NOAA GFDL - Geophysical Fluid Dynamics LaboratoryGFDL - Geophysical Fluid Dynamics Laboratory

Assessing Clouds in GFDL’s AM4.0 Using Alternative Microphysical Representations and Satellite Simulator Diagnostics

May 13th, 2026

Key Findings

  • Simulated cloud properties in the GFDL atmospheric model AM4.0 depend on the representation of cloud microphysical processes.
  • The Cloud Feedback Model Intercomparison Project Observation Simulator Package (COSP) allows comparison of model results with satellite observations using the same observational perspective.
  • Different microphysical representations influence simulated cloud amount, vertical structure, and radiative properties.
  • Satellite-based comparisons help evaluate how these modeling choices affect simulated cloud fields.

Huan Guo, Levi G. Silvers, David Paynter, Wenhao Dong, Songmiao Fan, Xianwen Jing, Ryan Kramer, Kristopher Rand, Kentaroh Suzuki, Yuying Zhang, Ming Zhao. Geophysical Research Letters. DOI: 10.1029/2024EA004053

Clouds affect Earth’s climate by influencing how sunlight and heat move through the atmosphere, as well as by producing precipitation. Because many cloud processes occur at scales smaller than the resolution of climate models, these processes are represented using simplified descriptions of cloud formation and evolution, and precipitation development.

In this study, the authors examine how different representations of cloud microphysics affect simulated cloud properties in the GFDL atmospheric model AM4.0. The analysis uses the Cloud Feedback Model Intercomparison Project Observation Simulator Package (COSP), which produces satellite-like measurements from model simulations. This approach allows model results to be evaluated using the same quantities measured by satellite instruments.

The findings demonstrate that changing how cloud microphysical processes are represented affects several aspects of simulated clouds, including cloud coverage, vertical distribution, and radiative characteristics. Comparisons with satellite observations help identify how these modeling choices influence simulated cloud fields. Specifically, these results detail how different cloud process representations influence simulated cloud behavior within the AM4.0 model, providing valuable insights for evaluating cloud simulations across atmospheric models.

 

This figure compares the total cloud fraction derived from four satellite observations—MODIS, ISCCP, CALIPSO, and MISR (Panels a-d)—with corresponding simulations from the AM4.0 model (Panels e-h) and the AM4-MG2 model (Panels i-l), both using COSP satellite simulators. The figure also quantifies the differences, or biases, between the AM4.0 simulation and the observations (Panels m-p), and between the AM4-MG2 simulation and the observations (Panels q-t). The top three rows present annual and global averages of cloud fraction, combining both land and ocean data, with ocean-only values provided in parentheses. The bottom two rows display the model biases and Root-Mean-Square Errors (RMSEs), also combining land and ocean data, with ocean-only values indicated parenthetically.