Professor Roy Grainger

Abstract:

The aerosol component of the Oxford-Rutherford Appleton Laboratory (RAL) Aerosol and Clouds (ORAC) retrieval scheme for the Advanced Along-Track Scanning Radiometer (AATSR) uses data derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) to constrain the brightness of the surface. However, the spectral response functions of the channels used (centred near 550 nm, 660 nm, 870 nm, and 1.6 μm) do not exactly match between the two sensors. It is shown that failure to account for differences between the instruments' spectral response functions leads to errors of typically 0.001-0.01 in spectral surface albedo, and distinct biases, dependent on wavelength and surface type. A technique based on singular value decomposition (SVD) is used to reduce these random errors by an average of 35% at 670. nm and over 60% at the other wavelengths used. The technique reduces the biases so that they are negligible. In principle, the method can be extended to any combination of sensors. The SVD-based scheme is applied to AATSR data from the month of July 2008 and found to increase the number of successful aerosol retrievals, the speed of retrieval convergence, and improve the level of consistency between the measurements and the retrieved state. Additionally, retrieved aerosol optical depth at 550. nm shows an improvement in correspondence when compared to Aerosol Robotic Network (AERONET) data. © 2011 Elsevier Inc.

Abstract:

The aerosol component of the Oxford-Rutherford Appleton Laboratory (RAL) Aerosol and Clouds (ORAC) retrieval scheme for the Advanced Along-Track Scanning Radiometer (AATSR) uses data derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) to constrain the brightness of the surface. However, the spectral response functions of the channels used (centred near 550nm, 660nm, 870nm, and 1.6μm) do not exactly match between the two sensors. It is shown that failure to account for differences between the instruments' spectral response functions leads to errors of typically 0.001–0.01 in spectral surface albedo, and distinct biases, dependent on wavelength and surface type. A technique based on singular value decomposition (SVD) is used to reduce these random errors by an average of 35% at 670nm and over 60% at the other wavelengths used. The technique reduces the biases so that they are negligible. In principle, the method can be extended to any combination of sensors. The SVD-based scheme is applied to AATSR data from the month of July 2008 and found to increase the number of successful aerosol retrievals, the speed of retrieval convergence, and improve the level of consistency between the measurements and the retrieved state. Additionally, retrieved aerosol optical depth at 550nm shows an improvement in correspondence when compared to Aerosol Robotic Network (AERONET) data.

Abstract:

The overlap function of a Raman channel for a lidar system is retrieved by non-linear regression using an analytic description of the optical system and a simple model for the extinction profile, constrained by aerosol optical thickness. Considering simulated data, the scheme is successful even where the aerosol profile deviates significantly from the simple model assumed. Application to real data is found to reduce by a factor of 1.4 – 2.0 the root-mean-square difference between the attenuated backscatter coefficient as measured by the calibrated instrument and a commercial instrument.