Parameterization of shortwave horizontal photon transport effects

 

Authors

Jerry Y. Harrington — Pennsylvania State University
Tamas Varnai — University of Maryland, Baltimore County

Category

Radiation

Description

In order to keep the computational demands of radiative calculations manageable, most dynamical atmospheric simulations use one-dimensional (1D) radiation models. These 1D models calculate radiative processes separately for each atmospheric column, without considering horizontal interactions between neighboring columns. As several studies and our earlier results have indicated, these horizontal interactions can lead to significant inaccuracies in 1D estimates of radiative heating. To help improve the accuracy of radiation calculations in regional and global cloud-resolving simulations, we are developing a fast parameterization that can adjust 1D radiation model outputs by considering horizontal radiative interactions. This parameterization provides corrections to 1D radiation model outputs by using a neural net, which estimates the needed corrections based on horizontal gradients in the dynamical simulation’s cloud field. The poster will discuss this neural net approach and will show results about the corrections improving the accuracy of 1D radiation calculations.