Annual and Seasonal Dynamics of CO 2 Emissions in Major Cities of China (2019–2022)

: To control the growth of CO 2 emissions and achieve the goal of carbon peaking, this study carried out a detailed spatio-temporal analysis of carbon emissions in major cities of China on a city-wide and seasonal scale, used carbon emissions as an indicator to explore the impact of COVID-19 on human activities, and thereby studied the urban resilience of different cities. Our research re-vealed that (i) the seasonal patterns of CO2 emissions in major cities of China could be divided into four types: Long High, Summer High, Winter High, and Fluctuations, which was highly related to the power and industrial sectors. (ii) The annual trends, which were strongly affected by the pan-demic, could be divided into four types: Little Impact, First Impact, Second Impact, and Both Impact. (iii) The recovery speed of CO2 emissions reflected urban resilience. Cities with higher levels of de-velopment had a stronger resistance to the pandemic, but a slower recovery speed. Studying the changes in CO2 emissions and their causes can help to make timely policy adjustments during the economic recovery period after the end of the pandemic, provide more references to urban resilience construction, and provide experience for future responses to large-scale emergencies.


Introduction
Global warming is broadly agreed to be mainly caused by human activity, and is currently an important issue facing sustainable human development [1].Evidence shows that the heat-trapping gases released by burning fossil fuels, referred to as greenhouse gases, are responsible for causing global warming [2], which is related to multiple sectors such as power, industry, transportation, and residential consumption [3].Thus, greenhouse gas emissions (GHG), such as carbon dioxide (CO 2 ), are critical for understanding and addressing the climate crisis [4].In this context, many countries have set clear targets to reduce their CO 2 emissions in order to face the challenges of climate change.China promises to peak carbon by 2030 and achieve carbon neutrality by 2060 [5].Correspondingly, despite facing many difficulties, China is taking a series of measures to achieve this goal, making a positive contribution to promoting global governance in response to climate change.These measures include developing and widely promoting low-carbon technologies; developing a new energy industry to promote energy transformation; and implementing comprehensive low-carbon policies and laws [6][7][8].So far, CO 2 emissions per unit of energy consumption, energy consumption per unit of gross domestic product (GDP), and CO 2 emissions per unit of GDP all present downward trends in China, but China is yet to reach its peak in CO 2 emissions per capita, which is the key factor of CO 2 emission measurements [9].Therefore, accurately characterizing the dynamics of CO 2 emissions (before the carbon peak is attained as a turning point) is critical for China to formulate and implement corresponding policies in order to fulfill its commitment of reducing CO 2 emissions.
At present, China's total CO 2 emissions rank first in the world, but their growth rate is constantly decreasing and has been effectively controlled.In addition, China's CO 2 emissions are unevenly distributed in space, with significant differences among different sectors [10].According to studies in recent years, there was an initial dip in CO 2 emissions due to the COVID-19 pandemic, followed by varying degrees of rebounds [11].Studying the dynamics of CO 2 emissions can observe the impact of the COVID-19 pandemic on human socio-economic activities and reveal patterns of urban resilience occurring during the pandemic period (2019-2022).Urban resilience is the capacity of a city to rebound from major and minor disasters [12], which shows whether a city can cope with risks or events and is an important goal in SDG 11 (making cities and human settlements inclusive, safe, resilient, and sustainable).According to the recovery of CO 2 emissions, we can adjust the relevant policies in the post-pandemic period in a timely manner and study the laws of urban resilience through the recovery rate of CO 2 emissions.This can also provide experience for dealing with disturbances in the future and improve the adaptability of cities to risks [13].
As the world's current major economic power and emitter of CO 2 , there are abundant CO 2 emissions studies focusing on China [14].However, due to the lack of highspatiotemporal-accuracy CO 2 emission data, studies of CO 2 dynamics are mostly analyzed on an annual basis [15], lacking a high temporal accuracy.Moreover, most studies choose the whole country or certain provinces as their research object, but less focus on the urban scale [16], which lacks fine-scale conclusions for the guidance of emission reduction.Furthermore, the contribution of each CO 2 emission sector is worth considering, but the specific classification of various CO 2 emission sources is needed now [17].According to relevant research, the influencing factors of CO 2 emissions are mainly divided into three categories: technology, scale, and structure, which are also the main ideas involved in implementing CO 2 reduction policies [18][19][20].In recent years, some studies have noticed the impact of seasonal changes on carbon emissions [21], which provides new possible methods for emission reduction, but there is a lack of more extensive and detailed research.The development of CO 2 emission datasets brings new possibilities for finer-scale studies.Fine temporal-spatial CO 2 emission management becomes the inevitable trend, keeping the goals of carbon peak and carbon neutrality.More precision in terms of the accuracy of timing allows us to grasp the dynamic changes occurring due to the impact of major social events on human activities in a more timely and accurate manner.The precise spatial scale allows us to focus on typical research areas.In addition, detailed sources of CO 2 emissions make it easier for us to study the causes of CO 2 emission changes.
In order to summarize the patterns of CO 2 emission changes occurring during the pandemic period and provide new ideas for the implementation of future emission reduction policies, this study separated the seasonal and annual trends of CO 2 emissions by time series decomposition.Taking COVID-19 from 2019 to 2022 as an example, we analyzed the relationship between the impact of major social events and changes in CO 2 emissions, as well as the relationship between seasons and CO 2 emissions during this period, which is lacking in previous relevant studies.This study fills the gap in related research by conducting more refined studies (urban and monthly scales), comparing and analyzing CO 2 emissions from different sectors and cities, and studying the seasonal trends of CO 2 emissions.In addition, we studied the recovery rates of CO 2 emissions in different cities after COVID-19 (2019.1-2022.12)and analyzed their patterns.

