Deliverable No. 4.5 Interim evaluation of initialization experiments investigating the impact of novel observations in the coupled atmosphere–land–ocean–sea ice system
Initial-value information is a key source of polar predictability at sub-seasonal to seasonal time scales. However, how to constrain numerical models with actual observations at initial prediction time is a technical and scientific challenge. The observational network is sparse and not all physicall...
Main Author: | |
---|---|
Format: | Report |
Language: | English |
Published: |
2018
|
Subjects: | |
Online Access: | https://zenodo.org/record/3569282 https://doi.org/10.5281/zenodo.3569282 |
Summary: | Initial-value information is a key source of polar predictability at sub-seasonal to seasonal time scales. However, how to constrain numerical models with actual observations at initial prediction time is a technical and scientific challenge. The observational network is sparse and not all physically relevant variables for initialization are well monitored. In addition, numerical models used in predictions have their own biases, which introduces numerical instabilities and shocks after initialization. One objective of APPLICATE WP4 is to deliver recommendations for optimal initialization strategies of weather and climate models. In this “interim” deliverable, progress is presented on the initialization of novel products and on the use of novel assimilation techniques. It is found that sea ice initialization can benefit the skill in the North Atlantic Oscillation (NAO) prediction, but that sea ice has to be initialized in a consistent way with the ocean underneath to avoid initial shocks. In retrospective seasonal prediction experiments, initialization of sea ice thickness gives promising results for forecasting the summer sea ice edge position, a diagnostic of interest for a wide range of stakeholders. Finally, a new scheme has been developed for assimilation of continental snow in a multi-layer model, opening prospects for enhanced land-surface- induced predictability at sub-seasonal time scales. The work summarized in this deliverable is still ongoing, will be continued until the end of the project, and will result in further recommendations for optimal initialization techniques to be applied in WP5. |
---|