Project Topics
Get Complete Project Material File(s) Now! »
The development of atmospheric models
Atmospheric models are mathematical representations based on a complete set of dynamic equations that can generate physical and numerical data of climate and chemical parameters (Li et al., 2016). Today, an atmospheric model becomes an essential tool in a variety of atmospheric sciences applications. Early in 1904, the Norwegian scientist Vilhelm Bjerknes listed seven basic variables (temperature; pressure; air density; humidity; the three components of wind velocity) and set down a two-step scientific viewpoint for the weather prediction: (i) the initial state of the atmosphere is determinates by observation giving the distribution of the variables at different levels in the diagnostic step first, (ii) then the changes over time calculated using the law of motion in the prognostic step (Lynch, 2008). By about 1922, the British scientist Lewis Fry Richardson presaged the numerical weather prediction after the advent of electronic computers in his book Weather Prediction by Numerical Process (Inness and Dorling, 2012; Somerville, 2011). In those days, the weather forecasting was a haphazard process, the forecaster used rough techniques of extrapolation based on their knowledge of local climatology and intuition, but the principles of theoretical physics played little role in practical forecasting, weather forecasting was more an art than a science (Lynch, 2002, 2008). Since the 1960s, with the development of computers and network communication techniques, the atmospheric model arises at the historic moment (Hersbach et al., 2015; Paine, 2019). For the past half century, with the rapid development of high-performance computers, satellite detection and the continuous in-depth research of numerical prediction theory, numerical prediction has achieved great success and has become an important symbol of the atmospheric science and the main methods of weather forecasting operations (Meng et al., 2019). Although the processes of climate change are of global proportions, there are significant differences in its specific manifestations between regions (temperature; precipitation and land-use changes). Thus, atmospheric mesoscale models are being embedded to provide specific guidance to policy makers at all administrative levels. The development of mesoscale models goes back to the 1960’s, in which with a grid size of tens of kilometers to simulate regional weather over a few days (Dudhia, 2014). With the evolution of dynamics and physical representations and the availability of cheaper computing resources, high resolution simulation and forecast became possible and various ―local customized‖ models began to emerge. For example, a mesoscale meteorological model coupled with a chemical transport model (CTM) is an essential tool to provide the regional airshed information to help government develop strategies for manage regional air quality. In the past two decades, those models have been used to provide information for integrated models like GAINS (Greenhouse gas – Air pollution Interactions and Synergies) for key negotiations on the air pollution control agreements (Amann et al., 2011). New approaches based on statistics are in use or under development like the Screening for High Emission Reduction Potential on Air (SHERPA) tool. In SHERPA (Pisoni et al., 2017; Thunis et al., 2016a), a different approach is undertaken that reproduces the grid-cell-to-grid-cell approach but does not require anywhere near as many CTM runs. SHERPA assumes that the unknown parameters vary on a cell-by-cell basis but are no longer independent of each other.
According to the model design concepts and parameters, the development of CTM is roughly divided into three generations (Casado, 2013). The first generation of CTM originated in the 1970’s which mainly include the box model based on the law of conservation of mass (Seinfeld, 1988) such as Empirical Kinetic Modeling Approach (EKMA), the Gaussian dispersion model based on the statistical theory of turbulence diffusion (Atkinson et al., 1997) like Industrial Source Complex model (ISC). These models use simple, parameterized linear mechanisms to describe complex atmospheric physical processes, which are suitable for simulating the long-term average concentration of inert pollutants (Watson et al., 1988b). In the early 1980s, with the study of the turbulence characteristics of the atmospheric boundary layer, researchers found that the Gaussian model could not answer many questions, which gradually promoted the development of the second generation. The second generation of CTM considers nonlinear response mechanism and incorporates the treatment of gas and aqueous phase chemistry. These models divide the simulated domain into many three-dimensional grid cells, the cloud, fog and precipitation scavenging processes are calculated in each cell (Carey Jang et al., 1995; Liu et al., 1984; Stockwell et al., 1990). Representative models in this category include the Urban Air shed Model (UAM), the Regional Acid Deposition Model (RADM), etc. The second generation CTM are only designed to address individual pollutant issues such as ozone and acid deposition, which does not fully consider the mutual transformation and mutual influence of various pollutants (Reis et al., 2005). However, the physical and chemical reaction processes among various pollutants are complex in the real atmosphere. In 1990’s, the US Environmental Protection Agency (EPA) developed the third-generation air quality simulation platform Models-3 based on the concept of « one atmosphere » (Byun and Schere, 2006a). At present, the most widely used are the third-generation comprehensive CTM in clouding WRF-Chem, the Comprehensive Air-quality Model with extensions (CAMx), Community Multiscale Air Quality Model (CMAQ), CHIMERE, etc.