Data and Methods
This study contains three major steps (Figure 1): (1) Data collection and the preprocession of CO 2 emissions and relevant variables.

Raw Data Collection and Pre-Procession
Thirty-one major cities in China with intensive human activities were selected as the research areas to be analyzed, which have high carbon footprints and become the main locations for controlling CO2 emissions.The research period lasted from January 2019 to December 2022, which includes the period from the outbreak to the end of the pandemic.
The raw data for this article came from Global gRidded dAily CO2 Emissions Dataset (GRACED) [22][23][24][25], which provides near-real-time global gridded daily CO2 emissions data from fossil fuel and cement production with a global spatial resolution of 0.1° by 0.1° and a temporal resolution of 1 day, and the measure in kilograms of CO2 per hour (kgC/h).The data satisfy the demand for high-quality, high-precision, and near-real-time data on CO2 emissions to support global emissions monitoring across various spatial scales.
We calculated the monthly CO2 emission data for provincial capital cities in China.For each city, we obtained the total CO2 emission data, as well as specific CO2 emission data for each of the four sectors: Power, Industry, Residential, and Ground Transportation.GRACED also provided CO2 emission data from three sectors: International Aviation, International Shipping, and Domestic Aviation, which were not studied in detail in this work, due to their small contributions and their very low or even zero CO2 emissions.A data summary is provided in Table 1 to provide readers with a general understanding of the data.
Other relevant data, such as population and gross domestic product (GDP).etc., were obtained from the China Statistical Yearbook to serve as a reference for exploring the causes of CO2 emission changes in the subsequent experiments [26].

Raw Data Collection and Pre-Procession
Thirty-one major cities in China with intensive human activities were selected as the research areas to be analyzed, which have high carbon footprints and become the main locations for controlling CO 2 emissions.The research period lasted from January 2019 to December 2022, which includes the period from the outbreak to the end of the pandemic.
The raw data for this article came from Global gRidded dAily CO 2 Emissions Dataset (GRACED) [22][23][24][25], which provides near-real-time global gridded daily CO 2 emissions data from fossil fuel and cement production with a global spatial resolution of 0.1 • by 0.1 • and a temporal resolution of 1 day, and the measure in kilograms of CO 2 per hour (kgC/h).The data satisfy the demand for high-quality, high-precision, and near-real-time data on CO 2 emissions to support global emissions monitoring across various spatial scales.
We calculated the monthly CO 2 emission data for provincial capital cities in China.For each city, we obtained the total CO 2 emission data, as well as specific CO 2 emission data for each of the four sectors: Power, Industry, Residential, and Ground Transportation.GRACED also provided CO 2 emission data from three sectors: International Aviation, International Shipping, and Domestic Aviation, which were not studied in detail in this work, due to their small contributions and their very low or even zero CO 2 emissions.A data summary is provided in Table 1 to provide readers with a general understanding of the data.Other relevant data, such as population and gross domestic product (GDP).etc., were obtained from the China Statistical Yearbook to serve as a reference for exploring the causes of CO 2 emission changes in the subsequent experiments [26].

