Abstract
To study the effectiveness of methods to reduce errors for Arctic Sea ice initialization due to underestimation of background error covariance, an advanced ensemble analysis system has been developed. The system integrates the local ensemble transform Kalman filter (LETKF) with the community ice code (CICE). With a mixed layer ocean model used to compute the sea surface temperature (SST), the experiments on assimilation of observations of sea ice concentration (SIC) have been carried out. Assimilation experiments were performed over a 3-month period from January to March in 1997. The model was sequentially constrained with daily observation data. The effects of observation density, amplification factor for analysis error covariance, and relaxation of disturbance and spread on the results of SIC initialization were studied. It is shown that doubling the density of observation of SIC does not bring significant further improvement on the analysis result; when the ensemble size is doubled, most severe SIC biases in the Labrador, Greenland, Norwegian, and Barents seas are reduced; amplifying the analysis error covariance, relaxing disturbance, and relaxing spread all contribute to improving the reproduction of SIC with amplifying covariance with the largest magnitude.
Highlights
Initialization of slow surface processes such as sea ice variation may provide an important contribution to improve climate prediction [1,2]
Accurate initial conditions of sea ice state are vital for numerical sea ice prediction, which is important for human activities such as maritime shipping and oil exploration
It is shown that ice extent during winter in the Arctic
Summary
Initialization of slow surface processes such as sea ice variation may provide an important contribution to improve climate prediction [1,2]. Advanced data assimilation (DA) methods combine real observations with model forecasts to improve initial conditions for sea ice prediction [4]. The “relaxation-to-prior” method, which relaxes the analysis ensemble variance to the forecast ensemble value by mixing background and analysis ensemble perturbations [22] was introduced Other measures, such as adding stochastic physics [23,24] and adopting multi-model and/or multi-physical parametrizations [25], were used. There have been a few EnKF-based data assimilation systems for sea ice modeling (for example, [4,8,10]) and effects of reducing underestimation of error covariance due to uncertainty of background have been explored. An ensemble analysis system for Arctic Sea ice has been developed and systematic investigations on methods reducing error for sea ice initialization due to underestimation of error covariance was performed
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