The WRF-Chem is a regional atmospheric dynamic-chemical coupling model developed by the US National Center for Atmospheric Research (NCAR), which is integrated with the atmospheric chemistry module in the mesoscale Weather Research and Forecast model (WRF) (Grell et al., 2005). The WRF provides online large airflow fields for CTM, simulating pollutant transportation, dry and wet deposition, gas phase chemistry, aerosol formation, radiation, biological radiation, etc. (Lin et al., 2020). The advantage of WRF-Chem is that the meteorology mode and the chemical transport mode are fully coupled in time and space resolution to achieve true online feedback. The CAMx model is a comprehensive CTM developed by ENVIRON on the basis of the UAM model. In addition to the typical features of the third-generation air quality model, the most famous features of CAMx include: two-way nesting and flexible nesting, grid plume module, ozone source allocation technology, particulate source allocation technology(Bove et al., 2014; Pepe et al., 2016). The CMAQ, emission inventory processing model (SMOKE), mesoscale meteorological model (MM5 or WRF.) together constitute the Models-3 platform, of which CMAQ is the core of the entire system (Hanna, 2008). The CMAQ was originally designed to comprehensively deal with complex air pollution problems such as tropospheric ozone, acid deposition, and visibility. For this reason, the design concept of CMAQ can systematically simulate various scales and various complicated air pollution issues. The CMAQ model has become a quasi-regulatory model used by the US EPA in environmental planning, management and decision-making. The characteristics of this model are: simultaneously simulate the behavior of a variety of air pollutants, including ozone, PM, acid deposition, and visibility and other air quality problems in different spatial scales; make full use of the latest Computer hardware and software technologies, such as high-performance computing, modular design, visualization technology, etc., make air quality simulation technology more efficient and accurate, and the application fields tend to be diversified. Since CMAQ 5.0, the model has realized the on-line coupling of meteorological model, absorbing the advantages of WRF-CHEM model (Kong et al., 2015). CHIMERE is a three-dimensional CTM driven by meteorological drivers like MM5 or WRF. More than 80 kinds of species more than 300 reactions are described in the model. Processes include chemistry, transportation, vertical diffusion, photochemistry, dry deposition, absorption in and below clouds, and SO2 oxidation. Clouds are included in the process model. It can simulate processes including gas-phase chemistry, vertical diffusion, photochemistry, aerosol formation, deposition and transport at regional and urban scales (Bessagnet et al., 2004; Vautard et al., 2005). A new version CHIMERE-V2019 has realized the online coupling of meteorological model like WRF-Chem and CMAQ (Bessagnet et al., 2020).
Compare to the first and second generation CTM, the third generation CTM has distinct advances in:
(i) full consideration of various atmospheric physical processes, chemical reactions between pollutants and gas-solid two-phase transformation process (San Joséet al., 2009);
(ii) based on nested grid design, it can be used as a multi-scale atmospheric simulation and prediction tool (Xue et al., 2001);
(iii) simultaneously simulate the variety of air pollutants, including ozone, PM, acid deposition, visibility and other environmental pollution problems;
(iv) make full use of the latest computer technologies, such as high-performance computing, modular design, visualization technology to make air quality
simulation technology more efficient and accurate (Reis et al., 2005).
