An improved high-resolution passenger vehicle emission inventory for China using ride-hailing big data

Li, Baojie; Shen, Zhihui; Li, Yan; Zhao, Yongqi; Gu, Wanglijin; Liu, Junjie; Yang, Yunkai; Zhang, Weimeng; Ma, Ziqian; Liao, Hong

doi:10.5194/egusphere-2025-5554

Preprints

https://doi.org/10.5194/egusphere-2025-5554

Preprints

28 Dec 2025

| 28 Dec 2025

An improved high-resolution passenger vehicle emission inventory for China using ride-hailing big data

Baojie Li, Zhihui Shen, Yan Li, Yongqi Zhao, Wanglijin Gu, Junjie Liu, Yunkai Yang, Weimeng Zhang, Ziqian Ma, and Hong Liao

Abstract. As the global automotive industry continues to grow rapidly, the increasing number of passenger vehicles has contributed to worsening air pollution. However, previous studies have insufficiently addressed nationwide hourly vehicle emissions. This study firstly utilized big data of ride-hailing services and traffic flow model to obtain nationwide hourly gridded speed and traffic volume. Then we established a high spatiotemporal resolution (0.01° × 0.01°; 1h) emission inventory by using multiple correction factors. The annual amount of CO, VOCs, NO_x, PM and NH₃ emitted from national passenger vehicles in 2019 were 4087.8, 1069.4, 211.7, 1.9, 77.5 kt, respectively. Despite occupying merely 0.8% of the national territory, urban areas generated 35.3% of the country's total vehicle emissions, due to high local traffic volumes and relatively low vehicle speeds. From a temporal perspective, passenger vehicle emissions exhibit significant holiday effect and weekend effect. In addition, hourly average emissions on workday exceeded those of weekend and holiday by 8% and 5% during the morning peak, with these differences increasing to 12% and 18% during the evening peak. Current traditional emission methodology might underestimate emissions by 31.5%. We also used the WRF-Chem model for simulation validation. This hourly-scale inventory provides quantitative support for the precise implementation of pollution control and early warning.

Received: 10 Nov 2025 – Discussion started: 28 Dec 2025

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 1766 KB)

Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Preprint (1766 KB)

Supplement (904 KB)

Download & links

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Journal article(s) based on this preprint

11 May 2026

An improved high-resolution passenger vehicle emission inventory for China using ride-hailing big data

Baojie Li, Zhihui Shen, Yan Li, Yongqi Zhao, Wanglijin Gu, Junjie Liu, Yunkai Yang, Weimeng Zhang, Ziqian Ma, and Hong Liao

Atmos. Chem. Phys., 26, 6197–6211, https://doi.org/10.5194/acp-26-6197-2026,https://doi.org/10.5194/acp-26-6197-2026, 2026

Short summary

Baojie Li, Zhihui Shen, Yan Li, Yongqi Zhao, Wanglijin Gu, Junjie Liu, Yunkai Yang, Weimeng Zhang, Ziqian Ma, and Hong Liao

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2025-5554', Anonymous Referee #1, 01 Jan 2026
General Assessment
Passenger vehicle emissions constitute a significant source of urban air pollution, and accurate quantification of their emission characteristics is fundamental for formulating effective control measures. This study integrates ride-hailing big data with traffic flow modeling to get hourly vehicle speed data gridded at a 0.01° resolution across the nation, and constructs a high spatiotemporal resolution passenger vehicle emission inventory (0.01°×0.01°; hourly). The research quantifies the distribution characteristics of multiple pollutants across different time scales (from hourly to annual), with comparative analyses conducted against conventional algorithms. Furthermore, the WRF-Chem model was employed to validate the inventory through simulation. The methodology employed in this study offers a novel approach for compiling motor vehicle emission inventories and provides enhanced data precision for urban traffic pollution management and air quality modeling. The manuscript is well-organized. Several additional comments are provided for further improvement as follows:

Major Comments:
The core of this paper is to establish the relationship between vehicle speed and emission factors (EF). However, the existing figures and tables do not intuitively present the EF-Speed It is recommended to add a new figure in Section 3.1 showing the EF-Speed curves for NOx, CO, and VOCs, with some explanation.

207: Figures 1a and 1b present the national gridded vehicle speeds at 08:00 and 22:00. In my opinion, additional information should be included in this section, such as the differences in vehicle speeds across different road types (National Highways, Provincial Highways, and County Roads), and explain them properly in the text.

