Inverse modelling of Chinese NO_x emissions using deep learning: integrating in situ observations with a satellite-based chemical reanalysis

Abstract

Abstract. Nitrogen dioxide (NO2) column density measurements from satellites have been widely used in constraining emissions of nitrogen oxides (NOx = NO + NO2). However, the utility of these measurements is impacted by reduced observational coverage due to cloud cover and their reduced sensitivity toward the surface. Combining the information from satellites with surface observations of NO2 will provide greater constraints on emission estimates of NOx. We have developed a deep-learning (DL) model to integrate satellite data and in situ observations of surface NO2 to estimate NOx emissions in China. A priori information for the DL model was obtained from satellite-derived emissions from the Tropospheric Chemistry Reanalysis (TCR-2). A two-stage training strategy was used to integrate in situ measurements from the China Ministry of Ecology and Environment (MEE) observation network with the TCR-2 data. The DL model is trained from 2005 to 2018 and evaluated for 2019 and 2020. The DL model estimated a source of 19.4 Tg NO for total Chinese NOx emissions in 2019, which is consistent with the TCR-2 estimate of 18.5 ± 3.9 Tg NO and the 20.9 Tg NO suggested by the Multi-resolution Emission Inventory for China (MEIC). Combining the MEE data with TCR-2, the DL model suggested higher NOx emissions in some of the less-densely populated provinces, such as Shaanxi and Sichuan, where the MEE data indicated higher surface NO2 concentrations than TCR-2. The DL model also suggested a faster recovery of NOx emissions than TCR-2 after the Chinese New Year (CNY) holiday in 2019, with a recovery time scale that is consistent with Baidu “Qianxi” mobility data. In 2020, the DL-based analysis estimated about a 30 % reduction in NOx emissions in eastern China during the COVID-19 lockdown period, relative to pre-lockdown levels. In particular, the maximum emission reductions were 42 % and 30 % for the Jing-Jin-Ji (JJJ) and the Yangtze River Delta (YRD) mega-regions, respectively. Our results illustrate the potential utility of the DL model as a complementary tool for conventional data-assimilation approaches for air quality applications.

Ozone in the Desert Southwest of the United States: A Synthesis of Past Work and Steps Ahead

“…Process-based modeling studies to identify inconsistencies in NEI emission estimates (and trends) relative to observed pollutant levels must be pursued. The use of CTMs for multi-pollutant emission estimation and data assimilation studies, e.g., refs − , together with the application of source apportionment and attribution strategies, e.g., refs − , including machine-learning approaches, , provide opportunities to reduce uncertainties in these emission estimates. This should, however, be conducted in association with improvements in input and ancillary datasets that account for local heterogeneity especially land cover classification and urban development, e.g., refs and , as well as emission trends and scenarios like changes in mobile and stationary sources, e.g., ref , and even the shift to renewable energy, , increasing contribution of noncombustion sources, , all of which can drive potential shifts in chemical regimes.…”

TROPOMI NO2 Shows a Fast Recovery of China’s Economy in the First Quarter of 2023

“…However, the lack of real-time economic statistics makes it challenging to timely monitor the economic recovery . Previous studies have disclosed that ambient nitrogen dioxides (NO 2 ), a reactive and short-lived gas mainly released from fuel combustion, , respond to emergency actions swiftly and fluctuate dynamically with the changes in anthropogenic activities and emissions, particularly during and after the COVID-19 lockdowns. − The sensitivity of NO 2 concentrations to socioeconomic changes has been witnessed on a global scale. , NO 2 tropospheric vertical column densities (TVCDs) retrieved from satellite observations have played an important role in monitoring ambient NO 2 concentrations, economic activities, and anthropogenic emission variations. − Here we investigate the daily dynamics of satellite-observed NO 2 around the Chinese New Year (CNY) in China since 2005. Chemical transport models, high-resolution emission inventories, and population density distribution maps are utilized to attribute the NO 2 changes to emission sectors and regions, which helps us to understand the drivers of satellite-observed NO 2 changes.…”

The Capability of Deep Learning Model to Predict Ozone Across Continents in China, the United States and Europe

“…The data‐driven approach provides an efficient way of integrating multisource data and predicting atmospheric compositions, which could be used as a useful supplement to process‐driven chemical transport models (CTMs). Recent studies demonstrated applications of data‐driven techniques to provide air quality forecasts (Bi et al., 2022; Zhang et al., 2023), spatial extensions of atmospheric observations (Liu et al., 2022; Wei et al., 2023), more accurate or rapid CTM simulations (Shen et al., 2022; Xing et al., 2020), and atmospheric pollutant emission estimates (He et al., 2022b; Xing et al., 2022). Furthermore, recent studies have highlighted the importance of data‐driven techniques to provide a better understanding of atmospheric ozone (O 3 ), for example, the contributions of meteorological and anthropogenic sources to observed O 3 changes (Chen et al., 2023; Wang et al., 2023; Weng et al., 2022).…”