North African dust export and deposition: A satellite and model perspective

Abstract

We use a suite of satellite observations (Moderate Resolution Imaging Spectroradiometer (MODIS), Multiangle Imaging Spectroradiometer (MISR), Cloud‐Aerosol Lidar With Orthogonal Polarization (CALIOP)) to investigate the processes of long‐range transport of dust represented in the global GEOS‐Chem model in 2006–2008. A multiyear mean of African dust transport is developed and used to test the representation of the variability in the model. We find that both MODIS and MISR correlate well with the majority of Aerosol Robotic Network observations in the region (r> 0.8). However, MODIS aerosol optical depth (AOD) appears to be biased low (>0.05) relative to MISR in Saharan regions during summer. We find that GEOS‐Chem captures much of the variability in AOD when compared with MISR and MODIS (r> 0.6) and represents the vertical structure in aerosol extinction over outflow regions well when compared to CALIOP. Including a realistic representation of the submicron‐size distribution of dust reduces simulated AOD by ∼25% over North Africa and improves agreement with observations. The lifetime of the simulated dust is typically a few days (25%–50%) shorter than inferred from MODIS observations, suggesting overvigorous wet removal, confirmed by comparison with rain rate observations from the Tropical Rainfall Measuring Mission satellite. The simulation captures the seasonality of deposition in Florida and the observed magnitude and variability of dust concentrations at Barbados from 2006 to 2008 (r = 0.74), indicating a good simulation of the impacts of North African dust on air quality in North America. We estimate that 218 ± 48 Tg of dust is annually deposited into the Atlantic and calculate a lower estimate for the dust deposited in the Caribbean and Amazon to be 26 ± 5 Tg yr−1 and 17 ± 5 Tg yr−1, respectively. This suggests that the dust deposition in the Amazon derived from satellites may be an upper limit.