Catálogo de publicaciones - libros

Until the recent decade, air quality forecasts have been largely based on statistical modeling techniques. There have been significant improvements and innovations made to these statistically based air quality forecast models during past years (Ryan et al., 2000). Forecast fidelity has improved considerably using these methods. Nonetheless, being non-physically-based models, the performance of these models can vary dramatically, both spatially and temporally. Recent strides in computational technology and the increasing speed of supercomputers, combined with scientific improvements in meteorological and air quality models has spurred the development of operational numerical air quality prediction models (e.g., Vaughn et al., 2004, McHenry et al., 2004).

Recent extensive observations of the nocturnal boundary layer, such as taken in CASES99 (Poulos et al., 2001) and SABLE98 (Cuxart et al., 2000), have revealed important deficiencies in our ability to model the strongly stratified nocturnal boundary layer. Even with more or less continuous and relatively strong turbulence near the surface, the primary source of turbulence can be shear above the surface inversion layer (Mahrt, 1999; Mahrt and Vickers, 2002), often related to a low-level wind maximum (Cuxartet al., 2000; Banta et al., 2002, 2003). Consequently for the CASES99 data, mixing formulations with a – limit perform better than traditional parameterizations based on a definable boundarylayer depth (Mahrt and Vickers, 2003).

The ensemble-averaging approach is potentially a technique for improving the performance of real-time photochemical air-quality modeling. Ensemble photochemical air-quality forecasts are tested extensively using the Community Multiscale Air Quality (CMAQ) model-system with mesonet observations from the Emergency Weather Net (EmWxNet) and the AQ Data Set over the Lower Fraser Valley (LFV). The CMAQ model is run daily over a 12 km resolution domain (Figure 1 top) covering southern British Columbia, Washington State, and the northern portion of Oregon State. A 4 km resolution grid (Figure 1 bottom) is nested within the 12km grid, and it covers the southern tip of British Columbia (including Vancouver and the LFV) and the northern part of Washington State (including the Seattle area). CMAQ is driven by two different meteorological models: the Mesoscale Compressible Community Model (MC2), and the Fifth-Generation NCAR/Penn State Mesoscale Model (MM5).

The GEM-AQ model (Global Environmental Multiscale model with Air Quality processes) is based on Canada’s operational weather prediction model developed by the Meteorological Services of Canada (MSC). The chemical module is included in the host meteorological model “online”, so that the chemical species are advected at each dynamical timestep and the meteorological information can be used in the chemical process calculations. The processes include over 100 gas phase chemical reactions, anthropogenic and biogenic emissions, transport due to vertical diffusion and convection, dry deposition and wet scavenging. We have recently introduced size-resolved aerosols which undergo processes such as coagulation, nucleation, dry and wet scavenging. GEM-AQ is capable of running on a global uniform or global variable resolution domain. The variable resolution capability allows a high resolution simulation over an area of interest without the computational overhead of a global high resolution domain, as well as eliminating the need for boundary conditions in the case of limited area modeling. The capabilities of this modelling system give us a tool to help examine local air quality issues while keeping the global picture in mind. The design philosophy behind GEM-AQ allows for the integration of different physics and chemistry modules into a single computational platform. The major advantage of this approach allows for the development and improvement of parameterizations which can easily be included in the system. In addition, the use of GEMAQ as the core of the modelling system will permit the incorporation of data assimilation techniques into the model validation and application studies on processes by researchers participating in the Multiscale Air Quality Modelling Network (MAQNet).

Backward trajectories from Buenos Aires, Argentina, were used in an exploratory study to characterize the airflow and the relationship with synoptic patterns and to contribute to the identification of air pollution source regions. Buenos Aires is the most populated city of the country. Air pollution climatology studies have shown that there is an important frequency of reduced ventilation conditions, especially during the cold season (Ulke, 2000). Case studies have demonstrated the transport of biomass burning products from Brazil during the tropical dry season toward the region (Longo et al, 1999).

Time series decomposition methods were applied to meteorological and air quality data and their numerical model estimates. Decomposition techniques express a time series as the sum of a small number of independent modes which hypothetically represent identifiable forcings, thereby helping to untangle complex processes. Mode-to-mode comparison of observed and modeled data provides a mechanism for model evaluation.

In the last three decades, in order to model pollutant dispersion inside a street canyon, several models have been proposed to describe mean concentration and retention time of pollutant inside the canyon, in function of the flow dynamics of the external flow (Berkowicz, 2000, Soulhac, 2000, De Paul and Sheih, 1986, Caton et al., 2003). Some of these models take account just of the mean velocity at the roof height Uh, some other do also consider the turbulence intensity of the incoming flow.

Inverse problems is one of comparably new and quickly developing areas. The problems themselves were known for ages but their exact solutions in many cases were either non-existing or requiring a forbidding amount of computations. It is a development of both mathematical methods (first of all, regularization techniques and statistical optimal filters) and new generations of computers that made some of the inverse problems approachable. In the field of atmospheric dispersion, an example of inverse problem is a so-called “compliance regime control” problem introduced by the Convention on Long-Range Transboundary Air Pollution (LRTAP). The task is to evaluate the true emission of primary acid pollutants in Europe from long-term observations and their comparison with model simulations. Host (1996) has shown that such a problem can be approached at a qualitative level only, which has been confirmed by another attempt of Sofiev & Sofieva (2000). In both papers, it is shown that without a heavy use of a-priory assumptions on emission values the uncertainties by far exceed the values themselves, while utilization of a-priory information destroys the signal from the observations together with the noise.

Dispersion at microscale should be simulated using 3D codes, since they are able to account for very complex situations. Targets of these models can be road traffic simulations, such as the reconstruction of the high pollution episodes into street canyons or car parks or, more generally, the reconstruction of dispersion patterns generated by the presence of obstacles. These problems are traditionally solved using CFD models adapted to the atmospheric PBL, suitable only for short term simulations in a limited number of cases, due to their high CPU demand. Here we present a different approach to reproduce the microscale scenarios with a lower demand of computational time. It potentially allows a wider range of applications, such as emergency responses, statistical calculations of impacts due to different conditions and climatological or long term impacts. This method allows an exact representation of buildings directly generated by a GIS (as .shp files). A first guess of the 3D mean flow is computed using all available and relevant meteorological data (inside or outside the target domain). This field is then modified using analytical corrections due to the obstacles (, 1996).

In this work recent developments in the Regional Atmospheric Modelling System (RAMS) version 4.3 (Cotton et al., 2003) coupled with modules describing the atmospheric mercury cycle are presented. The mechanisms that describe reemission of Hg0 from water and soil as well as the wet and dry deposition of mercury species have been implemented in RAMS. A detailed chemistry module has also been tested. The model performance has been examined for a representative summer week during which observations of the wet deposited mercury were available. Model calculations have been compared with observations made during the 14 to 26 August 1997 simulation period in NE USA. An inter-comparison was also performed with the previous version of the model (Voudouri et al., 2004) used over the same domain, and simulation period.