Impacts of surface albedo models on high-resolution AOD retrieval

摘要

There is a strong need to improve the resolution of Aerosol Optical depth (AOD) products as more urbanized areas continue to grow. In particular, localized emission sources are likely to create highly localized pollutants that should be monitored. However, in urbanized areas, the land surface itself is a major difficulty since finding dark vegetation pixels becomes harder. Therefore, in order to determine aerosols, a better estimate of the land surface itself should be attempted and should depend strongly on the land surface classification. In order to see if this is possible, we make use of the high density Dragon Network which was deployed in the Washington DC area for summer 2011. The high density of AERONET monitors makes it possible to assess the 3km MODIS AOD retrievals and explore how the deviations of this product depend critically on land surface properties. We then show that we can use improved land surface spectral properties as a function of the different land classes to improve the retrievals. Finally, we explore additional high density cases such as the Dragon Network experiment over Houston from May 1-Nov 1 2013 where the assessment of urban land surface can be better isolated from variations in aerosol class and solar/view geometries. In both cases, sensitivity to urban surface type is observed and magnified when retrieving high resolution AOD products.