231: In Section 3.2, the manuscript states that "Urban areas, despite occupying only 0.8% of the country’s total land area, accounted for a high 35.3% of total vehicle emissions." What is the definition criterion for "urban and rural areas"? It is necessary to clarify the criteria for defining urban boundaries, so as to avoid potential interpretation bias caused by ambiguous definitions.

Among different urban agglomerations, the emission density difference between urban and rural areas is pretty noticeable. What are the primary reasons for such differences?

The authors obtained national vehicle speeds using ride-hailing big data, but could this data introduce spatial bias into the results? Can it cover all roads? The authors need to conduct further discussions on these aspect to make sure the national-scale emission estimates are reliable.

L.279: The manuscript notes:"During the morning peak, hourly average emissions on workday exceeded those on weekend and holiday by 8% and 5%, respectively, increasing to 12% and 18% during the evening peak." Why does the phenomenon occur?

When comparing with the traditional algorithm, what vehicle speed is adopted in the traditional method? And what are the differences between it and the inventory developed in this study?

L.295: Section 3.4 only provides an overall emission comparison with the traditional method. It doesn’t show the differences in underestimation rates across different time periods—like seasons or holidays.

There are some minor comments:
In Section 2.4, The authors should provide the total number of monitoring stations, as well as the number of those located in regions with the top 20% road density.

L.208: It is recommended to revise the caption of Figure 1 from "Average speed in China" to "The average vehicle speed of passenger vehicles in China in 2019".

L.146 and L.158: Some symbols (such as K_j, v_f, L_p,i) lack complete unit annotations when they first appear.

L.228: Section 3.2 states " The annual amount of CO, VOCs, NOx, PM and NH₃ emitted from national passenger vehicles in 2019 were 4087.8, 1069.4, 211.7, 1.9, 77.5 kt, respectively" while Table 1 shows "CO: 408.78 ten thousand tons…" The unit conversions should be consistent throughout to avoid data ambiguity.

Check the citation format of all references in the full text.
Citation: https://doi.org/10.5194/egusphere-2025-5554-RC1
- AC2: 'Reply on RC1', Baojie Li, 18 Mar 2026
  
  We thank the reviewer very much for the careful reading of our manuscript and helpful comments. We have revised the manuscript following the suggestions, as described in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2025-5554-AC2
CC1:
'Comment on egusphere-2025-5554', Nima Zafarmomen, 02 Jan 2026
This paper presents a significant advancement in the field of atmospheric science and transportation emissions modeling. By leveraging a massive dataset of 23.6 billion ride-hailing trajectories, the authors have successfully transitioned from traditional, often static, emission estimation methods to a dynamic, big-data-driven approach. The novelty of this study lies in its spatial and temporal granularity at a national scale. While previous high-resolution inventories have focused on specific urban agglomerations (like the Pearl River Delta or Beijing-Tianjin-Hebei), this work provides a 0.01∘×0.01∘ hourly grid for the entirety of China.
Technical Comments for Improvement
As the study uses 2019 data, a brief discussion on how the rapid rise of Electric Vehicles (EVs) in China (post-2020) might alter these "speed-emission" curves would add valuable foresight.

While 23.6 billion data points are extensive, ride-hailing vehicles often have different driving patterns compared to private vehicles (e.g., more idling or cruising for passengers). Further clarification on how these patterns were normalized for the general passenger vehicle population would be beneficial.

To further strengthen the discussion on the spatiotemporal characteristics of urban traffic pollutants and to provide a comparative perspective on localized monitoring versus nationwide modeling, I strongly suggest citing the following paper:Comprehensive spatiotemporal analysis of long-term mobile monitoring for traffic-related particles in a complex urban environment Atmospheric Pollution Research, 2025. https://doi.org/10.1016/j.apr.2025.102870
Citation: https://doi.org/10.5194/egusphere-2025-5554-CC1
- AC3: 'Reply on CC1', Baojie Li, 18 Mar 2026
  
  We thank the reviewer very much for the careful reading of our manuscript and helpful comments. We have revised the manuscript following the suggestions, as described in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2025-5554-AC3
RC2:
'Comment on egusphere-2025-5554', Anonymous Referee #2, 08 Feb 2026

