Open AccessA parametrization of 3‐D subgrid‐scale clouds for conventional GCMs: Assessment using A‐Train satellite data and solar radiative transfer characteristics
Author(s) -
Barker Howard W.,
Cole Jason N. S.,
Li Jiangnan,
von Salzen Knut
Publication year - 2016
Publication title -
journal of advances in modeling earth systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.03
H-Index - 58
ISSN - 1942-2466
DOI - 10.1002/2015ms000601
Subject(s) - parametrization (atmospheric modeling) , radiative transfer , cloud fraction , solar zenith angle , environmental science , satellite , atmospheric radiative transfer codes , gcm transcription factors , downwelling , cloud computing , zenith , meteorology , atmospheric sciences , upwelling , physics , remote sensing , geology , computer science , cloud cover , general circulation model , climate change , optics , oceanography , astronomy , operating system
A stochastic algorithm for generating 3‐D cloud fields based on profiles of cloud fraction C and mean cloud water content is presented and assessed using cloud properties inferred from A‐Train satellite data. The ultimate intention is to employ the algorithm, along with 3‐D radiative transfer (RT) models, in Global Climate Models (GCMs). The algorithm approaches cloud fields as whole objects demarcated by contiguous layers with C > 0 . This contrasts with conventional GCM radiation routines that deal with clouds on a per‐(arbitrary) layer basis. A‐Train cloud data for August 2007 were partitioned into ∼29,000 domains, each ∼280 km long, to represent nominal GCM columns. For each A‐Train/stochastic pair of domains, profiles of domain‐averaged fluxes were computed by a 1‐D broadband solar RT model in Independent Column Approximation mode. Globally averaged, mean bias error for upwelling radiation at top‐of‐atmosphere (TOA) is 6.8 W m −2 . Upon advancing the RT model to 2‐D, differences between 1‐D and 2‐D upwelling fluxes at TOA for A‐Train domains differed from corresponding differences for model‐generated domains by ∼1 W m −2 , on average, with differences for the model domains exhibiting stronger dependence on solar zenith angleθ 0 . Moving on to 3‐D RT for model domains, 1‐D–3‐D differences became slightly stronger functions ofθ 0thanks mostly to accentuated 3‐D effects at smallθ 0 . Simple parametrizations for the stochastic algorithm's variables that govern horizontal and vertical structure of clouds should be adequate to capture the ramifications of systematic neglect of 3‐D solar RT in GCMs.