Asynchronous evolutionary modeling for PM₁₀ spatial characterization

One of the main questions, describing the behavior of a pollutant in the atmosphere, is determining its concentration in some point within a study area, where we come across areas that are difficult to access and where it is impossible to carry out measuring campaigns, or where it is not known with certainty how and in which form the discharge of a pollutant from a source occurs. This without counting, that there is little information, that a group of m_monitoring stations, which monitor the quality of the air, provides about the spatial behavior of the phenomenon. To overcome these problems in an integral way, this article proposes and analyzes a computational model, based on the principles of evolutionary computation (EC), in order to determine the behavior in terms of space and time of the concentration of the particulate matter PM₁₀ within a defined area. The model consists of a solution structure or individual with two submodels or genetic substructures that in turn determines two evolutionary submodels that evolve in an asynchronous manner: an estimate submodel which permits to know the emissions in n_sources based on the principles of a BGPT model (Backward Gaussian puff Tracking) from m_monitoring stations in an inverse way, and a spatial interpolation submodel of the type Takagi Sugeno NUPFS (Non Uniform puffs Functions) in order to determine the spatiotemporal behavior in terms of the analytic representation that defines each one of the puffs emitted from each one of the considered n_sources. In accordance with this structure, the asynchronous evolution mechanism is given mainly by the dependence that the interpolation submodel presents with respect to the estimate submodel, as this fixes and defines the base functions or NUPFS that serve as a base for the interpolator. The proposed evolutionary model was validated using for the estimate a series of concentration measurements for PM₁₀, which were taken starting from a group of m_monitoring stations, which monitor the quality of the air, and starting from a series of n_selected spatial sources within the study area. For the case of the spatial validation, a series of analytic surfaces of concentration for PM10 were obtained from the interpolation model. Each of these surfaces was duly validated by using the CALMET/CALPUFF model and it was validated for each measurement campaign. In this way, the proposed evolutionary model allowed to determine the spatial behavior of the concentration for PM₁₀ in a dynamic way over time, mainly due to the construction which the estimate model uses of the NUPFS base functions applied by the interpolator with reference to the phenomenon.