Urban surface models for better aerosol retrieval with MODIS and Landsat

Aerosols are notoriously hard to measure on a global scale since they do not have unique spectral signatures like trace green house gases. Accurate global characterization of Aerosol Optical Depth (AOD) is essential because aerosols are the most uncertain mechanism in climate forecast models, and have known impact on human health. In particular, fine mode particulates (PM2.5) can penetrate deep into the lung tissue contributing to lung damage and cardiac distress. Because of these effects on human health, the Environmental Protection Agency has strict monitoring standards for PM2.5.;Aerosols measurements over urban areas are critical because extended urban centers can have significant aerosol loadings with air quality levels that are above EPA standards. For global studies, satellite measurements are the only realistic approach. Making this monitoring possible from space is the observation that column AOD is quite remarkably related to PM2.5. Dark vegetative surfaces make such correlations strongest and more accurate aerosols retrieval. However, over urban scenes, it is particularly complicated due to the confusion between the ground signal and the aerosol signal. The satellite sensors cannot distinguish if the incoming photons come from the surface or from atmosphere scattering.;For global retrieval of aerosols, the MODIS sensor is perhaps the most suited for global observations, because it can cover almost the entire planet in less than 2 days. The general approach is to use the Long Wave Channel (2130nm) as a good estimate of the surface albedo, since the aerosols contribution in this channel is almost always negligible (especially urban aerosols). Then, the surface albedos in the visible channels, where aerosols are important, can be inferred from empirical relations. However, it has become more apparent that the relations used by MODIS algorithms are not optimized for urban areas and tend to overestimate the AOD. This thesis provides a more extensive study of these relationships. By using high spatial resolution data, we analyzed the urban and vegetation areas separately to determine more accurate relations coefficients between the visible and 2130 nm channels to improve the MODIS algorithm for aerosols detection over urban areas. This work shows that for urban areas like New York City, higher values of these coefficients should be used. We showed that a much better estimation of AOD occurred when using these adjusted coefficients.