Achievements of Three Year Aerosol_cci Work
Holzer-Popp, Thomas1; de Leeuw, Gerrit2; Bingen, Christine3; Fischer, Jürgen4; Kinne, Stefan5; North, Peter6; Poulson, Caroline7; Poulson, Caroline7; Ramon, Didier8; Schulz, Michael9; Stebel, Kerstin10; Tanre, Didier11; Thomas, Gareth12; Veefkind, Pepijn13; Veefkind, Pepijn13; Vountas, Marco14; Zieger, Paul15; Pinnock, Simon16
1DLR, GERMANY; 2FMI, FINLAND; 3BIRA, BELGIUM; 4FUB, GERMANY; 5MPI, GERMANY; 6Swansea University, UNITED KINGDOM; 7RAL, UNITED KINGDOM; 8HYGEOS, FRANCE; 9MetNo, NORWAY; 10NILU, NORWAY; 11LOA, FRANCE; 12Oxford University, UNITED KINGDOM; 13KNMI, NETHERLANDS; 14Bremen University, GERMANY; 15PSI, SWITZERLAND; 16ESA-ESRIN, ITALY
Within the ESA Climate Change Initiative (CCI) the Aerosol_cci project (mid 2010 - mid 2013) has conducted intensive work to improve algorithms for the retrieval of aerosol information from European sensors AATSR (3 algorithms), PARASOL, MERIS (3 algorithms), synergetic AATSR/SCIAMACHY, OMI and GOMOS. Whereas OMI and GOMOS were used to derive absorbing aerosol index and stratospheric extinction profiles, respectively, Aerosol Optical Depth (AOD) and Angstrom coefficient were retrieved from the other sensors. Global datasets for 2008 were produced and validated versus independent ground-based data and other satellite data sets (MODIS, MISR). An additional 17-year dataset will be produced using ATSR-2/AATSR data.
During the three years of the project, intensive collaborative efforts were made to improve the retrieval algorithms focusing on the most critical modules. The team agreed on the use of a common definition for the aerosol optical properties. Cloud masking was evaluated, but a rigorous analysis with a pre-scribed cloud mask did not lead to improvement for all algorithms. Better results were obtained using a post-processing step in which sudden transitions, indicative of possible occurrence of cloud contamination, were removed. Surface parameterization, which is most critical for the nadir only algorithms (MERIS and synergetic AATSR/SCIAMACHY) was studied to a limited extent.
The retrieval results for AOD, ångstrom exponent (AE) and uncertainties were evaluated by comparison with data from AERONET (and a limited amount of MAN) sun photometer and with satellite data avaialble from MODIS and MISR. Both level2 and level3 (gridded daily) datasets were validated. Several validation metrics were used (standard statistical quantities such as bias, rmse, Pearson correlation, linear regression, as well as scoring approaches to quantitatively evaluate the spatial and temporal correlations against AERONET), and in some cases developed further, to evaluate the datasets and their regional and seasonal merits. The validation showed that most datasets have improved significantly and in particular PARASOL (ocean only) provides excellent results. The metrics for AATSR (land and ocean) datasets are similar to those of MODIS and MISR, with AATSR better in some regions and less good in some others. However, AATSR coverage is smaller than that of MODIS due to swath width. The MERIS dataset provides better coverage than AATSR but has lower quality (especially over land) than the other datasets. Also the synergetic AATSR/SCIAMACHY dataset has lower quality. The evaluation of the pixel uncertainties shows first good results but also reveals that more work needs to be done to provide comprehensive information for data assimilation. Users (MACC/ECMWF, AEROCOM) confirmed the relevance of this additional information and encouraged Aerosol_cci to release the current uncertainties.
A thorough inter-comparison was conducted for the three AATSR algorithms. Care was taken to compare equal data amounts by common point filtering. The analysis suffered from limitations of the available reference datasets over open ocean and in the Southern hemisphere. The validation shows that all three AATSR algorithms produce almost equal accuracy, but with differences in the resulting datasets (similar to those between MODIS and MISR).
The paper will summarize and discuss the results of three year work in Aerosol_cci and conclude with lessons learned and recommendations for the future.