Preprints
https://doi.org/10.5194/gmd-2022-76
https://doi.org/10.5194/gmd-2022-76
Submitted as: development and technical paper
23 Mar 2022
Submitted as: development and technical paper | 23 Mar 2022
Status: this preprint is currently under review for the journal GMD.

Downscaling Atmospheric Chemistry Simulations with Physically Consistent Deep Learning

Andrew Geiss1, Sam Silva1,2, and Joseph Hardin1,3 Andrew Geiss et al.
  • 1Pacific Northwest National Laboratory, Richland, WA, USA
  • 2University of Southern California, Los Angeles, CA, USA
  • 3ClimateAi, Inc. San Francisco, CA, USA

Abstract. Recent advances in deep convolutional neural network (CNN) based super resolution can be used to downscale atmospheric chemistry simulations with substantially higher accuracy than conventional downscaling methods. This work both demonstrates the downscaling capabilities of modern CNN-based single image super resolution and video super resolution schemes and develops modifications to these schemes to ensure they are appropriate for use with physical science data. The CNN-based video super resolution schemes in particular incur only 39 % to 54 % of the grid-cell-level error of interpolation schemes and generate outputs with extremely realistic small-scale variability based on multiple perceptual quality metrics while performing a large (8 x 10) increase in resolution in the spatial dimensions. Methods are introduced to strictly enforce physical conservation laws within CNNs, perform large and asymmetric resolution changes between common model grid resolutions, account for non-uniform grid cell areas, super resolve log-normally distributed datasets, and leverage additional inputs such as high-resolution climatologies and model state variables. High-resolution chemistry simulations are critical for modeling regional air quality and for understanding future climate, and CNN-based downscaling has the potential to generate these high resolution simulations and ensembles at a fraction of the computational cost.

Andrew Geiss et al.

Status: open (until 24 Jun 2022)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • CC1: 'Comment on gmd-2022-76', Patrick Obin Sturm, 24 Mar 2022 reply
    • AC3: 'Reply on CC1', Andrew Geiss, 06 May 2022 reply
  • CEC1: 'Comment on gmd-2022-76', Juan Antonio Añel, 25 Apr 2022 reply
    • AC1: 'Reply on CEC1', Andrew Geiss, 29 Apr 2022 reply
  • AC2: 'Comment on gmd-2022-76', Andrew Geiss, 29 Apr 2022 reply
  • RC1: 'Comment on gmd-2022-76', Anonymous Referee #1, 12 May 2022 reply

Andrew Geiss et al.

Model code and software

Atmos. Chem. Downscaling CNN Andrew Geiss, Sam J. Silva, Joseph C. Hardin https://github.com/avgeiss/chem_downscaling

Video supplement

Ozone Super Resolution Andrew Geiss, Sam J. Silva, Joseph C. Hardin https://youtu.be/JPJX1k-5yew

Andrew Geiss et al.

Viewed

Total article views: 617 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
470 129 18 617 4 4
  • HTML: 470
  • PDF: 129
  • XML: 18
  • Total: 617
  • BibTeX: 4
  • EndNote: 4
Views and downloads (calculated since 23 Mar 2022)
Cumulative views and downloads (calculated since 23 Mar 2022)

Viewed (geographical distribution)

Total article views: 566 (including HTML, PDF, and XML) Thereof 566 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 20 May 2022
Download
Short summary
This work demonstrates using modern machine learning techniques to enhance the resolution of atmospheric chemistry simulations. We evaluate the schemes for an 8 x 10 increase in resolution and find that they perform substantially better than conventional methods. Methods are introduced to target the machine learning methods towards this type of problem, most notably, by ensuring they do not break known physical constraints.