Articles | Volume 15, issue 19
Development and technical paper
05 Oct 2022
Development and technical paper |  | 05 Oct 2022

Repeatable high-resolution statistical downscaling through deep learning

Dánnell Quesada-Chacón, Klemens Barfus, and Christian Bernhofer

Related authors

Simulating sub-hourly rainfall data for current and future periods using two statistical disaggregation models – case studies from Germany and South Korea
Ivan Vorobevskii, Jeongha Park, Dongkyun Kim, Klemens Barfus, and Rico Kronenberg
Hydrol. Earth Syst. Sci. Discuss.,,, 2023
Revised manuscript under review for HESS
Short summary
Meteorological, impact and climate perspectives of the 29 June 2017 heavy precipitation event in the Berlin metropolitan area
Alberto Caldas-Alvarez, Markus Augenstein, Georgy Ayzel, Klemens Barfus, Ribu Cherian, Lisa Dillenardt, Felix Fauer, Hendrik Feldmann, Maik Heistermann, Alexia Karwat, Frank Kaspar, Heidi Kreibich, Etor Emanuel Lucio-Eceiza, Edmund P. Meredith, Susanna Mohr, Deborah Niermann, Stephan Pfahl, Florian Ruff, Henning W. Rust, Lukas Schoppa, Thomas Schwitalla, Stella Steidl, Annegret H. Thieken, Jordis S. Tradowsky, Volker Wulfmeyer, and Johannes Quaas
Nat. Hazards Earth Syst. Sci., 22, 3701–3724,,, 2022
Short summary
Modelling evaporation with local, regional and global BROOK90 frameworks: importance of parameterization and forcing
Ivan Vorobevskii, Thi Thanh Luong, Rico Kronenberg, Thomas Grünwald, and Christian Bernhofer
Hydrol. Earth Syst. Sci., 26, 3177–3239,,, 2022
Short summary
An analysis of temporal scaling behaviour of extreme rainfall in Germany based on radar precipitation QPE data
Judith Marie Pöschmann, Dongkyun Kim, Rico Kronenberg, and Christian Bernhofer
Nat. Hazards Earth Syst. Sci., 21, 1195–1207,,, 2021
Short summary
Evapotranspiration at four sites representing land-use and height gradient in the Eastern Ore Mountains (Germany)
Uta Moderow, Stefanie Fischer, Thomas Grünwald, Ronald Queck, and Christian Bernhofer
Hydrol. Earth Syst. Sci. Discuss.,,, 2020
Preprint withdrawn
Short summary

Related subject area

Atmospheric sciences
Metrics for evaluating the quality in linear atmospheric inverse problems: a case study of a trace gas inversion
Vineet Yadav, Subhomoy Ghosh, and Charles E. Miller
Geosci. Model Dev., 16, 5219–5236,,, 2023
Short summary
Improved representation of volcanic sulfur dioxide depletion in Lagrangian transport simulations: a case study with MPTRAC v2.4
Mingzhao Liu, Lars Hoffmann, Sabine Griessbach, Zhongyin Cai, Yi Heng, and Xue Wu
Geosci. Model Dev., 16, 5197–5217,,, 2023
Short summary
Use of threshold parameter variation for tropical cyclone tracking
Bernhard M. Enz, Jan P. Engelmann, and Ulrike Lohmann
Geosci. Model Dev., 16, 5093–5112,,, 2023
Short summary
Passive-tracer modelling at super-resolution with Weather Research and Forecasting – Advanced Research WRF (WRF-ARW) to assess mass-balance schemes
Sepehr Fathi, Mark Gordon, and Yongsheng Chen
Geosci. Model Dev., 16, 5069–5091,,, 2023
Short summary
The High-resolution Intermediate Complexity Atmospheric Research (HICAR v1.1) model enables fast dynamic downscaling to the hectometer scale
Dylan Reynolds, Ethan Gutmann, Bert Kruyt, Michael Haugeneder, Tobias Jonas, Franziska Gerber, Michael Lehning, and Rebecca Mott
Geosci. Model Dev., 16, 5049–5068,,, 2023
Short summary

Cited articles

Alahmari, S. S., Goldgof, D. B., Mouton, P. R., and Hall, L. O.: Challenges for the Repeatability of Deep Learning Models, IEEE Access, 8, 211860–211868,, 2020. a, b, c, d
Allaire, J. J., Ushey, K., Tang, Y., and Eddelbuettel, D.: Reticulate: R Interface to Python, GitHub [code], (last access: 12 December 2021​​​​​​​), 2017. a
Association for Computing Machinery (ACM): Artifact Review and Badging Version 2.0, ACM,, 2021. a
Baño-Medina, J., Manzanas, R., and Gutiérrez, J. M.: Configuration and intercomparison of deep learning neural models for statistical downscaling, Geosci. Model Dev., 13, 2109–2124,, 2020. a, b, c, d, e, f, g, h, i, j, k, l
Short summary
We improved the performance of past perfect prognosis statistical downscaling methods while achieving full model repeatability with GPU-calculated deep learning models using the TensorFlow, climate4R, and VALUE frameworks. We employed the ERA5 reanalysis as predictors and ReKIS (eastern Ore Mountains, Germany, 1 km resolution) as precipitation predictand, while incorporating modern deep learning architectures. The achieved repeatability is key to accomplish further milestones with deep learning.