General comments:
The article “An improved high-resolution passenger vehicle emission inventory for China using ride-hailing big data” by Li et al., provides information about emission inventory for pollutants emitted by passenger cars in China. The study highlights the importance of average vehicle speed on emission factors. These findings were established in comparison with traditional methods, both showing discrepancies between urban and rural vehicle emissions, weekend relative to workdays emissions and seasonal distribution of pollutants released from vehicular emissions.
Major comments:
The authors should provide more detailed information regarding the vehicular fleet, type of motorization, percentual distribution over the entire country, type of fuel, etc. Also, there is a need to discuss about the type of industrial vehicles and those used for agriculture, passenger car legislation related to the restrictions on pollution and how this apply to other nations.
The authors should expand their consideration to the traffic worldwide and not over the few cities in China and India, even if those cities are very polluted. The emission inventories steady state data should be presented not over the cities in China and India. Even if the present study construct emission inventory of atmospheric pollutants in China the introduction should include a wider view.
The data collected from Amap Ride-hailing Platform are representative for the entire fleet? All the vehicles on the road send data to the platform? It is possible that, systematically, old vehicles emit constantly more pollutants but are not equipped with the tracking system?
The sections 2.1 and 2.2 include many equations which are not well described and their parameters are not always clearly explained in terms of units and meaning (e.g. congestion delay index, Kj, Ct, etc.).
Please include in the papers the problems which usually national monitoring stations have in the terms of trustable data provided. Which parameters are usually provided with huge errors and how these uncertainties are affecting the model?
Please describe in the paper the possible effect on the overall model of the average speed of (42.42 ± 5) km h-1 for example. This average speed should be considered with a range of uncertainties and to extend this uncertainty to the model output data.
Please include the uncertainty bars in the figure 2 for frequency on speed range. Three different days could have different frequency for the same speed range but there is an absolute number as average speed.
There is not convincingly that new model simulations with a better attribution of average speed parameter are the only responsible for the difference between the results in the model validation. The inventory optimization are actually not an important improvements since 0.36% in NMB and 0.02 for R2 are insignificant changes. Please discuss more in detail about the other advantages of the speed average model improvement.
Minor:
Line 95: please use consistent representation of the units „km per year, and g km-1”
Line 110: please include the year.
Line 135: congestion delay index (λ)

Citation: https://doi.org/10.5194/egusphere-2025-5554-RC2
- AC1: 'Reply on RC2', Baojie Li, 18 Mar 2026
  
  We thank the reviewer very much for the careful reading of our manuscript and helpful comments. We have revised the manuscript following the suggestions, as described in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2025-5554-AC1

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2025-5554', Anonymous Referee #1, 01 Jan 2026
General Assessment
Passenger vehicle emissions constitute a significant source of urban air pollution, and accurate quantification of their emission characteristics is fundamental for formulating effective control measures. This study integrates ride-hailing big data with traffic flow modeling to get hourly vehicle speed data gridded at a 0.01° resolution across the nation, and constructs a high spatiotemporal resolution passenger vehicle emission inventory (0.01°×0.01°; hourly). The research quantifies the distribution characteristics of multiple pollutants across different time scales (from hourly to annual), with comparative analyses conducted against conventional algorithms. Furthermore, the WRF-Chem model was employed to validate the inventory through simulation. The methodology employed in this study offers a novel approach for compiling motor vehicle emission inventories and provides enhanced data precision for urban traffic pollution management and air quality modeling. The manuscript is well-organized. Several additional comments are provided for further improvement as follows:

Major Comments:
The core of this paper is to establish the relationship between vehicle speed and emission factors (EF). However, the existing figures and tables do not intuitively present the EF-Speed It is recommended to add a new figure in Section 3.1 showing the EF-Speed curves for NOx, CO, and VOCs, with some explanation.

207: Figures 1a and 1b present the national gridded vehicle speeds at 08:00 and 22:00. In my opinion, additional information should be included in this section, such as the differences in vehicle speeds across different road types (National Highways, Provincial Highways, and County Roads), and explain them properly in the text.

231: In Section 3.2, the manuscript states that "Urban areas, despite occupying only 0.8% of the country’s total land area, accounted for a high 35.3% of total vehicle emissions." What is the definition criterion for "urban and rural areas"? It is necessary to clarify the criteria for defining urban boundaries, so as to avoid potential interpretation bias caused by ambiguous definitions.

Among different urban agglomerations, the emission density difference between urban and rural areas is pretty noticeable. What are the primary reasons for such differences?

The authors obtained national vehicle speeds using ride-hailing big data, but could this data introduce spatial bias into the results? Can it cover all roads? The authors need to conduct further discussions on these aspect to make sure the national-scale emission estimates are reliable.

L.279: The manuscript notes:"During the morning peak, hourly average emissions on workday exceeded those on weekend and holiday by 8% and 5%, respectively, increasing to 12% and 18% during the evening peak." Why does the phenomenon occur?

When comparing with the traditional algorithm, what vehicle speed is adopted in the traditional method? And what are the differences between it and the inventory developed in this study?