Time Series Decomposition of CO 2 Emissions Trend
A time series Y t with seasonal factors can be decomposed into the following three components: (1) a trend component (T t ) representing long-term changes; (2) a seasonal component (S t ) which reveals periodic changes over time; and (3) an error component (E t ) as a mutation caused by errors [27].Thus, Y t can be decomposed by an additive model.
This seasonal trend decomposition based on LOESS (STL) is a time series data analysis method, which makes the curve smoother and employs locally weighted regression models to decompose a time series into trend, seasonal, and remaining contributions.STL consists of an inner and an outer loop; the inner loop calculates the trend and seasonal components and the outer loop provides the robustness weights for the next inner loop.The process of STL is shown in Figure 2.

Time Series Decomposition of CO2 Emissions Trend
A time series Yt with seasonal factors can be decomposed into the following three components: (1) a trend component (Tt) representing long-term changes; (2) a seasonal component (St) which reveals periodic changes over time; and (3) an error component (Et) as a mutation caused by errors [27].Thus, Yt can be decomposed by an additive model.
This seasonal trend decomposition based on LOESS (STL) is a time series data analysis method, which makes the curve smoother and employs locally weighted regression models to decompose a time series into trend, seasonal, and remaining contributions.STL consists of an inner and an outer loop; the inner loop calculates the trend and seasonal components and the outer loop provides the robustness weights for the next inner loop.The process of STL is shown in Figure 2. STL is a simple and easy-to-use method for obtaining robust trend and seasonal components, and has a better applicability for complex long-term time series.The advantages of STL is that it can identify seasonal components changing over time, it is responsive to nonlinear trends, and it is robust in the presence of outliers [28].The STL procedure was developed and implemented in the R software (version 4.2.1), and decomposed the CO2 emissions into seasonal and annual patterns.Further, we analyzed the types and causes of the seasonal and annual patterns from the perspectives of CO2 emission sectors and urban development, respectively.STL is a simple and easy-to-use method for obtaining robust trend and seasonal components, and has a better applicability for complex long-term time series.The advantages of STL is that it can identify seasonal components changing over time, it is responsive to nonlinear trends, and it is robust in the presence of outliers [28].The STL procedure was developed and implemented in the R software (version 4.2.1), and decomposed the CO 2 emissions into seasonal and annual patterns.Further, we analyzed the types and causes of the seasonal and annual patterns from the perspectives of CO 2 emission sectors and urban development, respectively.

Evaluation of Urban Resilience
A city is a huge artificial ecosystem in the sense that we can deal with the city problem as an ecological entity.We found that when cities were facing the impact of the pandemic, their human activity levels reflected in CO 2 emissions exhibited different fluctuations, which were related to their different levels of development.In this study, we used the per capita GDP of 2019 to represent the development levels of different cities during normal periods.Some cities, which revealed strong resistance to external influences and quickly recovered after being affected, we identified as urban resilient.Due to the fact that CO 2 emissions levels can reflect the strength of human activity levels, the recovery rate of CO 2 emissions can reflect the recovery rate of human activity levels, thereby reflecting the resistance of different cities (Equation ( 2)).Note that, in order to quantitatively compare cities with different levels of CO 2 emissions, their variations in CO 2 emission levels were measured by the ratio of the increase in CO 2 emissions to the initial value per time interval of the recovery period, with the recovery rate of CO 2 emissions V, the total increase in CO 2 emissions during the recovery period is represented by ∆E.E 0 is the initial value of CO 2 emissions (the average CO 2 emissions in the first half of 2019, which represents the normal CO 2 emissions without experiencing the pandemic); T star denotes the time when CO 2 emissions begin to show a long-term upward trend; and T end defines the time when CO 2 emissions do not show an upward trend or the monthly increase is very small.Furthermore, we used per capita GDP as an indicator to represent the development level of the city to investigate the relationship between the development level of the city and its own recovery rate, and then reveal the patterns of urban resilience.

