Recommendations on Benchmarks for Photochemical Air Quality Model Applications in China — NO2, SO2, CO and PM10
Hehe Zhai,Ling Huang,Chris Emery,Xinxin Zhang,Yangjun Wang,Greg Yarwood,Joshua S. Fu,Li
DOI: https://doi.org/10.1016/j.atmosenv.2023.120290
IF: 5
2024-01-01
Atmospheric Environment
Abstract:Photochemical air quality models (AQMs) are a vital tool for atmospheric pollution research and have been widely used in various applications, such as air quality prediction and evaluation of pollution control strategies. Before using these models for further studies, it is essential to thoroughly evaluate their reliability and accuracy. While previous guidelines and benchmarks have primarily focused on fine particulate matter (PM2.5) and ozone (O3), there is still a lack of benchmarks for evaluating the model performance on primary criteria pollutants such as sulfur dioxide (SO2), nitrogen dioxide (NO2), carbon monoxide (CO), and particulate matter with aerodynamic diameter less than or equal to 10 mu m (PM10). The use of air quality models in China has increased significantly in the past decades. However, there is still a lack of standardized benchmarks for the performance evaluation of these models. Building upon our previous work on PM2.5 and its chemical species, we propose a set of benchmarks for evaluating the performance of the aforementioned four air pollutants. Initially, we identified a total of 475 papers published during 2007-2019 that utilized at least one of the five commonly used AQMs in China. From these papers, we selected 164 articles that provided model performance evaluation (MPE) results of the four primary air pollutants. The three most frequently used Model Performance Evaluation (MPE) metrics were selected to analyse the impact of different model configurations on the reported statistics, including modelling region, season, and emission inventory. Lastly, three commonly used statistical indicators, including normalized mean bias (NMB), normalized mean error (NME), and correlation coefficient (R), were proposed for the validation of simulated NO2, SO2, CO, and PM10. Two sets of benchmarks are given, including the "goal" and "criteria". The "goal" represents the best range of performance that a model can be expected to achieve, and the "criteria" represents performance that the majority of studies have achieved. We recommend R values above 0.50, 0.35, 0.45, and 0.40 for NO2, SO2, PM10, and CO, respectively, in order to meet the "criteria" benchmark. If the "goal" benchmark is to be achieved, the corresponding R values are 0.60, 0.55, 0.60, and 0.60. The "goal" benchmarks of NMB for NO2, SO2, PM10, and CO are within +/- 20%, +/- 25%, +/- 20%, and +/- 25%, respectively; while the "goal" benchmarks of NME for the four pollutants are less than 40%, 45%, 45%, and 60%, respectively. These benchmarks supplement our previous benchmarks for PM2.5 and its components and provide a more comprehensive guideline for the air quality modelling community in China.