Preprints
https://doi.org/10.5194/gmd-2022-163
https://doi.org/10.5194/gmd-2022-163
Submitted as: model description paper
18 Jul 2022
Submitted as: model description paper | 18 Jul 2022
Status: this preprint is currently under review for the journal GMD.

Deep learning for stochastic precipitation generation – Deep SPG v1.0

Leroy Bird1, Matthew Walker1, Greg Bodeker1, Isaac Campbell1, Guangzhong Liu1, Swapna Josmi Sam1, Jared Lewis2, and Suzanne Rosier3 Leroy Bird et al.
  • 1Bodeker Scientific, Alexandra, New Zealand
  • 2Climate & Energy College, The University of Melbourne, Parkville, Victoria, Australia
  • 3National Institute of Water and Atmospheric Research, Wellington, New Zealand

Abstract. We present a deep neural network based single site stochastic precipitation generator (SPG), capable of producing realistic time series of daily and hourly precipitation. The neural network outputs a wet day probability and precipitation distributions in the form of a mixture model. The SPG was tested in four different locations in New Zealand, and we found it accurately reproduced the precipitation depth, the autocorrelations seen in the original data, the observed dry-spell lengths and the seasonality in precipitation. We present two versions of the hourly and daily SPGs: (i) a stationary version of the SPG that assumes that the statistics of the precipitation are time independent (ii) a non-stationary version that captures the secular drift in precipitation statistics resulting from climate change. The latter was developed to be applicable to climate change impact studies, especially, studies reliant on SPG projections of future precipitation. We highlight many of the pitfalls associated with the training of a non-stationary SPG on observations alone, and offer an alternative method that replicates the secular drift in precipitation seen in a large-ensemble regional climate model. The SPG runs several orders of magnitude faster than a typical regional climate model, and permits the generation of very large ensembles of realistic precipitation time series under many climate change scenarios, these ensembles will also contain many extreme events not seen in the historical record.

Leroy Bird et al.

Status: open (until 12 Sep 2022)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse

Leroy Bird et al.

Model code and software

SPG code Leroy Bird, Matthew Walker https://doi.org/10.5281/zenodo.6801733

Leroy Bird et al.

Viewed

Total article views: 236 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
196 36 4 236 4 5
  • HTML: 196
  • PDF: 36
  • XML: 4
  • Total: 236
  • BibTeX: 4
  • EndNote: 5
Views and downloads (calculated since 18 Jul 2022)
Cumulative views and downloads (calculated since 18 Jul 2022)

Viewed (geographical distribution)

Total article views: 218 (including HTML, PDF, and XML) Thereof 218 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 08 Aug 2022
Download
Short summary
Deriving the statistics of expected future changes in extreme precipitation is challenged by these events being rare. Regional climate models (RCMs) are computationally prohibitive for generating ensembles capable of capturing large numbers of extreme precipitation events with statistical robustness. Stochastic precipitation generators (SPGs) provide an alternative to RCMs. This paper describes a novel single-site SPG that learns the statistics of precipitation using a machine learning approach.