The summer season of 1997–1998 marked an unprecedented extent of open water in the southern Weddell Sea, not observed since satellite observations began in the early 1970s. Wind patterns during the ...summer period tended to be more southerly than in years with typical summer ice coverage, leading to the hypothesis that the polynya opened due to ice advection away from the Ronne Ice Shelf and subsequent enhanced melting through the open water‐albedo feedback mechanism. A numerical model is run using a bulk forcing data set from the 1980s for model spin‐up and the control run. Initiation of the polynya by the anomalous wind pattern is investigated numerically by substituting winds from the 1997–1998 period for the control winds. While a narrow polynya is present in the control run, it opens considerably with the 1997–1998 winds, to the extent observed in the satellite images. If solar radiation flux is not allowed to heat the mixed layer, the sea ice in front of the Ronne Ice Shelf shows some opening, but the polynya does not form to the extent observed, demonstrating the important role played by open water‐albedo feedback. The mixed layer heat budget analysis shows the predominant balance in these three runs was derived from surface forcing and that the upwelling of heat from the deep ocean was not a significant driving force for the 1997–1998 Ronne Polynya. Further analysis of the fresh water budget indicates that the mixed layer salinity increased in areas of open water during the 1997–1998 season by as much as 0.5 psu.
We implement a variance‐based distance metric (Dn) to objectively assess skill of sea ice models when multiple output variables or uncertainties in both model predictions and observations need to be ...considered. The metric compares observations and model data pairs on common spatial and temporal grids improving upon highly aggregated metrics (e.g., total sea ice extent or volume) by capturing the spatial character of model skill. The Dn metric is a gamma‐distributed statistic that is more general than the χ2 statistic commonly used to assess model fit, which requires the assumption that the model is unbiased and can only incorporate observational error in the analysis. The Dn statistic does not assume that the model is unbiased, and allows the incorporation of multiple observational data sets for the same variable and simultaneously for different variables, along with different types of variances that can characterize uncertainties in both observations and the model. This approach represents a step to establish a systematic framework for probabilistic validation of sea ice models. The methodology is also useful for model tuning by using the Dn metric as a cost function and incorporating model parametric uncertainty as part of a scheme to optimize model functionality. We apply this approach to evaluate different configurations of the standalone Los Alamos sea ice model (CICE) encompassing the parametric uncertainty in the model, and to find new sets of model configurations that produce better agreement than previous configurations between model and observational estimates of sea ice concentration and thickness.
Plain Language Summary
We proposed a statistically robust distance metric to objectively quantify the skill of sea ice models in reproducing observations. The scheme is more general than other statistics (Chi2) in that it can incorporate both model and observational uncertainties in the analysis and does not require assumptions that are hard to hold for computer models, for instance, that the models are unbiased and mismatches with observations only arise from observational errors. We applied this scheme to evaluate the Los Alamos Sea Ice model with respect to its uncertainty in 40 parameters, and were able to rank 398 different model configurations according to their skill in simultaneously simulate the arge scale distributions of sea ice concentration and thickness during 2003‐2009.
Key Points
A framework for probabilistic validation of sea ice models which accounts for observational and model uncertainties is developed
The probabilistic yardstick uses the gamma distribution which is more general than the most commonly used Chi2 goodness‐of‐fit test
The skill of different CICE configurations in simulating sea ice concentration and thickness is determined over parameter uncertainty
We use a global coupled ocean-sea ice model to test the hypothesis that the disintegration of the West Antarctic ice sheet (WAIS), or just its ice shelves, may modify ocean circulation and sea-ice ...conditions in the Southern Ocean. We compare the results of three model runs: (1) a control run with a standard (modern) configuration of landmask in West Antarctica, (2) a no-shelves run with West Antarctic ice shelves removed and (3) a no-WAIS run. In the latter two runs, up to a few million square kilometres of new sea surface area opens to sea-ice formation, causing the volume and extent of Antarctic sea-ice cover to increase compared with the control run. In general, near-surface waters are cooler around Antarctica in the no-shelves and no-WAIS model runs than in the control run, while warm intermediate and deep waters penetrate further south, increasing poleward heat transport. Varying regional responses to the imposed changes in landmask configuration are determined by the fact that Antarctic polynyas and fast ice develop in different parts of the model domain in each run. Model results suggest that changes in the extent of WAIS may modify oceanographic conditions in the Southern Ocean.