The effective coupling between the dynamical processes and physical parameterizations during the short-term weather changes and extremes is still a significant challenge. Under the background of global warming, the chaotic nature of the atmosphere becomes stronger, which puts forward higher requirements for long-term accurate forecasting, a stronger fusion between model and observations for its data assimilation and bias corrections are needed. Maybe the next-generation model with a focus on chemical and physical processes, neural networks, machine learning are the ways in the future model evolution.
Urbanization and air quality
In recent decades, air pollution in urban areas, especially megacities, have aroused people’s attention (Baklanov et al., 2016b). With the accelerating process of industrialization and urbanization, an increasing number of people will be affected by such process especially in developing regions (Beirle et al., 2011; Fang et al., 2015; Hopke et al., 2008). Urbanization was positively related to global health in the short term and long term. In the short run, 1% increase in urbanization was associated with reduced mortality, under-five mortality, and infant mortality of 0.05%, 0.04%, and 0.04%, respectively, as well as increased life expectancy of 0.01 year (Wang, 2018). However, the process of urbanization has also made important contributions to regional climate changes and has caused a bad effect on ecosystem and extreme climate significantly increased (Changnon et al., 1996; Semenza et al., 1996). Since the end of World War II, the world’s population has grown from 2.5 billion to about 7 billion today, the urban pollution has risen from below 30% to 58% in the past seventy years. Air pollution in cities is exacerbated by increased human activity in urban areas due to population growth (Miranda et al., 2015). The recent global lockdown events response to COVID-19 pandemic have reduced the NO2concentrations and particulate matter levels by about 60% and 31% respectively by using satellite data and a network of more than 10000 air quality observation stations in 34 countries (Venter et al., 2020), the NO2 and PM2.5 concentration in China was reduced by around 20% and 17% for 30–50 days (Z. Liu et al., 2020; Zinke, 2020). Using the mesoscale meteorological and chemical transport model, Menut et al.(2020) noted a decrease of -30% to -50% for NO2 and -5% to -15% for particulate matter in all western Europe counties during the lockdown events. In general, air pollution is positively correlated with the initial and acceleration stages of urbanization, with pollution in major cities tending to increase during the construction phase, passing through a maximum pollution level and then again decreased at the end of the urbanization process as abatement strategies are developed (Fenger, 1999). In the industrialized Europe countries, urban air pollution is in some respects in the last stage with effectively reduced levels of many pollutants such as Sulphur dioxide, carbon monoxide and lead (Fenger, 2002). Figure1.5 shows that ground‐level concentrations of almost all pollutants have declined steadily since 2000 in 28 countries Europe Union (EU28). The two notable results concerning emission reductions are the total SOx from the EU28 member states were cut by 69 % since 2000 and the almost no changes of total ammonia (Koolen and Rothenberg, 2019).
