A study on radiative transfer effects in 3‐D cloudy atmosphere using satellite data

Abstract

This study evaluates 3‐D cloud effects on the radiation budget with a combined use of active sensor cloud profiling radar/CloudSat and imager Moderate Resolution Imaging Spectroradiometer/Aqua data on the A‐train. An algorithm is devised for constructing 3‐D cloud fields based on satellite‐observed cloud information. The 3‐D cloud fields thus constructed are used to calculate the broadband solar and thermal radiative fluxes with a 3‐D radiative transfer code developed by the authors. The aim of this study is to investigate the effects of cloud morphology on solar radiative transfer in cloudy atmosphere. For this purpose, 3‐D cloud fields are constructed with the new satellite‐based method, to which full 3D‐RT (radiative transfer) simulations are applied. The simulated 3‐D radiation fields are then used to examine and quantify errors of existing typical plane‐parallel approximations, i.e., Plane‐Parallel Approximation, Independent Pixel Approximation and Tilted Independent Pixel Approximation. Such 3D‐RT simulations also serve to address another objective of this study, i.e., to devise an accurate approximation and to characterize the observed specific 3D‐RT effects by the cloud morphology based on knowledge of idealized 3D‐RT effects. We introduce a modified approach based on an optimum value of diffusivity factor to better approximate the radiative fluxes for arbitrary solar zenith angle determined from the results of 3‐D radiative transfer simulations to redeem the overcorrections of these approximations for large solar zenith angles (SZAs). This new approach, called Slant path Independent Pixel Approximation, is found to be better than other approximations when SZA is large for some cloud cases. Based on the SZA dependence of the errors of these approximations relative to 3‐D computations, satellite‐observed real cloud cases are found to fall into either of three types of different morphologies, i.e., isolated cloud type, upper cloud‐roughened type and lower cloud‐roughened type. Such a classification offers a novel insight into error characteristics of the approximations that are interpreted in the context of specific cloud morphology.

Analysis of 3D cloud effects in OCO-2 XCO2 retrievals

“…Várnai and Marshak (2009) demonstrated that MODIS reflectance at various wavelengths between 0.47 and 2.12 µm increases as cloud distances decrease at cloud distances less than 10 km, and the effect is strongest at shorter wavelengths. Okata et al (2017) modeled 3D cloud effects, finding positive 3D-1D radiance differences at solar zenith angles greater than 5 • for periodic cuboid clouds of 2.5 km height. Merrelli et al (2015) applied the SHDOM 3D radiative transfer code and the OCO-2 retrieval code, and they concluded that the OCO-2 cloud screening algorithm had difficulty in rejecting clouds that filled less than half of the field of view.…”

Observational evidence of 3‐D cloud effects in OCO‐2 CO2 retrievals

“…From a 1‐D (IPA) perspective in the absence of clouds, an increase in CO 2 leads to a decrease in molecular line radiances. This is consistent with the case (a) calculations of Okata et al []. Three‐dimensional clouds, however, can enhance the radiation field of a clear‐sky footprint when clouds are several kilometers away from a clear‐sky footprint and/or cloud fragments are located in the footprint, perturbing continuum, and molecular line radiances, as a function of wavelength and total optical depth.…”

“…The recent study of Okata et al [2017] points out the different signed effects of 3-D radiative transfer that are possible, dependent upon several types of idealized cloud morphology. They calculated 3-D radiative transfer effects for three types of low-level clouds: (a) periodic cuboid clouds of 2.5 km height with no material between the clouds, (b) periodic cuboid clouds with near surface cloud opacity between the clouds, and (c) periodic voids with lower material between and above the voids.…”

“…They calculated 3-D radiative transfer effects for three types of low-level clouds: (a) periodic cuboid clouds of 2.5 km height with no material between the clouds, (b) periodic cuboid clouds with near surface cloud opacity between the clouds, and (c) periodic voids with lower material between and above the voids. Figure 4 of Okata et al [2017] presents IPA-3D top of atmosphere flux differences as a function of solar zenith for the three cases. For case (a), which corresponds most closely to the cloud fields we study, the IPA-3D differences are negative for solar zenith angles greater than 5°, and positive (though small) for case (b) clouds.…”

Analysis of 3D Cloud Effects in OCO-2 XCO2 Retrievals

“…Várnai and Marshak (2009) demonstrated that MODIS reflectance at various wavelengths between 0.47 and 2.12 µm increases as cloud 105 distances decrease at cloud distances less than 10 km, and that the effect is strongest at shorter wavelengths. Okata et al (2017) modeled 3D cloud effects, finding positive 3D-1D radiance differences, for solar zenith angles greater than 5°, for periodic cuboid clouds of 2.5 km height. Merrelli et al (2015) applied the SHDOM 3D radiative transfer code, and the OCO-2 retrieval code, and concluded that the OCO-2 cloud-screening algorithm 110 had difficulty in rejecting clouds that filled less than half of the field of view.…”