Results
We decomposed the CO 2 emission variation patterns into seasonal and annual variations by STL, and then classified and summarized the results.In Section 3.1, we discuss the seasonal patterns of total CO 2 emissions and further investigate the seasonal patterns of the four specific CO 2 emission sectors, power, industry, residential, and others.As the power and industry sectors were the main sources of CO 2 emissions, we focused on their discussion.In Section 3.2, we analyze the annual patterns of total CO 2 emissions divided into four types.Then, we choose a city as an example for more detailed analysis.In Section 3.3, we calculate the recovery rate of the city and plot it.

Seasonal Patterns of CO 2 Emissions
The seasonal patterns of the total CO 2 emissions in major cities in China can be generally classified into the following four categories in Figure 3.The number on the vertical axes in Figure 3 represents the proportion of seasonal components of CO 2 emissions relative to the initial value.The distribution of these patterns is shown in Figure 4.More detailed data can be found in Table 2. Power and Industry, respectively, represent the proportion of CO 2 emissions from the power sector and industry sector in the total CO 2 emissions.Type represents the classification of the seasonal components of CO 2 emissions in cities.No data represents missing relevant raw data.
Long-term High: CO 2 emissions have a low peak around February and a high to medium level during other time periods, such as Shanghai.
Summer High: CO 2 emissions have a high peak during summer and a low level during other time periods, such as Beijing.
Winter High: CO 2 emissions have a high peak during winter and a low level during other time periods, such as Guangzhou.
Fluctuations: CO 2 emissions fluctuate frequently throughout the year, such as Hohhot.
Considering the sectors of CO 2 emissions, we found that CO 2 emissions from the same sector in different cities had similar patterns.The power sector peaked in summer, in winter, or both, and often fluctuated in the other months.For the industry sector, all cities' CO 2 emissions revealed a low peak around February, and most of them stayed at a high level in the other months.Note that all cities' CO 2 emissions from the residential sector showed the same pattern: peaking around January and remaining at low levels at all other times.Due to the significant impact of the pandemic on CO 2 emissions from the transportation sector, the data could not be decomposed into a trend component and seasonal component.But we can find patterns by observing the original data in Figure 5: during the pandemic, there were two time periods (January to October 2021 and March to December 2022) during which CO 2 emissions were at very low levels, while CO 2 emissions fluctuated greatly at other times.The other sectors were not studied in detail, due to their small contributions and the very low or even zero CO 2 emissions.

Annual Patterns of CO2 Emissions
In our study, city CO2 emissions showed an upward trend most of the time.However, almost all cities fluctuated during specific time periods, which coincided highly with the outbreaks of the pandemic.Based on these specific fluctuations, we divided the annual

Annual Patterns of CO 2 Emissions
In our study, city CO 2 emissions showed an upward trend most of the time.However, almost all cities fluctuated during specific time periods, which coincided highly with the outbreaks of the pandemic.Based on these specific fluctuations, we divided the annual patterns of total CO 2 emissions into the following four types in Figure 6.First and Second represent the time of the first and second outbreaks of the pandemic.Little Impact: CO 2 emissions showed a stable upward trend or little change, such as in Hohhot.
First Impact: CO 2 emissions showed a low peak in early 2020, followed by a slow upward trend, such as in Fuzhou.
Second Impact: CO 2 emissions showed a low peak at the end of 2021, such as in Beijing.Both Impacts: CO 2 emissions showed low peaks in early 2020 and at the end of 2021, such as in Changchun.
For each type, we selected a typical city to display the STL results (Figure 7).The number on the vertical axes in Figure 7 represents the proportion of seasonal components of CO 2 emissions relative to the initial value.To more specifically describe the relationship between the impact of the pandemic and the changes in CO 2 emissions, we selected Changchun as an example.During the period from January 2020 to April 2020, Changchun City had a cumulative increase of 49 new infected individuals, which led to the implementation of strict control measures and a reduction in human activities, so, accordingly, CO 2 emissions decreased during this period.For a long time afterwards, there were no new infections and CO 2 emissions increased during this period.At the beginning of 2022, there was a large number of new infections, while CO 2 emissions decreased again at the same time.Due to China's implementation of the policy of liberalization at the end of the pandemic, this subsequent part was not within the scope of this study.It can be seen that the CO 2 emissions of Changchun significantly decreased during the two time periods of sudden increases in the number of infected individuals, since the levels of human activities were significantly affected by the pandemic at those times.Almost all cities showed a similar pattern: when the impact of the pandemic was severe or aggravated, CO 2 emissions decreased, and vice versa, when the impact was mild, CO 2 emissions rose.
the same time.Due to China's implementation of the policy of liberalization at the end of the pandemic, this subsequent part was not within the scope of this study.It can be seen that the CO2 emissions of Changchun significantly decreased during the two time periods of sudden increases in the number of infected individuals, since the levels of human activities were significantly affected by the pandemic at those times.Almost all cities showed a similar pattern: when the impact of the pandemic was severe or aggravated, CO2 emissions decreased, and vice versa, when the impact was mild, CO2 emissions rose.

