capita significantly higher than other streets. The carbon emission per capita of the central urban area is considerably lower than those of other regions due to its high population concentration.
4.2. Carbon Emission Characteristics of Different Regions
To reflect carbon emission characteristics of different agglomerations, this study mainly analyzed carbon emissions and social-economic indicators in the central city, functional area, new urban area, built-up area, and metropolitan development area (Table 3
The city centre of Wuhan contains 57.51% of the city’s population within 7.48% of the city’s area. Due to the high population concentration, residential land area is significantly higher than other parts, which accounts for 55.28% of the total residential land. Residential energy consumptions and waste disposal amounts are also large. Carbon emissions account for 43% and 41.70% of the total emissions. The average road density is 3.28 km/km2. The city road densities in Jianghan and Qiaokou Districts, 6.20 and 5.12 km/km2, are considerably larger than those in other districts in the central area. The high-density road network, population aggregation, and heavy traffic flow make transportation carbon emissions in the city center account for 43% of the city’s total carbon emission. The industrial carbon emission patterns are relatively different, due to various development strategies. For example, Hanyang District has priority in manufacturing, and Qingshan District is dominated by Wuhan Iron and Steel Company, while Jianghan District is dominated by modern business industry. The industrial carbon emissions in Hanyang District are 7.6 times that in Jianghan District and 3.13 times that in Qingshan District.
The new urban area is larger than the central area and the population is relatively scattered. Carbon emissions are still dominated by industrial carbon emissions, accounting for 88.49% of all emissions, followed by transportation and residential living carbon emissions. The mileage of roads accounts for 77.51% of the city’s road, which are mainly rural roads with light traffic pressure, but the transportation carbon emissions only account for 48.54% of total transportation carbon emissions. Further, 92.08% of agricultural land distributed in the new urban area and the new urban area contributes 92.65% of agricultural carbon emissions. Carbon emissions and social-economic indices of the functional area are also higher than those of the new urban area, but lower than those of the central city.
The built-up area has a higher population density, road network density, residential area ratio, and industrial land ratio than the central city. Carbon emissions in the built-up area are still dominated by industrial carbon emissions, but the total carbon emissions and carbon emissions per capita are significantly lower than other areas. The reason is that the built-up area is the construction-completed area with the most dense population and the highest floor area ratio. The industrial enterprises in this area are mainly light industry and service-oriented enterprises, which generally have less carbon emissions.
The metropolitan development area is the major construction area in Wuhan. This region contains 77.61% of the population, 94.75% of residential areas, 80.20% of industrial land, 49.47% of road network and contributes 86.67% of carbon emissions in Wuhan. The carbon emissions per capita is 12.63 tons, which is significantly higher than those of the city center and built-up areas. The allocation and distribution of industry in the metropolitan development area is the epitome of Wuhan’s present stage, and its carbon emissions can be regarded as the real carbon emission level of Wuhan.
4.3. Carbon Emission Characteristics in Different Sectors
We also discussed carbon emission characteristics in different sectors. Figure 6
shows carbon emissions and emission intensity of different sectors at street scale.
Industrial and transportation carbon emissions in Wuhan show a similar circle distribution trend from the center to the periphery. Industrial carbon emissions and intensity in the city center are generally low, and high carbon emissions are concentrated in the functional areas, especially areas between the second and third belt highways. On one hand, the reason lies in the urban development planning and functional orientation of each district. On the other hand, these regions have significant geographical location advantages, such as abundant raw materials and labor resources, and convenient transportation. Transportation carbon emission intensity in Hankou District is generally higher than that in Wuchang and Hanyang Districts, because the density of the urban road network is considerably higher. In addition, some areas in Qingshan District and WCIP have large transportation carbon emissions. The reason is that large enterprises, such as Wuhan Iron and Steel Company, have large transportation needs, resulting in high traffic carbon emissions. Residential living carbon emissions and waste disposal carbon emissions in Wuhan have a similar distribution tendency, which are both closely related to population distribution. The agricultural production carbon emission tends to decrease from the periphery to the city center. The areas with high carbon emission and emission intensity are mainly concentrated in the new urban area.
4.4. GW Spatial Correlation Analysis
In this section, we explore the relationships between carbon emissions and social-economic indicators via GW correlation coefficients with the GWmodel package in R. Summaries of GW correlation coefficients are demonstrated in Table 4
. Density plots are illustrated in Figure 7
Carbon emissions are closely correlated with GDP, industrial land area and population density. Correlations with road density and residential land area are relatively small. Figure 8
shows spatial distribution of GDP (Figure 8
a), GW correlation of GDP (Figure 8
b), industrial land area (Figure 8
c), GW correlation of industrial land area (Figure 8
d), population density (Figure 8
e) and GW correlation of population density (Figure 8
Carbon emissions in the metropolitan development area are positively correlated with GDP, which means the higher the GDP, the more carbon emissions. The correlation is distinctly high at the central area, and this means that carbon emissions in this region would increase significantly with the increase of GDP.
