Development of a satellite based PM10 concentration product for urban areas

Dissertação para obtenção do Grau de Doutor em Engenharia do Ambiente

Suspended particulate matter (PM) is one of the most important atmospheric constituents with significant impacts in the climate system and human health. The wide spatiotemporal scale of their effects represents a monitoring challenge that current ground based systems, with limited spatial coverage, cannot fully address. Satellite aerosol optical thickness (AOT) products can complement those monitoring systems by providing, in a periodical basis, spatially resolved ground PM distribution patterns, estimated using linear regression models. Since most satellite AOT products have relevant limitations over high reflective urban areas, where most anthropogenic PM emission sources and world population are concentrated, the development and validation of algorithms to tackle these constraints assumes a paramount importance. The main objective of this thesis was to assess the suitability of one of those alternative AOT retrieval algorithms, based on the contrast reduction effect measured between two images (one reference and one polluted image), for urban PM10 estimation, using MODIS imagery. In order to achieve this objective the work focused primarily on the optimization and validation of a contrast reduction algorithm based on the Differential Texture Analysis (DTA). Several factors were considered such as spatial resolution, reference and polluted viewing geometry similarity and land cover combination to provide more stable and comparable results, while minimizing viewing angle and surface reflectance change effects on algorithm’s accuracy. These optimal AOT retrieval conditions led to a relatively good agreement with the AERONET AOT measurements(r=0.78) for a set of selected European cities and a higher number of valid retrievals when comparing with the MODIS standard aerosol product, supporting the use of this algorithm for urban satellite AOT retrieval. However, the need to define multiple reference images with variable aerosol content as well as the AOT’s coarse spatial resolution limits its ability to extract absolute AOT values towards establishing spatially consistent urban aerosol distribution patterns. The Lisbon Metropolitan air quality stations were used to assess AOT’s ability to reproduce PM10 concentrations. A characterization of the two variables correlation as a function of PM10 time and spatial averaging, season, station type and seasonal reflectance differences between reference and polluted images provided a comprehensive optimal framework for the development of the regression models. This analysis was complemented with the assessment of the relationship of these variables with several other meteorological parameters thus providing additional information to improve PM10/AOT correlation or directly estimate PM10 concentrations. Three PM10 ground concentration estimation regression models were developed to fully assess AOT’s direct capability as the main estimator of daily ground PM10 concentration values. Furthermore, the impact of selected meteorological variables was also assessed using a multivariate approach, enhancing the representation of PM10 dispersion conditions. The univariate regression model, using only AOT as an explanatory variable, was able to reproduce between 37 and 61% (95% confidence interval) of the PM10 ground concentration variance. Values increased to 50-68%, when wind speed was added to the model and to 53-88% when both variables were used to estimate PM10 daily average values, aggregated for all stations. Furthermore the AOT regression coefficients were found to be comparable to other independent studies conducted in different urban regions. PM10 underestimation across all models seem to indicate that the satellite based estimations are more representative of background conditions and have a limited capability to reproduce peak concentrations, usually characteristic of smaller areas heavily influence by point sources (traffic or industry).

Portuguese Foundation for Science and Technology (FCT)- European Union under Operational Program “Science and Innovation” (POCI 2010), PhD grant ref. SFRH/BD/6516/2001