Urban Resilience
We calculated the rates of increase in CO2 emissions in cities and indicated which post-pandemic recovery rate they belonged to; the results are presented in Table 3.The Recovery Speed I and Recovery Speed II represent the recovery speeds of CO2 emissions in cities after experiencing the impact of the first and the second outbreak of the pandemic.Recovery Value I and Recovery Value II represent the proportion of CO2 emissions increasing relative to the initial value after experiencing the impacts of the first and second pandemic outbreaks.No data means that the city did not experience a decrease in CO2 emissions due to the pandemic during this period, or it did not recover until the end of the research period.Then, we plot the calculation results in Figure 8. Blue means the recovery rate after the first outbreak of the pandemic, while red means the second.

Urban Resilience
We calculated the rates of increase in 2 emissions in cities and indicated which postpandemic recovery rate they belonged to; the results are presented in Table 3.The Recovery Speed I and Recovery Speed II represent the recovery speeds of CO 2 emissions in cities after experiencing the impact of the first and the second outbreak of the pandemic.Recovery Value I and Recovery Value II represent the proportion of CO 2 emissions increasing relative to the initial value after experiencing the impacts of the first and second pandemic outbreaks.No data means that the city did not experience a decrease in CO 2 emissions due to the pandemic during this period, or it did not recover until the end of the research period.Then, we plot the calculation results in Figure 8. Blue means the recovery rate after the first outbreak of the pandemic, while red means the second.

Seasonal Patterns and the Causes
The data sources used in this study, which provide the CO2 emissions, inclu sources (such as industrial electricity and building electricity).For the power sec half of the cities saw seasonal fluctuations in CO2 emissions, while the others pea during summer or winter.In most cities, the CO2 emissions from the industry se

Seasonal Patterns and the Causes
The data sources used in this study, which provide the CO 2 emissions, include many sources (such as industrial electricity and building electricity).For the power sector, about half of the cities saw seasonal fluctuations in CO 2 emissions, while the others peaked only during summer or winter.In most cities, the CO 2 emissions from the industry sector were at high levels throughout the year, but lower peaks were experienced during the Spring Festival (traditional Chinese festival during which many companies, factories, enterprises, and schools give employees and students a long holiday).However, different from the industry sector, due to the impact of the pandemic, most people choose to celebrate the Spring Festival at home, resulting in an increase in CO 2 emissions from the residential sector during this period.As for CO 2 emissions from the transportation sector, due to the uncertainty of people's travel willingness during the pandemic and other possible factors, various transportation modes, such as express delivery, saw fluctuation.When implementing travel restriction policies, it will drop to a very low or even close to zero level.
For all cities in this study, the CO 2 emissions of the power and industrial sectors accounted for at least 75% of the total CO 2 emissions; that is, these two sectors are the main contributors to changes in the total CO 2 emissions.In general, cities with the industry sector as their main source of CO 2 emissions reveal more obvious seasonal patterns of CO 2 emissions, while cities with the power sector dominating as the main source of CO 2 emissions are more prone to fluctuations in their seasonal CO 2 emission patterns.In addition, we also found that most cities with similar seasonal patterns in their CO 2 emissions are geographically clustered (see Figure 3).As for the residential and transportation sectors, both have undergone significant changes at certain times, but CO 2 emissions from the industrial and electricity sectors have not shown significant fluctuations during these times.This indicates that emissions reduction in the power and industrial sectors is important for achieving carbon peaking, while low-carbon transportation and residential energy saving cannot make sufficient contributions to this.Although seasonal components account for a relatively small proportion of CO 2 emissions, they are more prone to significant changes in the short-term time scale.Therefore, in order to observe the changes in CO 2 emissions more accurately, it is necessary to remove seasonal effects.This way, we can grasp the rising and falling trends of CO 2 emissions in a timely and accurate way, which is conducive to a fast-responding policy to achieve carbon peaking.