Figure1.5 Emission trends in the EU28 on the basis of fuel sold (Koolen and Rothenberg, 2019)
Besides, geographical condition also plays an important role in transportation and dispersion of pollutants. Pollutants dispersion processes over the valley-basin city are much more complicated than over flat areas. Therefore, pollution episodes have been frequently witnessed over complex terrain, especially in wintertime (Chen et al., 2019; Sabatier et al., 2020). In coastal cities, shipping emissions contribute to air quality degradation. Viana et al. (2014) found that shipping emissions contribute with 1% to 7% for PM10, with 1% to 20% for PM2.5, and with 7% to 24% for NO2 annual mean concentrations in European coastal areas. In addition, continuing urbanization has resulted in cities that are almost always warmer than the surround, which are known as urban heat island (UHI). A recent research shows that urban region has higher heat index than rural areas by a difference of about 1.5– 2°C (Bhati and Mohan, 2018). Compared with the natural vegetation canopy, the city has its unique features: taking buildings as an example, different buildings have different functions, shapes, heights, orientations, styles, etc. (Wong and Chen, 2008). Figure1.6 shows the differences in surface temperature between buildings and vegetation canopy. Meanwhile, UHI can significantly exacerbate building energy consumption (Palme et al., 2017). The growing cooling energy consumption caused by UHI will increase CO2 emissions by up to five times in 2050 than in 2000 for buildings in cities (Y. Liu et al., 2020). Besides, there are various types of artificial surfaces in urban areas, such as asphalt roads, concrete sidewalks, glass curtain walls, parking lots, and green spaces which have impact on urban air quality (Wise, 2016). For example, computational fluid dynamics studies indicated that the flow resistance of tree crowns decreased wind speed and the dispersion of pollutants is limited in the ground level and the particulate matter concentrations accumulate within the canyon (Gromke and Blocken, 2015). These diversified types of artificial areas appear alternately, leading to complex, diverse, and non-uniform urban canopy characteristics, which give more complex turbulent airflows in urban region, increased the difficulty of air quality simulation and forecast in cities than in other regions. The urban canopy, urban street canopy and urban boundary layer consist of the urban micro environment. The boundary layer is the atmospheric structure closest to the underlying surface. The meteorological conditions within the boundary layer play an important role in the formation of air pollution (Bianco et al., 2008). The wind field determines the regional transportation of pollutants. The height of the atmospheric boundary layer determines the amount of ventilation and the dilution capacity of pollutants. Heavy pollution episodes are always accompanied by a low boundary layer height(Quan et al., 2014; Yin et al., 2019). In cities, urbanization causes thermodynamic perturbations and facilitates the development of the boundary layer (Miao et al., 2019). Moreover, some aerosols like black carbon can warm the upper PBL which help to stabilize the boundary layer height and weaken turbulent mixing, resulting in a decrease of the boundary layer height (BLH), which enhances the accumulation of air pollutants (Ding et al., 2016; Li et al., 2017).
Overall, cities are highly sensitive to the impact of meteorological disasters. How to be efficient in extreme climates risk analysis and early warning, firmly keeping the bottom line of city safe operation and ecological environment protection is a concern of scientists and politicians. The impact of the accelerated urbanization process on the regional climate and atmospheric environment mainly includes the following aspects:
(1) The urban underlying surface is defined as the part of the city in direct exchange with the atmosphere (Y. Li et al., 2020), the changes of the urban underlying surface have changed the environmental properties of the natural underlying surface;
(2) The urban canopy has a « shading effect » on solar radiation, which reduces the solar radiation reaching the surface. At the same time, there is the exchange and storage of radiation energy between the buildings in the canopy, which affects the surface energy balance;
(3) Human activities will increase atmospheric pollutants and artificial heat generation. In short, in addition to bringing economic and social benefits to humanity, urbanization has also brought many problems to the urban climate and atmospheric environment Statement.
The climate effect brought by urbanization and its impact on the atmospheric environment is becoming one of the key issues of the world.
Objectives and outline of the PhD thesis
The main goal of this PhD work is to improve the air quality simulation and forecast in France especially in urbanized areas. For its operational requirements, the system requires continuous improvements, particularly the ability to better predict certain types of episodes and exceedances of limit values for criteria pollutants. Patterns of urban air pollution are rather variable and spatially heterogeneous. Turbulence plays an important role on the vertical mixing of physical parameters and pollutants.. Understanding the impact of multiple parameterization schemes on air quality simulation can help improve model configuration to address local areas.
The third generation of CTM system generally has three basic modules: a meteorological process module (input information from mesoscale meteorological models), emission inventory pre-processor and the core air quality simulation module. Studies indicate that the physics parameterizations play the key role in both meteorological and air quality simulation particularly in urban areas. The buildings and cement pavement form the urban canopy layers and surface roughness, therefore change thermal and dynamic characteristics of the surface layer. These changes will significantly influence the surface heat accumulation that has a negative effect of the planetary boundary layer (PBL) height and surface wind speed which affect the transport and dispersion of pollutants. The PBL height is a key factor on the formation and evolution of pollutants. A low boundary layer height is regarded as a major meteorological process for haze formation. Thus, a smart selection of physics parameterizations plays a crucial role in urban air quality prediction.