L.295: Section 3.4 only provides an overall emission comparison with the traditional method. It doesn’t show the differences in underestimation rates across different time periods—like seasons or holidays.

There are some minor comments:
In Section 2.4, The authors should provide the total number of monitoring stations, as well as the number of those located in regions with the top 20% road density.

L.208: It is recommended to revise the caption of Figure 1 from "Average speed in China" to "The average vehicle speed of passenger vehicles in China in 2019".

L.146 and L.158: Some symbols (such as K_j, v_f, L_p,i) lack complete unit annotations when they first appear.

L.228: Section 3.2 states " The annual amount of CO, VOCs, NOx, PM and NH₃ emitted from national passenger vehicles in 2019 were 4087.8, 1069.4, 211.7, 1.9, 77.5 kt, respectively" while Table 1 shows "CO: 408.78 ten thousand tons…" The unit conversions should be consistent throughout to avoid data ambiguity.

Check the citation format of all references in the full text.
Citation: https://doi.org/10.5194/egusphere-2025-5554-RC1
- AC2: 'Reply on RC1', Baojie Li, 18 Mar 2026
  
  We thank the reviewer very much for the careful reading of our manuscript and helpful comments. We have revised the manuscript following the suggestions, as described in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2025-5554-AC2
CC1:
'Comment on egusphere-2025-5554', Nima Zafarmomen, 02 Jan 2026
This paper presents a significant advancement in the field of atmospheric science and transportation emissions modeling. By leveraging a massive dataset of 23.6 billion ride-hailing trajectories, the authors have successfully transitioned from traditional, often static, emission estimation methods to a dynamic, big-data-driven approach. The novelty of this study lies in its spatial and temporal granularity at a national scale. While previous high-resolution inventories have focused on specific urban agglomerations (like the Pearl River Delta or Beijing-Tianjin-Hebei), this work provides a 0.01∘×0.01∘ hourly grid for the entirety of China.
Technical Comments for Improvement
As the study uses 2019 data, a brief discussion on how the rapid rise of Electric Vehicles (EVs) in China (post-2020) might alter these "speed-emission" curves would add valuable foresight.

While 23.6 billion data points are extensive, ride-hailing vehicles often have different driving patterns compared to private vehicles (e.g., more idling or cruising for passengers). Further clarification on how these patterns were normalized for the general passenger vehicle population would be beneficial.

To further strengthen the discussion on the spatiotemporal characteristics of urban traffic pollutants and to provide a comparative perspective on localized monitoring versus nationwide modeling, I strongly suggest citing the following paper:Comprehensive spatiotemporal analysis of long-term mobile monitoring for traffic-related particles in a complex urban environment Atmospheric Pollution Research, 2025. https://doi.org/10.1016/j.apr.2025.102870
Citation: https://doi.org/10.5194/egusphere-2025-5554-CC1
- AC3: 'Reply on CC1', Baojie Li, 18 Mar 2026
  
  We thank the reviewer very much for the careful reading of our manuscript and helpful comments. We have revised the manuscript following the suggestions, as described in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2025-5554-AC3
RC2:
'Comment on egusphere-2025-5554', Anonymous Referee #2, 08 Feb 2026

General comments:
The article “An improved high-resolution passenger vehicle emission inventory for China using ride-hailing big data” by Li et al., provides information about emission inventory for pollutants emitted by passenger cars in China. The study highlights the importance of average vehicle speed on emission factors. These findings were established in comparison with traditional methods, both showing discrepancies between urban and rural vehicle emissions, weekend relative to workdays emissions and seasonal distribution of pollutants released from vehicular emissions.
Major comments:
The authors should provide more detailed information regarding the vehicular fleet, type of motorization, percentual distribution over the entire country, type of fuel, etc. Also, there is a need to discuss about the type of industrial vehicles and those used for agriculture, passenger car legislation related to the restrictions on pollution and how this apply to other nations.
The authors should expand their consideration to the traffic worldwide and not over the few cities in China and India, even if those cities are very polluted. The emission inventories steady state data should be presented not over the cities in China and India. Even if the present study construct emission inventory of atmospheric pollutants in China the introduction should include a wider view.
The data collected from Amap Ride-hailing Platform are representative for the entire fleet? All the vehicles on the road send data to the platform? It is possible that, systematically, old vehicles emit constantly more pollutants but are not equipped with the tracking system?
The sections 2.1 and 2.2 include many equations which are not well described and their parameters are not always clearly explained in terms of units and meaning (e.g. congestion delay index, Kj, Ct, etc.).
Please include in the papers the problems which usually national monitoring stations have in the terms of trustable data provided. Which parameters are usually provided with huge errors and how these uncertainties are affecting the model?
Please describe in the paper the possible effect on the overall model of the average speed of (42.42 ± 5) km h-1 for example. This average speed should be considered with a range of uncertainties and to extend this uncertainty to the model output data.
Please include the uncertainty bars in the figure 2 for frequency on speed range. Three different days could have different frequency for the same speed range but there is an absolute number as average speed.
There is not convincingly that new model simulations with a better attribution of average speed parameter are the only responsible for the difference between the results in the model validation. The inventory optimization are actually not an important improvements since 0.36% in NMB and 0.02 for R2 are insignificant changes. Please discuss more in detail about the other advantages of the speed average model improvement.
Minor:
Line 95: please use consistent representation of the units „km per year, and g km-1”
Line 110: please include the year.
Line 135: congestion delay index (λ)