Resilience of Annual Patterns in CO 2 Emissions
The impact of the pandemic on cities varied in degree and duration, and thus on human activities, which could be reflected through the changes in the CO 2 emission trends.At the same time, different cities had varying levels of resistance and resilience to the impact of the pandemic.We compared the changes in the number of infected individuals and the annual patterns in CO 2 emissions during the pandemic period, and obtained the relationship between the two: (i) When the impact of the pandemic did not exceed the city's resistance to the pandemic, CO 2 emissions showed an upward trend associated with the normal development of the economy and the acceleration of various production activities.
(ii) When the impact of the pandemic exceeded the city's resistance, CO 2 emissions rapidly decreased in the short-term time scale and then rose after the impact ended.
During the pandemic, the period when urban CO 2 emissions suddenly decreased is similar to the period when the number of infected people in the city suddenly increased significantly (which also indicates that the impact of the pandemic is increasing).This result indicates that it is feasible to measure the impact of large-scale emergencies on cities through real-time observations of changes in CO 2 emissions.When a city is affected by large-scale emergencies, we can react in a timely manner and respond accordingly.It can be seen that cities with a higher per capita GDP have a lower recovery rate.In fact, a higher level of development means that the internal structure of the city is more complex, which means that a small amount of impact can not only lead to the paralysis of the urban system, but also to a more difficult recovery after destruction.
In this sense, urban resilience means "resistance", "quick transformation", and "rapid recovery".But from the results, most cities have been more or less affected by the pandemic, indicating that the current construction of urban resilience is insufficient.Therefore, the construction of urban resilience is important and should consider different levels of urban development.The improvement in urban resilience can reduce the impact of large-scale emergencies such as COVID-19, and accelerate the recovery rate of cities after the city is affected.In addition, our study found that the CO 2 emission levels of most cities after the recovery period had exceeded the initial levels, with some even approaching a 20% increase.For cities affected by two outbreaks, we found that the recovery rate of CO 2 emissions had an unreasonably high value after the first impact ended.In addition, comparing the two trend lines in Figure 8, we found that the recovery rate of CO 2 emissions after the first impact of the pandemic was faster than the second.This means that the rising rate of CO 2 emissions during the period was not under control, which may be due to the lack of experience in responding to the pandemic.This also indicates that there are still many risks that need to be carefully addressed during the urban recovery period.Achieving a fast recovery speed of cities after disasters is important, but ensuring that recovery is not too excessive is also highly important.How to achieve these goals is what we need to focus on in the future.

