Arctic ocean–sea ice reanalysis for the period 2007–2016 using the adjoint method

et al. 2022

Preprint

Abstract. The adjoint technique has been applied to the coupled ocean and sea ice models for sensitivity studies and Arctic state estimation. However, the accuracy of the adjoint model is degraded by simplifications on the adjoint of the sea ice model, especially adjoint sea ice rheology. As part of ongoing developments of coupled ocean and sea ice estimation system, we incorporate and stabilize the adjoint of viscous-plastic sea ice dynamics (adjoint-VP) and compare it with the adjoint of a free drift sea ice model (adjoint-FD) through assimilation experiments. Using the adjoint-VP resulted in a further cost reduction of 7.9 % in comparison to adjoint-FD with noticeable improvements in ocean temperature over the open water and intermediate layers of the Arctic Ocean. Adjoint-VP more efficiently adjusts uncertain parameters than adjoint-FD by involving different sea ice retreat processes. For instance, adjoint-FD melts sea ice up to 1.0 m in the marginal seas from May to June through over-adjusting air temperature (>8 °C); adjoint-VP reproduces the sea ice retreat with smaller adjustments on the atmospheric state within the prior uncertainty range. The developments of the adjoint model here lay the foundation for further improving Arctic ocean and sea ice estimation through comprehensively adjusting the initial conditions, atmosphere forcings, and model parameters.

Section: The Coupled Ocean-sea Ice Model and Assimilation Systemmentioning

confidence: 99%

Section: Viscos-plastic Sea Ice Dynamics and Its Adjointmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Effects of including the adjoint sea ice rheology on estimating Arctic ocean-sea ice state

et al. 2022

Preprint

“…Integrated Arctic Observation System Project, INTAROS) by extending and unifying different Arctic observing systems from different nations. To help the process of the Arctic observing system development, we performed observing system simulation experiments (Errico et al ., 2013) using a pan‐Arctic ocean–sea ice model and its adjoint (Koldunov et al ., 2017; Lyu et al ., 2021) as part of the European Union INTAROS effort. The goal was to test the degree by which the existing Arctic Ocean observing system can help constraining model simulations, including SIC, SIT, sea ice volume (SIV), ocean temperature, freshwater changes, and exchanges with the Atlantic Ocean, and to identify how much better the solution can become by assimilating additional mooring and high‐density hydrographic profile observations.…”

Section: Introductionmentioning

confidence: 99%

Assessing the current and future Arctic Ocean observing system with observing system simulating experiments

Quart J Royal Meteoro Soc

Serra

et al. 2021

Self Cite

Arctic Ocean Observing System Simulation Experiments (OSSEs) were performed with a pan-Arctic coupled ocean-sea ice data assimilation system to assess the impacts of assimilating available observations on the Arctic ocean-sea ice state. To this end, the adjoint method with a 3-year assimilation window was used to assimilate synthetic observations sampled from a 4 km model simulation at the spatio-temporal distribution of the existing observing system.After data assimilation, the sea ice state, including sea ice concentration (SIC), sea ice thickness (SIT), and sea ice volume (SIV), were significantly improved, benefiting from the high spatio-temporal coverage of SIC observations and the wintertime SIT observations. In contrast, the ocean state is not very well constrained with the existing hydrographic observing system. An additional 1 • × 1 • ocean profiling array with a 10-day sampling frequency was seen to substantially improve the estimated ocean temperature and freshwater content. Data from additional moorings deployed in the Fram Strait and continental slope of the Laptev Sea could also improve the pathway of Atlantic inflow to the Arctic Ocean and the temperature of the Atlantic inflow but degrade the mean volume transport through the Fram Strait. Moreover, estimated exchanges between the Arctic Ocean and the Atlantic Ocean through the Fram Strait, Davis Strait and the St. Anna Trough were found to benefit from the high-density profiling array.

Effects of including the adjoint sea ice rheology on estimating Arctic Ocean–sea ice state

et al. 2023

Ocean Sci.

Abstract. The adjoint assimilation method has been applied to coupled ocean and sea ice models for sensitivity studies and Arctic state estimations. However, the accuracy of the adjoint model is degraded by simplifications of the adjoint of the sea ice model, especially the adjoint sea ice rheologies. As part of ongoing developments in coupled ocean and sea ice estimation systems, we incorporate and approximate the adjoint of viscous-plastic sea ice dynamics (adjoint-VP) and compare it with the adjoint of free-drift sea ice dynamics (adjoint-FD) through assimilation experiments. Using the adjoint-VP results in a further cost reduction of 7.9 % in comparison to adjoint-FD, with noticeable improvements in the ocean temperature over the open water and the intermediate layers of the Arctic Ocean. Adjoint-VP adjusts the model input more efficiently than adjoint-FD does by involving different sea ice retreat processes. For instance, adjoint-FD melts the sea ice up to 1.0 m in the marginal seas from May to June by overadjusting air temperature (>8 ∘C); adjoint-VP reproduces the sea ice retreat with smaller adjustments to the atmospheric state within their prior uncertainty range. These developments of the adjoint model here lay the foundation for further improving Arctic Ocean and sea ice estimations by comprehensively adjusting the initial conditions, atmospheric forcings, and parameters of the model.