Improving MISR AOD retrievals over land at medium to high aerosol loadings

<p>Multiangle Imaging SpectroRadiometer (MISR) multi-wavelength and multi-angle observations offer a unique capability for retrieving aerosol optical depth (AOD) and particle properties over most surface types, including land. The excellent quality of MISR aerosol retrievals has been documented in a number of studies. The current MISR land algorithm combines the benefits of two powerful approaches: (1) the heterogeneous surface algorithm (HET) that makes use of surface contrasts and empirical orthogonal functions to represent the surface hemispherical-directional reflectance factors (HDRFs), and (2) the homogenous surface algorithm (HOMOG) that applies spectral and angular invariance constraints to the surface HDRFs. HOMOG was introduced in 2005 to limit occasional outliers and unphysical region-to-region noise in basic HET AOD retrievals. Recently, extensive testing has shown that HOMOG works very well in the low- to medium-AOD range (0.8), but its application under higher aerosol loadings is problematic due to reduced sensitivity to surface reflection. As a result, retrieved MISR AODs over land tend to become more and more underestimated as pollution levels increase. Here we report on a prototype MISR aerosol retrieval algorithm over land that alleviates this AOD underestimation at medium to high aerosol loadings. The improvements introduced to the basic HET algorithm do not require use of the HOMOG constraint at high AODs, thus removing the main cause for AOD underestimation. Simultaneously, the prototype algorithm shows very high AOD retrieval accuracy (quantitatively) as well as spatial consistency (qualitatively). The results are a testament to the physical principles and empirical prowess of the original multi-angle aerosol retrieval algorithm over heterogeneous surfaces.</p>