Conclusions
The impact of the pandemic on human activities is analyzed accurately and in more detail, thereby studying the urban resilience of different cities.This included the seasonal and annual patterns of CO 2 emissions by STL, and urban resilience by calculating and comparing the recovery rates of CO 2 emissions in different cities.
Seasonal trends can be classified into four types, mainly generated by the power and industrial sectors, which are also the focus of emission reduction.The seasonal portion of CO 2 emissions generally does not exceed 15% overall, but can change significantly in the short term, which means that we must consider it when observing the short-term trend of CO 2 emissions.When facing large-scale emergencies such as the pandemic, we need to consider the seasonal trends of human activities when evaluating the short-term impacts of the event.In addition, due to the significant differences in seasonal patterns between different seasons, implementing emission reduction policies tailored to the seasons can achieve better results.Due to the fact that cities with the same seasonal pattern are not geographically dispersed, it is feasible to implement this differentiated policy nationwide.However, this study still has the following limitations: (1) It is difficult to explain seasonal trends related to the power sector due to the lack of more detailed classifications from the raw data.(2) There is no analysis of urban resistance to compare with urban resilience (existing research indicates an inverse relationship between the two).(3) Due to the significant impact of the pandemic on the transportation sector, time series decomposition cannot be performed.
The annual trend can also be classified into four types, which are related to the impact of the pandemic.During the pandemic, CO 2 emissions in many cities rose or recovered with normal economic development, and rapidly decreased if severely affected by the pandemic.Many cities are experiencing economic development or recovery, but uncontrolled CO 2 emissions are detrimental to achieving the goal of carbon peaking.In our study, most cities had already exceeded their initial CO 2 emission levels after rebounding.Especially after the end of the first pandemic outbreak, the rebound rate in some cities showed abnormally high values, and their CO 2 emissions eventually exceeded the initial value.When facing large-scale emergencies, apart from the degree of impact, we also need to consider factors such as urban development, economy, and population, etc., which can reflect the strength of the city's resistance to such events and play an important role in the achievement of SDG 11.During the recovery period after the impact of the event, we also need to achieve economic recovery while preventing excessive rebounds in CO 2 emissions.Especially in recent years, when SDG11 and the carbon peak target date are approaching, it is crucial to control the growth rate of CO 2 emissions and even achieve a reduction.In future research, we can construct an urban resilience evaluation system based on CO 2 emissions to achieve the goals of urban health assessment, sustainable development, and carbon peaking.

( 2 )
Time series decomposition of CO 2 emission trends into annual and seasonal patterns, and then an analysis of both patterns by types and causes.(3) Evaluation of urban resilience during the COVID-19 pandemic based on the annual patterns of CO 2 emissions.The following sub-sections detail the data and main methods.
ISPRS Int.J. Geo-Inf.2024, 13, x FOR PEER REVIEW 3 of 14 by types and causes.(3) Evaluation of urban resilience during the COVID-19 pandemic based on the annual patterns of CO2 emissions.The following sub-sections detail the data and main methods.

Figure 1 .
Figure 1.The process in this study.

Figure 1 .
Figure 1.The process in this study.

Figure 2 .
Figure 2. The process of STL.

Figure 2 .
Figure 2. The process of STL.

Figure 4 .
Figure 4. Major cities analyzed in this study and their seasonal patterns.

Figure 4 .
Figure 4. Major cities analyzed in this study and their seasonal patterns.

Figure 4 .
Figure 4. Major cities analyzed in this study and their seasonal patterns.

Figure 6 .
Figure 6.Conceptual charts of the four trend patterns of CO 2 emissions: (a) Little Impact, (b) First Impact, (c) Second Impact, and (d) Both Impacts.

Figure 7 .
Figure 7. Seasonal trend decomposition result of total CO2 emissions in (a) Hohhot, (b) Fuzhou, (c) Shanghai, and (d) Changchun.(The shaded areas represent that the city was in a period of recovery.)

Figure 7 .
Figure 7. Seasonal trend decomposition result of total CO 2 emissions in (a) Hohhot, (b) Fuzhou, (c) Shanghai, and (d) Changchun.(The shaded areas represent that the city was in a period of recovery.)

Figure 8 .
Figure 8. Scatter plot of urban per capita GDP and recovery rate of CO 2 emissions.

Table 1 .
Summary of total CO 2 emissions data for the studied cities (unit: kgC/h).

Table 1 .
Summary of total CO2 emissions data for the studied cities (unit: kgC/h).

Table 2 .
The composition and seasonal classification of CO2 emissions in cities.

Table 2 .
The composition and seasonal classification of CO2 emissions in cities.

Table 2 .
The composition and seasonal classification of CO 2 emissions in cities.

Table 3 .
The recovery rate of CO2 emissions in cities and other data.

Table 3 .
The recovery rate of CO 2 emissions in cities and other data.
Figure 8. Scatter plot of urban per capita GDP and recovery rate of CO2 emissions.