With the development of computer performances, high resolution meteorological and air quality simulations become possible. However, studies demonstrate that the higher resolution does not represent better simulation results, horizontal resolution below 1km is not necessary. Meanwhile, the building clusters modified surface roughness and zero-plane displacement height, an overestimated vertical resolution can cause unreasonable wind flows near surface.
The K-theory shows certain advantages in dealing with the dispersion of air pollution at the mesoscale. It can use the observed wind speed profile data to obtain the concentration distribution of pollutants without assuming a certain form of distribution. However, high pollutants concentrations are frequently found under stable and cold weather conditions, such conditions are characterized by a stratified lower atmosphere and weak turbulent diffusion. In cities, traffic, residential and local industrial emissions have a major share of the pollutants; those emissions can accumulate and have longer residence times near the ground level due to the inefficient transport and mixing. The transient turbulent flow plays an important role on the transportation and deposition of pollutants, thus, an advanced vertical eddy coefficient may helpful to provide an accurate prediction of pollution during the short-term episode. It is noteworthy that meteorological sciences have focused on extreme events to forecast heavy precipitations, thunderstorms, high wind speed conditions while air quality is mainly driven by stable and calm meteorological situations and these latter conditions are difficulty captured by meteorological models.
Different geographical areas will be investigated in France, especially mainly for short-term pollution episodes. The ability of models to assess air quality management strategies is also important, it will be appropriate during this study to perform the diffusion and dispersion of main pollutants and possibly improve the vertical diffusion coefficient in the selected CTM (CHIMERE here) (Menut et al., 2013; Mailler et al., 2017)..
Table of contents :
Chapter I: Introduction
1.1 Air pollutants in the Atmosphere
1.2 Effects of air pollution on climate
1.3 Health effect of air pollution
1.4 The development of atmospheric models
1.5 Urbanization and air quality
1.6 Objectives and outline of the PhD thesis
Chapter II: Meteorology and atmospheric dispersion in the urban canopy– One-year air quality simulations in France using IFS/CHIMERE modeling system
2.1 General principle of the chemical transport model
2.2 Urban meteorology
2.3 Urban canopy modeling
2.4 The basic equations and dispersion in the urban canopy layer
2.5 Preliminary study – Model setup
2.6 Results and Discussions
2.7 Conclusions
Chapter III: Impact of physics parameterizations on high-resolution air quality simulations over urban region
3.1 Introduction
3.2 Model Description and Experiment Design
3.2.1 WRF Model Description
3.2.2 Description of the CHIMERE Model
3.2.3 Domains Setup and Observations Data
3.3. Results and Discussions
3.3.1 Urban Parameters and Nudging Tests
3.3.2 Impact of the Urban Canopy Model
3.3.3 Impact of Mixing Boundary Layer Height
3.3.4 Impact of Land surface model
3.4 Conclusions
Chapter IV: On the impact of WRF-CHIMERE vertical grid resolution and first layer height on mesoscale meteorological and chemistry transport modelling
4.1 Introduction
4.2 Method
4.3. Results and Discussion
4.3.1 Surface meteorological and pollutants concentration
4.3.2 Vertical profiles
4.3.3 Comparison of model results with observations
4.4 Conclusions
Chapter V: Improvement of the vertical mixing in chemistry transport modelling based on a 1.5 order turbulence kinetic energy-based eddy diffusivity closure scheme
5.1 Introduction
5.2 Model Description and Experiment Design
5.2.1 WRF and CHIMERE Model Description
5.2.2 Domains Setup and Observations Data
5.3 Results and Discussions
5.3.1 Impact of the first layer height on NED modelling
5.3.2 NO2 and O3 Simulations over Paris region
5.3.3 PM2.5 and PM10 Simulations over Paris region
5.3.4 Study in Lyon and Bordeaux regions
