Simulation of a retrieval process of atmospheric changes from spatially and temporally averaged nadir view spectral changes

Top-of-atmosphere (TOA) nadir view longwave spectral radiance changes caused by perturbations to the atmospheric properties are simulated by a radiative transfer model and cloud fields derived from Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) and CloudSat. The simulation process leads to building spectral radiative kernels. In addition, we compute TOA nadir view spectral radiance change using slightly different cloud fields derived only from Moderate Resolution Imaging Spectroradiometer (MODIS). Treating spectral differences computed from MODIS-derived cloud fields as CLARREO observations, we apply the spectral radiative kernel to derive atmospheric changes using a linear regression. We then quantify the error in estimating atmospheric changes from nadir-view spectral radiance changes to understand the sensitivity of retrieved change to cloud fields. Our earlier study indicates that perturbations to the surface temperature, cloud fraction and height, thin cirrus optical thickness, and upper tropospheric relative humidity cause unique changes to the TOA spectral radiances. In addition, the nadir-view spectral radiance changes are linearly dependent upon the perturbation, and the sum of all the individual spectral radiance changes is equal to the spectral change due to all changes made at once. These are necessary conditions to estimate atmospheric changes by a linear regression. The goal of this study is to understand information content of nadir-view spectral radiance that can be extracted from a linear regression and how it varies depending on cloud fields.