JRA55-based repeat year forcing datasets for driving ocean-sea-ice models

• Published in: Ocean Modelling
• Main Authors: Stewart, Kial, Kim, W. M., Urakawa, S., Hogg, Andy, Yeager, Steve G., Tsujino, Hiroyuki, Nakano, H., Kiss, Andrew, Danabasoglu, Gokhan
• Format: Article in Journal/Newspaper
• Language: English
• Published: Elsevier
• Subjects: JRA55-do; Repeat year forcing; CORE-NYF; Sea ice
• Online Access: http://hdl.handle.net/1885/219040; https://doi.org/10.1016/j.ocemod.2019.101557; https://openresearch-repository.anu.edu.au/bitstream/1885/219040/3/01_Stewart_JRA55-based_repeat_year_2020.pdf.jpg

JRA55-do, the new atmospheric dataset for driving ocean-sea-ice models based on the Japanese 55-year atmospheric reanalysis (JRA-55), has been recently proffered for use in future ocean-sea-ice hindcast simulations, including those under the Ocean Model Intercomparison Project (OMIP) umbrella, thereby replacing the Coordinated Ocean-ice Reference Experiments (CORE) interannual forcing dataset. The JRA55-do dataset contains numerous and substantial improvements over the existing CORE dataset, including refined resolution, self-consistency of forcing fields, and duration. However, one feature of CORE, that is not available in JRA55- do, is the "Normal Year Forcing" (CORE-NYF), a single repeating annual cycle of all forcing fields necessary to run an ocean-sea-ice model without imposed interannual variability. Here, we propose a process for obtaining and evaluating "Repeat Year Forcing" (RYF) datasets based on the JRA55-do dataset for driving ocean-sea-ice models. This process involves the identification of 12-month periods (not necessarily a single calendar year) that are most neutral in terms of major climate modes of variability. Three candidate periods are identified and evaluated with three global ocean-sea-ice models. We find that the largest differences between the simulations arise from model biases, indicating the ultimate choice of the candidate repeat year is not critical. By referencing to the respective CORE-NYF simulations (in an attempt to account for model bias), we find the differences between the three RYF periods to be generally consistent across the models, except for an anthropogenic warming signal for later candidate years. Based on the analysis presented here, we find all three candidate periods to be suitable for use as RYF datasets, subject to application, and we recommend the period from 1st May 1990 to 30th April 1991 to be the best available RYF dataset for driving ocean-sea-ice models. COSIMA is supported by an Australian Research Council Linkage Project (LP160100073). KDS was ...