Local scale air quality model system for diagnostic and forecasting simulations using the finite element method

Resumen

Air pollution is an important topic with a great social impact; it is related with public health, environment and ecology, and climate change. Scientists have developed several models in the last thirty years, and regional air quality operational systems are used routinely by governments and agencies. Efforts have also been done to simulate the air quality in the local scale; main models are Gaussian and Puff models, that are based on a Lagrangian approach. In contrast with these models, in this thesis we have developed a system using an Eulerian approach. This model is specifically designed for regions with complex orography where the Lagrangian models have problems computing the trajectory of the particles. This model can be used for diagnostic or prediction simulations. Air quality operational systems depend on the orography, meteorological data, and emission data. Air quality models use processors to incorporate these data into the model. The data can come from numerical weather prediction systems, experimental data, or databases. In this thesis we have developed processors, specifically designed for the local scale, to incorporate these data into our system. To incorporate the orography, we have developed a mesh generation algorithm suitable for complex terrain discretization; it also allows to insert layers that can match the regional models. A wind field model has also been used; it can interpolate a three-dimensional wind field from some station measurements using a log-linear vertical wind profile, or can interpolate it from a numerical weather prediction system. Once an interpolated wind field is computed, a mass-consistent model is applied to ensure null divergence and impermeability in the terrain. The wind field is modified to take into account the injection of the pollutants into the atmosphere. Briggs studied the trajectory of the plume rise giving some empirical equations that will be used in our model. Briggs' equations describe the trajectory in a plane; our model will modify this trajectory adapting it to the ambient wind field. This modification allows the plume rise to surround the mountains or channel into the valleys. The transport and reaction of pollutants in the atmosphere is then computed using an splitting method, so the transport and the chemical reactions are computed independently. To solve the transport of pollutants we have used a finite element method stabilized using least squares. The chemical reaction is simulated using simplified models such as RIVAD, or more complex ones such as CB05. To obtain more accurate results we have used adaptation. An error indicator has been used to adapt the mesh to the solution. To adapt the mesh to the concentration distribution of all the species is very demanding, for this reason we have used a multimesh method where every chemical specie has its own mesh where we solve the transport and the chemical reactions are simulated in a common mesh. The system developed in this thesis has diagnostic and forecasting capabilities. For this reason we present two different applications. The first one is a diagnostic application in La Palma island (Spain), where wind measurements are given, and SO2 and NO2 emissions from a stack are considered. The topography of the island is real, from a digital elevation model, but the wind field measurements, and the stack location and emissions, are simulated. The second application is a forecasting application data from the CMAQ benchmark test. It is located in the surrounding of Pineville Kentucky. In this application we have used all the data from CMAQ and the chemical reaction model CB05.