Description of task
This activity provides improved models, a knowledge of model sensitivities and an understanding of processes needed to guide the activities of the project and advise the wider scientific community.
Observations from core themes 1-3, the IPY and other sources (candidate data to be used by the individual models are listed in table 4.1) will be compared with different ocean-ice-atmosphere coupled models as well as ocean-ice and atmosphere standalone versions. Thereby model deficiencies can be identified, and in particular, the description and understanding of processes relevant to the disappearance of sea ice can be improved by classical model development and by development of adjoint models. Improvement of certain parameterizations (cloud and albedo formulations) is supported by local process-oriented model studies in core theme 2. Considering the recent changes that have moved the Arctic system beyond the range of our past experience, this modelling effort will ultimately result in better simulation capabilities covering a wider range of climates, i.e. better predictive skill.
In order to obtain a solid understanding of the system, it is necessary to use a variety of ocean, ice and atmosphere models in coupled and standalone versions, covering different domains, process parameterizations and numerics. The candidate models are RCAO (a regional coupled ocean-ice-atmosphere model by SMHI), NAOSIM (regional ocean model by AWI), ADNAOSIM (the adjoint version of NAOSIM by OASys and FastOpt), ORCM (met.no’s Oslo regional atmosphere-ocean-ice model with isopycnal coordinates in the ocean), BCM (global coupled model by NERSC), HELMI (multi-category sea ice model coupled to MPI-OM1 by FIMR), TOPAZ (the operational ocean model and data system of NERSC, including a local focus on the Fram Strait), HIRLAM (the atmospheric NWP model of SMHI) and a 1D vertical column model of ocean, sea ice and atmosphere by UGOT and SMHI.
These models complement each other with their different capabilities. The coupled models take regional atmosphere-ocean feedbacks into account (e.g. sea ice extent – arctic atmospheric vortex feedback). The 1D model directly supports 3D models with its capability to explore parameter spaces over wide ranges (e.g. cloud and albedo specifications that will be extensively explored using the results of core theme 2. Adjoint sensitivities can be explored with ADNAOSIM. The isopycnal coordinates of ORCM allow a differing treatment of the tracking of water masses throughout the Arctic Ocean. And the local focus on the FRAM Strait in the operational TOPAZ model allows for continuous real time validation with DAMOCLES observational data with feedback to the field operations.
Sensitivities of the arctic system will be evaluated with the above model hierarchy. This will help with the task of rating the relative importance of processes, parameters and boundary conditions, and will improve the identification of causal relationships and the understanding of processes and mechanisms. Sensitivities and associated model improvements will be addressed with respect to key parameters such as the model resolution, vertical density stratification and atmospheric boundary layer parameterization in climate models (in collaboration with NWP-related work in the atmospheric core theme 2), lateral boundary conditions of regional arctic domains, the initial states of ocean and sea ice, ocean-ice-atmosphere exchange fluxes and specific processes underlying sea ice dynamics as analyzed in core theme 1 (sea ice) with the help of a multi-category sea ice model. Other candidates for sensitivity tests will be the role of regional water transformation processes in the Barents Sea for the Arctic, the role of shelf processes such as brine formation and river runoff, the impact of regional coupled ice-ocean-atmosphere interaction and the impact of advection numerics on the establishment of Atlantic inflow advection regimes. These model experiments will be coordinated by cross checking sensitivities so that it will be possible to draw common conclusions. According to the nature of these experiments, most integrations will cover several decades of the ERA period (about 40 years). Model improvements in this task are linked to the outcome of sensitivity runs in comparison with observations.
The key candidate quantities for sensitivity runs will be the sea ice volume, the inflow and outflow transports of volume, heat, sea ice and freshwater, the vertical T/S structure in different locations of the Atlantic water inflow under the sea ice and outside the sea ice margins (with special emphasis on the upper halocline), the structure and composition of boundary currents in the Arctic ocean and Nordic Seas, the amount of brine release and salinity underneath sea ice, ocean-atmosphere and ice-atmosphere heat fluxes and arctic cloud and radiation properties. The vast amount of data created by the sensitivity runs will be analyzed and condensed using different techniques spanning the range from simple mean value comparison to statistical projections onto large-scale variability patterns such as those associated with NAO.
