Cloud Property Retrieval Based on DISAMAR Using Oxygen-A and Oxygen-B Band Measurements by TROPOMI on Sentinel-5P

Determining instrument specifications and analyzing methods for atmospheric retrieval (DISAMAR) is a radiative-transfer (RT) model developed to simulate retrievals of atmospheric components and surface properties. This article evaluates the performance of the DISAMAR retrieval algorithm using Tropospheric Monitoring Instrument (TROPOMI) spectral measurements of the O2-A (757–770 nm) and O2-B (685–691 nm) bands for cloud-free scenes and fully cloudy scenes. For fully cloudy scenes, cloud pressure and cloud optical thickness (COT) are retrieved and compared with Fast Retrieval Scheme for Clouds from the O2-A band-Sentinel (FRESCO-S), Retrieval of Cloud Information Using Neural Network (ROCINN), and Visible Infrared Imager Radiometer Suite (VIIRS) cloud products. For cloud-free scenes, the surface pressure and surface albedo are fit; the results are compared with ERA5 surface pressure and TROPOMI DLER datasets, respectively. Several parameters in DISAMAR retrieval settings are tested, including the stray-light settings, the fitting wavelength windows, the a priori values of the fitting parameters, and the performance of oxygen cross-section datasets (JPL2008, HITRAN 2008/2012/2020). As expected, DISAMAR retrieves cloud mid-level pressure (CLP) values closely matching FRESCO-S and ROCINN, especially when fitting the O2-A band. COT and cloud-top pressure (CTP) show better correlation with VIIRS data when fitting the O2-A band than the O2-B band. For thick clouds, fitting stray light is not advisable, and a specific a priori value of COT is unnecessary. Using the HITRAN2020 dataset provides more accurate spectral simulations. Fitting the O2-A band yields results closer to both FRESCO-S and VIIRS, while the O2-B band yields results closer to ROCINN and serves as a useful supplement. Excluding high-residual wavelengths in the fitting window further improves retrievals.

https://ieeexplore.ieee.org/document/11298292