Diagnosing the Model Bias in Simulating Daily Surface Ozone Variability Using a Machine Learning Method: The Effects of Dry Deposition and Cloud Optical Depth

Machine learning methods are increasingly used in air quality studies to predict air pollution levels, while few applied them to diagnose and improve the underlying mechanisms controlling air pollution represented in chemical transport models (CTMs). Here, we use the random forest (RF) method to diagnose high biases of surface daily maximum 8 h average (MDA8) ozone concentrations in the GEOS-Chem CTM evaluated against measurements from the nationwide monitoring network in summer 2018 over China. The feature importance results show that cloud optical depth (COD), relative humidity, and precipitation are the top three factors affecting CTM high biases. Such results indicate that the high ozone biases in summer over China mainly occur on wet/cloudy days (∼40% biased high), while biases on dry/clear days are small (within 5%). We link the important features with model parameterizations and variables, identifying model underestimates in the dry deposition velocity and COD on wet/cloudy days. By accounting for the enhanced dry deposition on wet plant cuticles and using satellite observation constrained COD, we find that CTM high ozone biases can be halved with an improved agreement in the temporal variability, highlighting the effects of dry deposition and COD on ozone, as suggested by the RF outcomes.

A Feasible Methodological Framework for Uncertainty Analysis and Diagnosis of Atmospheric Chemical Transport Models

The current state of quantifying uncertainty in chemical transport models (CTM) is often limited and insufficient due to numerous uncertainty sources and inefficient or inaccurate uncertainty propagation methods. In this study, we proposed a feasible methodological framework for CTM uncertainty analysis, featuring sensitivity analysis to filter for important model inputs and a new reduced-form model (RFM) that couples the high-order decoupled direct method (HDDM) and the stochastic response surface model (SRSM) to boost uncertainty propagation. Compared with the SRSM, the new RFM approach is 64% more computationally efficient while maintaining high accuracy. The framework was applied to PM_2.5 simulations in the Pearl River Delta (PRD) region and found five precursor emissions, two  pollutants in lateral boundary conditions (LBCs), and three meteorological inputs out of 203 model inputs to be important model inputs based on sensitivity analysis. Among these selected inputs, primary PM_2.5 emissions, PM_2.5 concentrations of LBCs, and wind speed were identified as key uncertainty sources, which collectively contributed 81.4% to the total uncertainty in PM_2.5 simulations. Also, when evaluated against observations, we found that there were systematic underestimates in PM_2.5 simulations, which can be attributed to the two-product method that describes the formation of secondary organic aerosol.

Impacts of using an ensemble Kalman filter on air quality simulations along the California-Mexico border region during Cal-Mex 2010 field campaign

The purpose of this study is to investigate the impact of using an ensemble Kalman filter (EnKF) on air quality simulations in the California-Mexico border region on two days (May 30 and June 04, 2010) during Cal-Mex 2010. The uncertainties in ozone (O3) and aerosol simulations in the border area due to the meteorological initial uncertainties were examined through ensemble simulations. The ensemble spread of surface O3 averaged over the coastal region was less than 10ppb. The spreads in the nitrate and ammonium aerosols are substantial on both days, mostly caused by the large uncertainties in the surface temperature and humidity simulations. In general, the forecast initialized with the EnKF analysis (EnKF) improved the simulation of meteorological fields to some degree in the border region compared to the reference forecast initialized with NCEP analysis data (FCST) and the simulation with observation nudging (FDDA), which in turn leading to reasonable air quality simulations. The simulated surface O3 distributions by EnKF were consistently better than FCST and FDDA on both days. EnKF usually produced more reasonable simulations of nitrate and ammonium aerosols compared to the observations, but still have difficulties in improving the simulations of organic and sulfate aerosols. However, discrepancies between the EnKF simulations and the measurements were still considerably large, particularly for sulfate and organic aerosols, indicating that there are still ample rooms for improvement in the present data assimilation and/or the modeling systems.

Uncertainties influencing health-based prioritization of ozone abatement options

The primary goal of air quality management is protection of human health. Therefore, formulation of ground-level ozone mitigation policies could be informed by considering not just attainment of regulatory standards but also how control measures benefit public health. However, evaluation of health impacts is complicated by uncertainties associated with photochemical modeling and epidemiological studies. This study demonstrates methods to characterize uncertainties influencing health-benefits estimation of ozone reduction (averted premature mortalities due to short-term exposure) in the Dallas-Fort Worth (DFW) region. Uncertainty in photochemical modeling and the selection of temporal metric (duration of ozone exposure) for concentration-response relationships can each affect the health-based prioritization of ozone control options. For example, deterministic results (neglecting uncertainties) based on 8-h daily maximum ozone reduction shows DFW anthropogenic NO(x) controls to yield 9.23 times as much benefit per ton as VOC controls. However, the rankings reverse under 5.7% of the cases (including 2.8% cases that exhibit incremental mortalities due to NO(X) control) when uncertainties in the photochemical model are considered. Evaluated ozone exposure on a 24-h rather than an 8-h basis also reverses the rankings.

Procedures for estimation of modelling uncertainty in air quality assessment

The main objectives of this work focus, firstly, on a review of the current existent methodologies to estimate air quality modelling uncertainty, and, secondly, in the preparation of guidelines for modelling uncertainty estimation, which can be used by local and regional authorities responsible for air quality management. From the application exercise, it was concluded that it is possible to define a subset of statistical parameters able to reproduce the general uncertainties estimation. Concerning the quality indicators defined by EU directives, the results show that the legislated uncertainty estimation measures are ambiguous and inadequate in several aspects, mainly in what concerns the error measures for hourly and daily indicators based on the highest observed concentration. A relative error at the percentile correspondent to the allowed number of exceedances of the limit value was suggested and tested, showing that is a more robust and appropriate parameter for model performance evaluation.

Air quality assessment for Portugal

According to the Air Quality Framework Directive, air pollutant concentration levels have to be assessed and reported annually by each European Union member state, taking into consideration European air quality standards. Plans and programmes should be implemented in zones and agglomerations where pollutant concentrations exceed the limit and target values. The main objective of this study is to perform a long-term air quality simulation for Portugal, using the CHIMERE chemistry-transport model, applied over Portugal, for the year 2001. The model performance was evaluated by comparing its results to air quality data from the regional monitoring networks and to data from a diffusive sampling experimental campaign. The results obtained show a modelling system able to reproduce the pollutant concentrations' temporal evolution and spatial distribution observed at the regional networks of air quality monitoring. As far as the fulfilment of the air quality targets is concerned, there are excessive values for nitrogen and sulfur dioxides, ozone also being a critical gaseous pollutant in what concerns hourly concentrations and AOT40 (Accumulated Over Threshold 40 ppb) values.