For the industrial land, since the industry sector is the major source of carbon emissions in Wuhan, carbon emissions are positively correlated with the industrial land area in most streets. Streets in the functional area generally have higher industrial land areas and carbon emissions. The increase of industrial land areas will also raise carbon emissions, but the degree of improvement is less than that in the central area.
For the population density, the GW correlation distribution shows very interesting trend. The population density in the central area is obviously higher than other regions. However, the more people gathered at the central area, the less carbon emissions in this region. The reason for this phenomenon is that increasing population will occupy industrial land, which will greatly reduce industrial carbon emission. For the periphery area of metropolitan development area, population density is positively correlated with carbon emissions. There is adequate unused but suitable land in this region, increasing population density would improve the land utilization rate, which would also increase carbon emissions.
4.5. Comparisons with Other Studies
The latest released EDGAR 4.3.3 gridded emissions cover 1970–2013 time series, FFDAS datasets cover 1997–2010 time series, and CHRED release carbon emission dataset in 2007 and 2012. Due to data unavailability, we can’t make quantitative comparisons. Therefore, we analyze differences in methods and data used in these datasets and our work.
Both EDGAR and FFDAS datasets use fossil fuel data from IEA statistics (http://www.iea.org/
) to evaluate fossil carbon emissions. CHRED evaluate fossil fuel carbon emissions using official statistical yearbooks, which is more accurate for China. Then geo-information data is used to disaggregate carbon emissions at different resolutions. EDGAR uses the location of energy and manufacturing facilities, road networks, shipping routes, human and animal population density and agricultural land use. FFDAS utilizes population, per capita GDP, energy intensity of the economy, carbon intensity of energy production, nightlights, and emissions from listed power stations as control variables for the application of spatial operators. CHRED uses industrial facilities, landfills and wastewater treatment plants.
All these datasets are derived using bottom-up methods and the accuracy is determined by the geo-information data. Due to the geo-information data availability, EDGAR, FFDAS and CHRED datasets are more suitable for large scale carbon evaluation, such as the national and provincial scale. For example, they omit traffic flow and traffic congestion in city. At the city and street scales, our method needs much finer geo-information data in order to better reflect carbon emission characteristics.
In this study, we construct a carbon inventory in Wuhan based on the IPCC inventory and evaluated carbon emissions in 2015. On the basis of the geographical national conditions data, carbon emission disaggregation models for different sectors were proposed. We analyzed carbon emission characteristics at different regions and in different sectors, especially the spatial correlations at the metropolitan development area. Comparisons were made to prove the feasibility and effectiveness of the geographical conditions data in the application of carbon emissions. With the continuous accumulation of geographical conditions monitoring data in the future, spatial-temporal changes in of carbon emissions can be recognized, which would help evaluate low-carbon city development. The main conclusions are as follows:
(1) Main carbon emissions in Wuhan are still from industrial production, whereas living and transportation carbon emissions account for a relatively small part, indicating a large gap with developed countries. The carbon emission per capita in Wuhan is 11.31 tons/person, which is considerably higher than the national average level. Therefore, industrial restructuring and energy consumption optimization will take time to be achieved.
(2) The carbon emission patterns in all regions of Wuhan vary significantly. Carbon emissions from residential living and transportation in the center area are palpably higher than those in other regions, while industrial carbon emissions in the new urban areas are significantly higher. Moreover, emissions from various sectors in the functional areas are relatively balanced. The metropolitan development area can better represent the carbon emission characteristics of Wuhan than other regions. The spatial correlation analysis show that carbon emissions are positively correlated with GDP and industrial land, and negatively correlated with population density in Wuhan.
Evaluation and disaggregation of carbon emissions at the city scale remain great challenges of significant interest in the carbon emissions field. Although some exploratory works have been conducted in this study, several problems need to be further addressed.
(1) The census of geographical conditions has a strict data collection standard. Some land coverage less than the minimum acquisition area indices is ignored, which might cause errors in the spatialization process.
(2) This study considers the Scope 1 emission but not the Scope 2 and Scope 3 emissions, and therefore may underestimate carbon emissions. For example, carbon emissions caused by cross-boundary transportation and outbound transportation are not included.