Automatic differentiation (AD) is used to perform a multiple parameter sensitivity analysis for the Los Alamos sea-ice model CICE. Numerical experiments are run by six-hourly, 1987 forcing data with ...a two-hour time step, and the AD-based sensitivity scheme is validated by comparison with derivatives calculated using the conventional finite-difference approach. Twenty-two thermodynamic and dynamic parameters are selected for simultaneous analysis. Of these, the simulated average sea-ice thickness is most sensitive to ice density; albedos and emissivity predominate in summer, while ice thickness is highly sensitive to the snow density in winter. Ice conductivity, the ice–ocean drag parameter, maximum ice salinity and ridging parameters significantly affect the simulation year-round. Gradient information computed by the AD-based sea-ice code is then used in an experiment designed to assess the efficacy of this technique for tuning the parameters against observational data. Preliminary results, obtained with a bound-constrained minimization method and with simulated observational data, show that satisfactory convergence is obtained.
A high‐resolution coupled ice‐ocean model, forced with 1983–1997 European Center for Medium‐Range Weather Forecasts data, is used to explore recent Arctic change. In response to changes in ...atmospheric circulation, stronger cyclonic circulation is present in Arctic sea ice and upper ocean in the late 1980s and early 1990s as compared to the early 1980s, manifested as the weakening of the Beaufort Gyre and the shifting of the Transpolar Drift Stream. Corroborating previous studies, ice divergence in the central Arctic Ocean is highly correlated with surface atmospheric vorticity in summer, suggesting that summer atmospheric circulation is more important than winter for inducing interannual variability of the central Arctic ice divergence and growth rate. The weakening of the summer atmospheric cyclonic circulation from the earlier period to the later period over the Canadian Basin leads to decreased ice divergence there, which then has significant impact on the ice growth rate by reducing ice formation in fall and winter. For the 15 year period, variability in the spatial distribution of ice concentration and thickness is largely determined by the ice dynamics, which is dominated by the atmospheric circulation, except over the Greenland and Labrador Seas, where the ice thermodynamics plays a more important role. The model simulation supports the recent observations of increased presence of Atlantic Water in the Arctic Ocean. The spatial pattern of warming and salinization of the Arctic Atlantic layer follows the pathways of the strengthened boundary currents along the continental slopes and over the ridges, thereby slowly spreading more Atlantic Water downstream from the eastern Arctic into the western Arctic. The integrations with and without surface temperature restoring indicate that the restoring leads to a warmer ocean surface temperature. However, the restoring has little impact on its interannual variability for the 15 year period.
Results from adding a tracer for age of sea ice to a sophisticated sea ice model that is widely used for climate studies are presented. The consistent simulation of ice age, dynamics, and ...thermodynamics in the model shows explicitly that the loss of Arctic perennial ice has accelerated in the past three decades, as has been seen in satellite-derived observations. Our model shows that the September ice age average across the Northern Hemisphere varies from about 5 to 8 years, and the ice is much younger (about 2--3 years) in late winter because of the expansion of first-year ice. We find seasonal ice on average comprises about 5% of the total ice area in September, but as much as 1.34 x 10{sup 6} km{sup 2} survives in some years. Our simulated ice age in the late 1980s and early 1990s declined markedly in agreement with other studies. After this period of decline, the ice age began to recover, but in the final years of the simulation very little young ice remains after the melt season, a strong indication that the age of the pack will again decline in the future as older ice classes fail to be replenished. The Arctic ice pack has fluctuated between older and younger ice types over the past 30 years, while ice area, thickness, and volume all declined over the same period, with an apparent acceleration in the last decade.
Here, we implement a variance-based distance metric (Dn) to objectively assess skill of sea ice models when multiple output variables or uncertainties in both model predictions and observations need ...to be considered. The metric compares observations and model data pairs on common spatial and temporal grids improving upon highly aggregated metrics (e.g., total sea ice extent or volume) by capturing the spatial character of model skill. The Dn metric is a gamma-distributed statistic that is more general than the χ2 statistic commonly used to assess model fit, which requires the assumption that the model is unbiased and can only incorporate observational error in the analysis. The Dn statistic does not assume that the model is unbiased, and allows the incorporation of multiple observational data sets for the same variable and simultaneously for different variables, along with different types of variances that can characterize uncertainties in both observations and the model. This approach represents a step to establish a systematic framework for probabilistic validation of sea ice models. The methodology is also useful for model tuning by using the Dn metric as a cost function and incorporating model parametric uncertainty as part of a scheme to optimize model functionality. We apply this approach to evaluate different configurations of the standalone Los Alamos sea ice model (CICE) encompassing the parametric uncertainty in the model, and to find new sets of model configurations that produce better agreement than previous configurations between model and observational estimates of sea ice concentration and thickness.
This chapter contains sections titled:
Introduction
Theoretical Framework
Expected Thickness Spread due to Longwave Flux Errors
20C3M Surface Energy Fluxes and Flux‐Derived Thickness
Discussion and ...Conclusions