Abstract: Manaus, located in the Central Amazon, is a city of scientific interest because it is surrounded by areas of tropical rainforest. The urban emissions of atmospheric pollutants form an urban plume that can extend throughout tens of kilometers away. In this context, this study aims to estimate the dispersion parameters of the urban plume generated by the city of Manaus using airplane data obtained in situ by the GoAmazon2014/5 experiment and satellite data from the TROPOMI sensor on board the S5P satellite. Obtaining these parameters is a necessary step towards a better understanding of the impacts of the urban plume on the forest and on the regional climate. Average horizontal distributions of air pollutant concentrations were evaluated over an area of about 1x1o in the downwind region of Manaus, during the wet and dry seasons in 2014, when the aerial measurements of the GoAmazon2014/5 experiment were carried out. Three air pollutants were considered: NOx (nitrogen oxides), CO (carbon monoxide) and N (particle number concentration). During the first part of this project, spatially interpolated aircraft data was used from 16 flights carried out during the wet season. In the interpolation, only observations made below 1500m altitude were considered in order to incorporate only data within the Planetary Boundary Layer, in which it is assumed that there is a vertical homogeneity of concentration. Thus, the data can be analyzed considering only latitude and longitude, in a plane. After that, a plume mask model was used to establish which data belong to the plume and which do not. From this, it was possible to develop a method to determine the center line of the plume (which corresponds to the average wind direction) by determining midpoints and performing a linear regression to obtain the center line. Once the center line was determined, concentration data was obtained along this line and an exponential function was fitted to this data. In addition, it is also possible to obtain lines transverse to the central line, whose data was submitted to a gaussian fit. Overall, the pollutant attenuation curves along the central line proved to be compatible with the proposed exponential fit, with an R2 greater than 0.9 in all cases. Among the three pollutants analyzed, NOx had the largest dispersion parameter (in absolute value), indicating that it disperses in the atmosphere near Manaus more rapidly than the other pollutants. This was expected, as NOx is highly reactive and participates, among other reactions, in ozone formation. The pollutant with the lowest rate of concentration change was the particle number concentration. To analyze the concentration along the transverse line, it was sought to determine nearby transverse lines for different pollutants. All pollutants showed a transverse concentration profile that fit well with the Gaussian model, with R2 greater than 0.9, and the transverse dispersion parameters were similar, with NOx having the largest parameter and N the smallest. Finally, an important point to note is that the Gaussian plume model takes into account advection and dispersion processes, without considering other phenomena such as chemical and physical reactions, as in the case of NOx and N. Nevertheless, the results from the fittings indicate that the applied model was compatible with the obtained data, which raises some hypotheses to explain this aspect. It is possible that the time during which pollutants travel the analyzed section of the Manaus plume is shorter than the average time of reactions, so that dispersion predominates. It is worth noting that the work is ongoing, and some of the next steps are the use of satellite data, interpolation of the data using geostatistical analysis techniques; and the use of simplified inversion methods to estimate emission rates.

Keywords: Amazon, air pollutants, urban emissions, atmospheric dispersion.