The dataset includes source codes, a regional example and a global example related to the draft paper "An Effective Communication Topology for Performance Optimization: A Case Study of Finite Volume Wave Modeling (FVWAM)". It also includes a user manual to run the FVWAM model.Compared to the previous version, we added a regional example and updates in source codes to support the GNU compiler for tests with the Open MPI library in this version.

The dataset includes source codes, a regional example and a global example related to the draft paper "An Effective Communication Topology for Performance Optimization: A Case Study of Finite Volume Wave Modeling (FVWAM)". It also includes a user manual to run the FVWAM model.Compared to the previous version, we added a regional example and updates in source codes to support the GNU compiler for tests with the Open MPI library in this version.

Arctic climatology datasetTemperature,Salinity0.25°×0.25°57levels0-1500m

Arctic Ocean Thermohaline Dataset Climatology0.25°×0.25°0-1500m57levels

The dataset includes source codes and an example related to the draft paper "An Effective Communication Topology for Performance Optimization: A Case Study of Finite Volume Wave Modeling (FVWAM)". It also includes a user manual to run the FVWAM model.

Röhrs and Kaleschke (2012) found that thin ice has a unique signature in the emissivity radio in the vertical polarized brightness temperature channels at frequencies between 89.0 GHz and 18.7 GHz in winter. The different sea ice classes especially for water and thin ice are characterized by the emissivity ratios above one. Following the proposed algorithm, we derive sea ice lead fraction from the AMSR2 brightness temperature data for the freezing season (November-April) north of 65°S. The spatial resolution of the data is 6.25 km. The steps are as follows. Firstly, the AMSR2 L1B brightness temperatures at frequencies of 18.7 GHz (TB,18.7V) and 89.0 GHz (TB,89V) are interpolated onto the NSIDC EASE grid with a spatial resolution of 6.25 km. Secondly, the brightness temperature radio r=TB,89V/TB,18.7V is calculated. Thirdly, a mean filter is used to enhance the signal of the leads. Finally, lead fraction which is defined as the area fraction of thin ice compared to other ice classes is computed. The proposed algorithm exhibits advanced ability in detecting sea ice leads in pack ice zone. It can detects leads wider than 3 km and resolves about 50% of the lead area compared to MODIS satellite images.File descriptionsPeriod and temporal resolution: November 1, 2012, to April 30, 2020;Daily for freezing season: November 1 to April 30Coverage and spatial resolution: Arctic north of 65°NSpatial resolution: 6.25 km x 6.25 km, NSIDC EASE grid.Geographic longitude: -180°E to 180°EGeographic latitude: from 65°N to 90°NDimension: 1792 rows x 1216 columnFormat: NetCDF

Modeled wintertime sea ice drift, sea ice thickness, dynamic sea surface height and sea ice drift budget terms in the Arctic between 1981 and 2020.

Modeled wintertime sea ice drift, sea ice thickness, dynamic sea surface height and sea ice drift budget terms in the Arctic between 1981 and 2020.

The regular ice observation was carried out during March to December 2012 every 2~7 days near Chinese Zhongshan Station in Prydz Bay. Snow and ice thickness, freeboard, ice type and characters were recorded at two sea ice observation positions (SIP), with one was first-year ice (SIP-FYI) and another was multi-year ice (SIP-MYI). The summary of gap layer records observed at two sea ice observation position SIP-MYI and SIP-FYI near Zhongshan Station in 2012 was presented here. NA means no available data. MYI and FYI represents multi-year ice and first-year ice, respectively.

We applied a 1‐D high‐resolution Thermodynamic Sea Ice and snow model HIGHTSI (Launiainen & Cheng, 1998). To simulate the evolution of snow and ice temperature profiles and mass balance, HIGHTSI solves nonlinear partial differential heat conduction equations, as well as melting and freezing processes. Solar radiation absorbed in the ice highly depends on the bulk extinction coefficient κ and the depth of ice surface scattering layer (SSL), i.e., a layer where a major part of the incoming solar radiation is either scattered or absorbed. The depth of SSL typically increases towards the end of the melt season, finally reaching some 0.1 m (Light et al., 2015). The solar radiation attenuates more rapidly in SSL than in the internal layer below. The parameterization of penetration of solar radiation within snow and ice allows HIGHTSI to quantitatively simulate sub-surface melting of snow and ice. HIGHTSI has been validated extensively and applied widely in both process studies and operational services (Cheng et al., 2008, 2013; Wang et al., 2015; Merkouriadi et al., 2017, 2019; Mäkynen et al., 2020; Zhao et al., 2020). Detailed model parameterizations are given in Supporting Information Table S2. We made a control model run on an MYI floe covering the entire melting season from late spring (1 November 2011) until autumn (31 March 2012). The initial snow depth and ice thickness were 0.17 m and 1.5 m, respectively, based on in situ observations. In early November, in-ice temperature revealed a linear profile (Lei et al., 2010) and therefore used as an initial condition for the control run. The meteorological parameters observed by an automatic weather station (AWS) at the Chinese Zhongshan Station were used as model forcing. The wind speed (Va), air temperature (Ta), and relative humidity (Rh) were observed at 10 m height with one-minute time interval. The total cloud fraction (CN) was observed visually four times daily. Total precipitation (Prec) was observed at the Russian Progress Station, 1 km southeast of Zhongshan Station. On the basis of previous studies in the Prydz Bay region, the oceanic heat flux (Fw) has an annual cycle with a maximum value in March and a minimum in September (Heil, 1996; Lei et al., 2010). Unfortunately, no summer observations are available due to unsafe ice conditions. We therefore assumed a simple linear increase of monthly mean oceanic heat flux from an observed 16 W/m2 in November to an observed 32 W/m2 in March (Zhao et al., 2019a). Figure S1 shows the time series of weather data and oceanic heat flux used for the control run. Snow depth and ice thickness were manually observed by members of the Russian Progress Station on a weekly basis between April and December 2011. The measurements were made by an ice gauge and a ruler. The accuracy of the measurements was 0.01 m. The measurements are compared with modeled values. In addition to the control run, several model sensitivity experiments were made. To study a regional pattern of gap layers, we run Fast Ice Prediction System (FIPS) for the domain of land-fast ice in Prydz Bay (Zhao et al., 2020). The AMSR2 ice concentration data were used to identify the aerial coverage of ice in the coastal region (Spreen et al., 2008). The ERA-Interim reanalysis products (ERA-I) from the European Centre for Medium-Range Weather Forecasts (ECMWF) were used as weather forcing for HIGHTSI. When comparing four atmospheric reanalysis over the Antarctic sea ice, Jonassen et al. (2019) found that ERA-Interim had generally the best skill scores. The spatial and temporal resolutions of ERA-Interim are 0.125° and 6-hours, respectively. The AMSR2 data showed that the entire FIPS domain was covered by ice with a typical concentration of 80~100% . On the basis of climatological results of FIPS, the initial snow depth and ice thickness on 1 November varied spatially from 0.3 to 0.5 m and from 1.0 to 1.5 m, respectively.