Interactive comment on “ Wave climate in the Arctic 1992 – 2014 : seasonality and trends ”

Abstract. Over the past decade, the diminishing Arctic sea ice has impacted the wave field, which depends on the ice-free ocean and wind. This study characterizes the wave climate in the Arctic spanning 1992–2014 from a merged altimeter data set and a wave hindcast that uses CFSR winds and ice concentrations from satellites as input. The model performs well, verified by the altimeters, and is relatively consistent for climate studies. The wave seasonality and extremes are linked to the ice coverage, wind strength, and wind direction, creating distinct features in the wind seas and swells. The altimeters and model show that the reduction of sea ice coverage causes increasing wave heights instead of the wind. However, trends are convoluted by interannual climate oscillations like the North Atlantic Oscillation (NAO) and Pacific Decadal Oscillation. In the Nordic Greenland Sea the NAO influences the decreasing wind speeds and wave heights. Swells are becoming more prevalent and wind-sea steepness is declining. The satellite data show the sea ice minimum occurs later in fall when the wind speeds increase. This creates more favorable conditions for wave development. Therefore we expect the ice freeze-up in fall to be the most critical season in the Arctic and small changes in ice cover, wind speeds, and wave heights can have large impacts to the evolution of the sea ice throughout the year. It is inconclusive how important wave–ice processes are within the climate system, but selected events suggest the importance of waves within the marginal ice zone.

per year that are ice-free (i.e. concentration <15%). Most of the ice covered areas are statistically significant and are ice-free 28 2 additional days each year. The strongest trends are located in the Barents and Kara Seas with 8 more ice-free days per year. 29 Only isolated regions near Svalbard, Greenland, and the Amundsen Gulf have increasing ice coverage. 30 Most of the basin has increasing wave heights shown by the altimeters and wave model in the top right and bottom panels. 31 The bottom panels show the co-located H s trends from the altimeters and the model agree, despite the stronger trends in 32 the altimeters. However, the altimeter confidence interval encompasses the model results so statistically they are equivalent.   Figure 8 shows the trends from other parameters using monthly averages. In this case the results are presented as percentages 11 relative to the mean to allow comparison. The trends in U10 are calculated using the entire dataset independent of ice cover; 12 otherwise all other variables are computed from ice-free statistics. The decreasing U10 trend in the Nordic Sea is significant 13 and is consistent with the H s trend. Across most of the sea ice, U10 is decreasing especially in the Beaufort Sea. Some regions 14 have weak increasing trends of 0.25% per year. Wind speeds in the Baffin Bay are increasing creating taller wave heights.

15
The trends in T m02 follow the same pattern as the wave heights in Figure 7 and with an increase of 2% (2-3 cs) per year. 16 The trends in H sw and H ss heights have similar spatial patterns as H s . However, H ss is increasing at a faster rate compared 17 to H sw in the Beaufort, East Siberia, Laptev, and Kara Seas. This is directly related to the higher occurrence of swells (i.e.  Table 1 presents corre-28 lation coefficients between area-averaged monthly time series of sea ice, U10, and H s and the NAO and PDO indices. The  Figure 9 graphically summarizes the regional trends through Sen's slope of the NAO, PDO, ice-free area, U10, and H s .

34
The NAO and PDO have statistically significant decreasing trends. The NAO is expected to cause the decreasing trend in the  So far, we have seen that the decreasing the sea ice has drastic impacts on the wave climate and has amplified sea states.

11
The waves also impact sea ice and their influence is unknown due to lack of observations and understanding of the wave-ice year average (dashed line) shows the ice cover climatology increases 6% from November through December . The first event,

19
November 23-29, indicates a decrease in ice cover by -3% corresponding to a loss of 60,000 km 2 . This event coincides with 20 H s , peak periods, and wind speeds exceeding 5 m, 12 s, and 14 m/s respectively. The second event in December has larger 21 wave heights (>6 m), however the ice cover remains the same. 22 We qualitatively compare and contrast these two events by considering the physical processes of the wind and wave condi-23 tions. Figure 11 shows snapshots of the peak wind speed, wave height, and period for the two selected events. The white and 24 black lines denote the ice edge defined by 15% ice concentration before and after the event. During the November event, the   The impact of waves on the sea ice is difficult to determine without detailed knowledge of wave-ice interaction. The evolution  Table A1 displays error H s metric at the buoys. Standard error metrics including the normalized bias (NBIAS), root mean 9 square error (RMSE), correlation coefficient (R), scatter index (SI), and normalized standard deviation (NSTD) are given below 10 assuming the x represents the observation and y represents the model, and n is the number of data pairs: All buoys have water depths less than 50 m. CFSR overestimates the H s by at least 5% at all locations while ERAI under-21 estimates by 3% (except WMO48213). The RMSEs are commonly 0.25 m with ERAI always having a slightly better match.

22
The scatter indices and correlation coefficients for ERAI and CFSR follow the same pattern at each buoy. The NSTD shows 23 CFSR has more variability than the observations while ERAI is a smoother model. In general both models are comparable with 24 a positive bias in CFSR and negative bias in ERAI. grow with the wave amplitude, leading to a dependence of β on the wave amplitude. 28 We thus revisit this question and propose a parametrization for the laminar to turbulent transition of the boundary layer 29 below the ice. In turbulent conditions, an important parameter is the roughness length below the ice z 0 . That roughness is dissipation by friction at the air-sea interface, where the significant orbital amplitudes of the surface velocity is, for deep water waves, and f e is the same dissipation factor used for bottom friction, that is a function only of the ratio a orb /z 0 where a orb is the 9 significant orbital displacement at the sea surface, here for deep water waves a orb = H s /2.