Spatial Representativeness of the Bogota Air Quality Monitoring Network

摘要

Based on the mutual information results, the most representative stations are Las Ferias, Parque Simon Bolivar and Puente Aranda which are located at the geographical center of Bogota. Conversely the stations located at the outskirts share a small amount of information, a fact that makes them relatively singular. These stations are Tunal, Carvajal, Usaquen and Suba. The stations that share most information represent common areas and/or similar meteorological patterns, especially wind speed and wind direction. This fact was confirmed based on the results from Lagrangian simulations. The mutual information method provides valuable approximations in order establish the spatial representativeness of a monitoring network such as the RMCAB. However, it lacks of a model that represents the chemical and physical behavior of the atmosphere. Therefore, the mutual information method by itself does not provide enough tools to observe in an objective way the spatial coverage of the RMCAB. Instead it gives insight of the stations that would potentially represent common features from the perspectives of air quality, spatial representativeness and meteorology. The results must be supplemented by means of the introduction of simulation tools that include meteorological and dispersion variables, in such a way that it would be feasible to observe in detail spatial and temporal variations. The Lagrangian simulations from STILT generated the influence footprints of the RMCAB stations, which are associated with the spatial representativeness of the measurement sites. The distribution of the influence has high spatial and temporal variability when it is evaluated in punctual moments and allows the identification of the different zones that are not covered by the network. It was observed that it is not possible to assign an equidistant influence from the monitoring sites because it constantly changes in direction, form and intensity. There are zones that in certain moments are not covered despite being located at the surroundings of the receptors. This fact warns about possible mistaken interpretations that could be deduced when analyzing air quality measurements. In concordance with the results provided by the mutual information method, the STILT footprints show that the geographical center of Bogota tends to be adequately represented most of the time. The explanation is associated with the relatively high number of stations located around this zone. These particularities are associated with common features in terms of meteorological behavior and soil use. Conversely, the footprints of the stations located at the outskirts reflect singular meteorological behaviors, which is concordant with the mutual information values. Sometimes the measurement sites located far from the center of the domain represent rural areas with low population densities which prevent an objective evaluation of the Bogota air quality and a poor evaluation of public policies. In summary, the evaluation of the spatial representativeness of the RMCAB requires methodologies that involve time periods of different length. It was demonstrated the mutual information method allows the evaluation of temporal series of considerable length and the Lagrangian simulations provide information about specific episodes. Both perspectives are concordant and complementary and they have the potential to be unified in a sole methodology. The inclusion of meteorological information is essential and therefore it is important that the numerical weather prediction systems reasonably represent the atmospheric physics of the geographical zones with the objective of reducing uncertainties in dispersion models.