Citation: https://doi.org/10.5194/egusphere-2025-5554-RC2
- AC1: 'Reply on RC2', Baojie Li, 18 Mar 2026
  
  We thank the reviewer very much for the careful reading of our manuscript and helpful comments. We have revised the manuscript following the suggestions, as described in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2025-5554-AC1

Peer review completion

AR – Author's response | RR – Referee report | ED – Editor decision | EF – Editorial file upload

AR by Baojie Li on behalf of the Authors (18 Mar 2026) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (25 Mar 2026) by Jayanarayanan Kuttippurath

RR by Anonymous Referee #1 (25 Mar 2026)

RR by Iustinian Bejan (08 Apr 2026)

ED: Publish subject to technical corrections (18 Apr 2026) by Jayanarayanan Kuttippurath

AR by Baojie Li on behalf of the Authors (22 Apr 2026) Manuscript

Journal article(s) based on this preprint

11 May 2026

An improved high-resolution passenger vehicle emission inventory for China using ride-hailing big data

Baojie Li, Zhihui Shen, Yan Li, Yongqi Zhao, Wanglijin Gu, Junjie Liu, Yunkai Yang, Weimeng Zhang, Ziqian Ma, and Hong Liao

Atmos. Chem. Phys., 26, 6197–6211, https://doi.org/10.5194/acp-26-6197-2026,https://doi.org/10.5194/acp-26-6197-2026, 2026

Short summary

Baojie Li, Zhihui Shen, Yan Li, Yongqi Zhao, Wanglijin Gu, Junjie Liu, Yunkai Yang, Weimeng Zhang, Ziqian Ma, and Hong Liao

Supplement

https://doi.org/10.5194/egusphere-2025-5554-supplement

Baojie Li, Zhihui Shen, Yan Li, Yongqi Zhao, Wanglijin Gu, Junjie Liu, Yunkai Yang, Weimeng Zhang, Ziqian Ma, and Hong Liao

Viewed

Total article views: 2,798 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
1,714	852	232	2,798	248	279	270

HTML: 1,714
PDF: 852
XML: 232
Total: 2,798
Supplement: 248
BibTeX: 279
EndNote: 270

Views and downloads (calculated since 28 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	330	105	30	465
Jan 2026	660	329	84	1,073
Feb 2026	222	85	18	325
Mar 2026	410	286	90	786
Apr 2026	71	33	8	112
May 2026	21	14	2	37

Cumulative views and downloads (calculated since 28 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	330	105	30	465
Jan 2026	660	329	84	1,073
Feb 2026	222	85	18	325
Mar 2026	410	286	90	786
Apr 2026	71	33	8	112
May 2026	21	14	2	37

Viewed (geographical distribution)

Total article views: 2,779 (including HTML, PDF, and XML) Thereof 2,779 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 13 May 2026

Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Preprint (1766 KB)
Metadata XML

Short summary

Currently, nationwide hourly emission data is lacking. This study integrated billion ride-hailing trajectory data points and traffic flow models to build a high-res emission inventory. Results show China's passenger vehicles account for 35.3% of urban emissions. Workday morning rush emissions are 8% (vs weekends) and 5% (vs holidays) higher; the gap widens to 12% and 18% in evening rush. Model validation confirms this study's better simulation of fine PM_2.5 and O₃ than traditional methods.


Total:	0
HTML:	0
PDF:	0
XML:	0

An improved high-resolution passenger vehicle emission inventory for China using ride-hailing big data

Journal article(s) based on this preprint

Interactive discussion

Interactive discussion

Peer review completion

Suggestions for revision or reasons for rejection

Journal article(s) based on this preprint

Supplement

Viewed

Viewed (geographical distribution)