Submitted as: model description paper 28 Oct 2020

Submitted as: model description paper | 28 Oct 2020

Review status: a revised version of this preprint is currently under review for the journal GMD.

Latent Linear Adjustment Autoencoders v1.0: A novel method for estimating and emulating dynamic precipitation at high resolution

Christina Heinze-Deml1, Sebastian Sippel1,2, Angeline G. Pendergrass3,2, Flavio Lehner2,3, and Nicolai Meinshausen1 Christina Heinze-Deml et al.
  • 1Seminar for Statistics, ETH Zurich, Zurich, Switzerland
  • 2Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland
  • 3Climate and Global Dynamics Laboratory, National Center for Atmospheric Research, Boulder, Colorado

Abstract. A key challenge in climate science is to quantify the forced response in impact-relevant variables such as precipitation against the background of internal variability, both in models and observations. Dynamical adjustment techniques aim to remove unforced variability from a target variable by identifying patterns associated with circulation, thus effectively acting as a filter for dynamically-induced variability. The forced contributions are interpreted as the variation that is unexplained by circulation. However, dynamical adjustment of precipitation at local scales remains challenging because of large natural variability and the complex, nonlinear relationship between precipitation and circulation particularly in heterogeneous terrain. Building on variational autoencoders, we introduce a novel statistical model – the Latent Linear Adjustment Autoencoders v1.0 – that enables estimation of the contribution of a coarse-scale atmospheric circulation proxy to daily precipitation at high-resolution and in a spatially coherent manner. To predict circulation-induced precipitation, the Latent Linear Adjustment Autoencoders v1.0 combines a linear component, which models the relationship between circulation and the latent space of an autocoder, with the autoencoder's nonlinear decoder. The combination is achieved by imposing an additional penalty in the cost function that encourages linearity between the circulation field and the autoencoder's latent space, hence leveraging robustness advantages of linear models as well as the flexibility of deep neural networks. We show that our model predicts realistic daily winter precipitation fields at high resolution based on a 50-member ensemble of the Canadian Regional Climate Model at 12-km resolution over Europe, capturing for instance key orographic features and geographical gradients. Using the Latent Linear Adjustment Autoencoders v1.0 to remove the dynamic component of precipitation variability, forced thermodynamic components are expected to remains in the residual, which enables the uncovering of forced precipitation patterns of change from just a few ensemble members. We extend this to quantify the forced pattern of change conditional on specific circulation regimes. In addition, we briefly illustrate one of multiple possible further applications of the method: a weather generator that emulates climate model simulations of regional precipitation at high resolution by bootstrapping circulation patterns. Other potential applications include addressing detection&attribution at sub-continental scales, statistical downscaling and transfer learning between models and observations to exploit the typically much larger sample size in models compared to observations.

Christina Heinze-Deml et al.

Status: final response (author comments only)
Status: final response (author comments only)
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
[Login for authors/topical editors] [Subscribe to comment alert] Printer-friendly Version - Printer-friendly version Supplement - Supplement

Christina Heinze-Deml et al.

Data sets

Original CRCM5-LE dataset description and data access provided by the ClimEx project Martin Leduc, Alain Mailhot, Anne Frigon, Jean-Luc Martel, Ralf Ludwig, Gilbert B. Brietzke, Michel Giguère, François Brissette, Richard Turcotte, Marco Braun, and John Scinocca

Sample data of the CRCM5-LE for applications of the Latent Linear Adjustment autoencoder Christina Heinze-Deml, Sebastian Sippel, Angeline G. Pendergrass, Flavio Lehner, and Nicolai Meinshausen

Model code and software

Pretrained Models and Link to Github Code archive Christina Heinze-Deml, Sebastian Sippel, Angeline G. Pendergrass, Flavio Lehner, and Nicolai Meinshausen

Christina Heinze-Deml et al.


Total article views: 632 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
413 212 7 632 5 7
  • HTML: 413
  • PDF: 212
  • XML: 7
  • Total: 632
  • BibTeX: 5
  • EndNote: 7
Views and downloads (calculated since 28 Oct 2020)
Cumulative views and downloads (calculated since 28 Oct 2020)

Viewed (geographical distribution)

Total article views: 474 (including HTML, PDF, and XML) Thereof 472 with geography defined and 2 with unknown origin.
Country # Views %
  • 1
Latest update: 14 Jun 2021
Short summary
Building on deep learning based variational autoencoders, we develop a novel statistical model that enables estimating daily precipitation at high resolution from a coarse-scale atmospheric circulation proxy (sea level pressure). The model predicts realistic daily winter precipitation fields in a 50-member regional climate model ensemble (CRCM5-LE) over Central Europe, capturing key orographic features, and helps to better isolate and analyse dynamically-induced precipitation variability.