Estimating nitrogen oxides emissions at city scale in China with a nightlight remote sensing model

Highlights

• An effective nightlight based model was developed to assess city scale NO_x emission.

• NO_x emission showed a significant linear correlation with night stable lights.

• The modeled NO_x emissions were about 4.1%–13.8% higher than previous inventory.

• Spatial analyses revealed key areas for NO_x mitigation management in China.

Abstract

Increasing nitrogen oxides (NO_x) emissions over the fast developing regions have been of great concern due to their critical associations with the aggravated haze and climate change. However, little geographically specific data exists for estimating spatio-temporal trends of NO_x emissions. In order to quantify the spatial and temporal variations of NO_x emissions, a spatially explicit approach based on the continuous satellite observations of artificial nighttime stable lights (NSLs) from the Defense Meteorological Satellite Program/Operational Linescan System (DMSP/OLS) was developed to estimate NO_x emissions from the largest emission source of fossil fuel combustion. The NSL based model was established with three types of data including satellite data of nighttime stable lights, geographical data of administrative boundaries, and provincial energy consumptions in China, where a significant growth of NO_x emission has experienced during three policy stages corresponding to the 9^th–11^th Five-Year Plan (FYP, 1995–2010). The estimated national NO_x emissions increased by 8.2% per year during the study period, and the total annual NO_x emissions in China estimated by the NSL-based model were approximately 4.1%–13.8% higher than the previous estimates. The spatio-temporal variations of NO_x emissions at city scale were then evaluated by the Moran's I indices. The global Moran's I indices for measuring spatial agglomerations of China's NO_x emission increased by 50.7% during 1995–2010. Although the inland cities have shown larger contribution to the emission growth than the more developed coastal cities since 2005, the High–High clusters of NO_x emission located in Beijing-Tianjin-Hebei regions, the Yangtze River Delta, and the Pearl River Delta should still be the major focus of NO_x mitigation. Our results indicate that the readily available DMSP/OLS nighttime stable lights based model could be an easily accessible and effective tool for achieving strategic decision making toward NO_x reduction.

Introduction

Estimating the emissions of nitrogen oxides (NO_x) is increasingly important due to the rapid emission growth in many regions of developing countries. The high NO_x emissions contribute directly or indirectly to the adverse environmental consequences including haze, acid deposition and climate change from local to the global scale via the nitrogen cascade (Erisman et al., 2011, Huang et al., 2014a, Liu et al., 2013, Zien et al., 2014). Driven by the increasing fossil fuel combustion for thermoelectric power supply, industrial production and transportation, NO_x emissions in the fast developing Asia and Middle East countries were reported to increase by 5% to 10% per year (Hilboll et al., 2013). With a much higher growth rate than the direct greenhouse gases such as carbon dioxide, methane and other atmospheric pollutants such as sulfur dioxide (IPCC, 2013, Klimont et al., 2013), the mitigation of NO_x emissions has become the most critical issue in these areas. As the basis for mitigation decision making, effective approaches to identifying emission sources and estimating emissions have been a necessity (Liu et al., 2015). Thus, it is essential to quantify the spatio-temporal patterns of NO_x emissions from fossil fuel combustions to improve the NO_x mitigation decision making.

Anthropogenic NO_x emissions at the large spatial–temporal scale have been estimated by several approaches. The bottom-up emission inventories calculate national or provincial scale NO_x emissions by statistical data on human activity and emission factors, then distribute the overall emissions to grids by spatial proxies such as human settlement, geographical coordinates of power plants and industrial facilities (Kurokawa et al., 2013, Olivier et al., 1996, Zhao et al., 2013). However, the lack of validated high-resolution human activity data especially for developing countries could result in a great deal of uncertainty (Zhao et al., 2011). Besides, the time duration required to collect valid statistical data can always lead to a delayed emission inventory, which is vital to a good strategy making on NO_x reduction. As a consequence, researchers have shown a growing interest in the top-down satellite approach to timely map the spatial variations of NO_x emissions. Typically the overall NO_x emissions were estimated by tropospheric NO₂ column data retrieved from satellite sensors, e.g., the Global Ozone Monitoring Experiment (GOME), the Ozone Monitoring Instrument (OMI) and the Scanning Imaging Absorption Spectrometer for Atmospheric Cartography (SCIAMACHY) (Hilboll et al., 2013, Lamsal et al., 2011). Nevertheless, uncertainties associated with NO₂ columns calculation and model parameters such as cloud interference (Boersma et al., 2004, Martin, 2008), difficulties in differentiating anthropogenic emission from the measured overall emission, have limited its application to strategic decision-making on NO_x mitigation.

A proof-of-concept approach developed here is to combine the provincial-scale human activity inventories and satellite data of nocturnal lights that reflect spatio-temporal variations. The Defense Meteorological Satellite Program's Operational Linescan System (DMSP/OLS) has been monitoring distribution and brightness of global nighttime lights since 1976 (Elvidge et al., 2007). The resultant nighttime stable lights (NSLs) from DMSP/OLS can detect artificial nighttime lights from human settlement, industrial sites and road network with a maximum resolution of 0.0083° (about 1000 m) while excluding interferences of clouds, sun and moon lights (Elvidge et al., 1997, NOAA, 2015). NSL has been regarded as an ideal indicator for human activities as it can provide spatially explicit imageries at a high resolution over a continuous time period. Its strong correlation with energy consumption has been demonstrated (Letu et al., 2010, Letu et al., 2014), and several regression models have been developed and verified to estimate fossil fuel related CO₂ emissions by pixel digital numbers of DMSP/OLS imageries (Doll et al., 2000, Ghosh et al., 2010, Su et al., 2014). Since fossil fuel consumption is also the largest source of NO_x emission (Kato and Akimoto, 1992), there is a good potential to estimate NO_x emissions by the nighttime lights remote sensing data while it has rarely be explored in previous studies. The only work by Toenges-Schuller et al. (2006) reported a correlation coefficient of 0.79 between global NSL imagery and gridded NO_x emissions estimated by EDGAR inventory. However, whether and how the nighttime lights data could be used to estimate the spatial and temporal variations of NO_x emissions at various scales of specific geographical area remain to be investigated.

China accounts for over 10% of the global NO_x emission (Hilboll et al., 2013, Richter et al., 2005). To reduce impacts of rapidly growing NO_x emissions in China, the government adopted a region specific mitigation strategy appropriate for Beijing-Tianjin-Hebei region, the Yangtze River Delta and the Pearl River Delta by adjusting energy structure, improving energy efficiency and promoting emission reduction technology in the recently issued Atmospheric Pollution Prevention Action Plan (State Council of China, 2013). Quantifying the spatial disparities and temporal variations of the energy derived NO_x emission in a timely manner therefore becomes especially important to assist a rational decision making for regional pollution control. However, there exists a data and knowledge gap using geographically specific data of energy consumption activities for spatial estimation of NO_x emissions. Therefore, the objective of this study was to develop a novel method based on DMSP/OLS nighttime light imageries for estimating NO_x emissions from fossil fuel combustion using data in China as a case study. A regression model for estimating NO_x emissions at the city scale in China was developed based on NSL data derived from DMSP/OLS and provincial-scale NO_x emissions estimated by the inventory approach, and feasibility of the NSL-based model was verified by a 16-year simulation over the 9th–11th Five Year Plan period (1995–2010). This work is expected to quantify spatio-temporal variations of NO_x emissions as a basis for decision-makers to develop timely NO_x mitigation strategies toward the cleaner air in China.