A covariant feature of optimal parameters for Arctic sea ice model


Contact
Hiroshi.Sumata [ at ] awi.de

Abstract

An automatic parameter optimization system for a coupled ocean-sea ice model is applied to investigate uniqueness of parameter set obtained from data assimilation. We set up a parameter optimization experiment, in which 15 model parameters are optimized simultaneously using a 23-years optimization window. A series of 11 independent experiments are conducted to examine spread of objective functions, optimized sea ice fields and associated optimal parameters. The result shows sufficiently small spreads of objective functions and ice fields, whereas a significantly large spread of optimized parameters. This indicates the system gives an unique solution regarding the simulated ice fields, whereas multiple solutions regarding the associated model parameters. A correlation analysis shows the optimal parameters are inter-related and covariant. A principal component analysis (PCA) reveals that the first 3 principal components explain 70% of the variation of the optimal parameter sets, indicating a contraction of the model parameter space.



Item Type
Conference (Poster)
Authors
Divisions
Primary Division
Programs
Primary Topic
Research Networks
Peer revision
Not peer-reviewed
Publication Status
Published
Event Details
REKLIM workshop 2018, 15 May 2018 - 17 May 2018, Boppard, Germanay.
Eprint ID
48635
Cite as
Sumata, H. , Kauker, F. , Karcher, M. and Gerdes, R. (2018): A covariant feature of optimal parameters for Arctic sea ice model , REKLIM workshop 2018, Boppard, Germanay, 15 May 2018 - 17 May 2018 .


Share

Research Platforms
N/A

Campaigns


Actions
Edit Item Edit Item