Data Description

Parameters

The main parameters for this data set are daily and weekly sea ice motion (cm/s). Sea ice motion is divided into along-x (u) and along-y (v) components. Error variance (for daily data) and the number of contributing files (for weekly data) are also provided.

Additionally, the input data used to calculate daily and weekly sea ice motion vectors are included in this data set. The input data consist of sea ice motion vectors derived from specific sensors (see Table 2 for more details).

File Information

Format

Daily and weekly sea ice motion are provided in georeferenced netCDF (.nc) format. Input data are also provided in netCDF (.nc) format.

PNG (.png) browse images are also included.

Daily and Weekly Sea Ice Motion File Contents

Daily and weekly sea ice motion netCDF file contents are described in Table 1 and represented in Figure 1.

Browse Image File Contents

One browse image displaying weekly sea ice motion is provided for every week of data. Figure 2 contains a sample browse image.

Input Data File Contents

Input data files are described in Table 2. All input data files contain the same variables and structure, except for the IABP buoy data. Since the buoy data are point-source data, they are stored as a list, not as a georeferenced grid. Input data fields are described in more detail in Tables 3 and 4.

Directory Structure

Data are available for download via HTTPS; the link is accessible through the "Download Data" tab. 

Within the file directory, data are sorted by hemisphere. There are two main subdirectories: north and south. Within these subdirectories, folders are subdivided by data type. Table 5 provides a listing of the subfolders.

Daily and Weekly Sea Ice Motion File Naming Convention

The daily and weekly data files are named according to the following convention and as described in Table 6:
icemotion_<daily|weekly>_hh_rrrr_<start-date>_<end-date>_v##.nc

Example:
icemotion_daily_sh_25km_19781101_19781231_v4.1.nc
icemotion_weekly_sh_25km_19790101_19791231_v4.1.nc

Browse Image File Naming Convention

The browse images are named according to the following convention and as described in Table 7:
icemotion_weekly_hh_rrrr_<start-date>_<end-date>_v##.png

Example:
icemotion_weekly_nh_25km_20000101_20000108_v4.1.png

Input Data File Naming Convention

The input data files are named according to the following convention and as described in Table 8:
im_from_<source>_hh_rrrr_<start-date>_<end-date>_v##.nc

Example:
​im_from_amsre_nh_37.5km_20020619_20021231_v4.1.nc
im_from_avhrr_nh_50km_19810724_19811231_v4.1.nc
im_from_buoy_nh_list_198790101_19791231_v4.1.nc
im_from_smmr_nh_75km_19781025_19781231_v4.1.nc
im_from_smmi_nh_75km_19870821_19871231_v4.1.nc
im_from_smmis_nh_75km_20070101_20071231_v4.1.nc
im_from_wind_nh_50km_19781025_19781231_v4.1.nc

File Size

Daily sea ice motion files range from approximately 18.0 - 22.0 MB.
Weekly sea ice motion files range from approximately 3.0 - 4.0 MB.
Input data files range from approximately 0.2 - 8.0 MB.

Browse images are approximately 0.5 - 1.5 MB.

Spatial Information

Coverage

Daily and weekly sea ice motion data are confined to the poles in the Northern and Southern Hemisphere, as represented in Figure 3.

Resolution

The final daily and weekly sea ice motion data are provided at a resolution of 25 km. Input data have different spatial resolutions (see Table 2). 

Geolocation

Daily and weekly sea ice motion data are georeferenced to the Northern and Southern 25 km original EASE-Grid projections. More details on these projections and grids are provided in the tables below. More details on EASE-Grid can be found on the EASE Grids website.

Temporal Information

Coverage and Resolution

The temporal coverage and resolution vary by data type and sensor. See Table 13 for more details.

Data Acquisition and Processing

Background

Sea ice movement is measured using imagery acquired by frequent, repeat coverage by remote sensing instruments. Ice motion computed from satellite imagery represents the displacement between two images with the same spatial coverage taken at different acquisition times. Researchers identify a feature, such as an ice floe, on two registered images and measure its pixel displacement. Ice velocity vectors are computed based on the pixel resolution and the time span between images.

A more automated method is to measure the correlation of groups of pixels between image pairs. A small target area in one image is correlated with several areas of the same size in a search region of the second image. The displacement of the ice is then defined by the location in the second image where the correlation coefficient is the highest. This is the spatial correlation method used to produce sea ice motion vectors for this data set.

The approach used for this data set is generally valid over short distances away from the ice edge in areas where ice conditions are relatively stable from day to day. Spatial correlation methods cannot find matches between images where a complete knowledge of ice dynamics is needed, e.g. in areas where ice is deforming or in the ice margins near the open ocean where the spatial or spectral characteristics of the ice within a pixel are changing rapidly (Emery, Fowler, and Maslanik 1995).

Acquisition

AVHRR Input Data

This input data derived sea ice motion vectors from AVHRR channel 2 (visible band) and channel 4 (infrared) Global Area Coverage (GAC) images. These images have a 50 km gridded resolution, are available for nearly two decades, provide an intermediate spatial resolution between passive microwave and buoys, and have a finer time sampling than microwave data.

Beginning with Version 3 of this data set, misregistered AVHRR estimates were removed.

For more information on AVHRR input data, please refer to: AVHRR Polar Pathfinder Twice-Daily 5 km EASE-Grid Composites.

Buoy Input Data

This input data derived sea ice motion vectors from International Arctic Buoy Program (IABP) C buoy position data. IABP provides buoy location information through satellite tracking of buoys placed on sea ice. Several buoy locations are determined each day and corresponding sea ice motions are calculated. Sea ice motion estimates from buoys are very accurate, but they are limited since the numbers and locations of buoys are driven by cost and logistics. In addition, buoys have not been placed on ice in the Eastern Arctic.

IABP buoy locations are generally provided every 12 hours: at noon and at midnight Greenwich Mean Time (GMT). This sea ice motion product uses 24-hour motion estimates from the IABP. For example, the IABP motion estimate for a buoy at noon on 01 January 2010 is derived by taking the difference of the buoy's location at noon on 02 January 2010 and its location at noon on 01 January 2010 and then dividing by 24 hours. The intervening midnight location value is not factored into the noon-to-noon 24-hour motion estimate. Similarly, the IABP motion estimate for midnight is calculated the same way, ignoring the intervening noon location information. Therefore, each buoy generally has two independent, 24-hour motion estimates: one for midnight and one for noon.  Beginning with Version 4 of this data set, the noon-time and midnight buoy vectors were averaged together to provide one buoy-derived sea ice motion per day. 

For more information the buoy input data, please refer to Brown and Kerut (1978) and the IABP website.

NCEP/NCAR Wind Input Data

This input data derived sea ice motion vectors from the NCEP/NCAR Reanalysis wind data set. The data, called U-wind and V-wind at 10 m, are available from the NOAA Earth System Research Laboratory (ESRL) Physical Sciences Division (PSD).

For more information on the NCEP/NCAR input data, please refer to Kistler et al. (2000).

Passive Microwave Input Data

The passive microwave input data come from four different instruments: SMMR, SSM/I, SSMIS, and AMSR-E. All the source data are available for download from the National Snow and Ice Data Center Distributed Active Archive Center (NSIDC DAAC). The SMMR input data are derived from the NIMBUS-7 SMMR Pathfinder Brightness Temperatures data set. Due to satellite limitations, full Arctic coverage is only available every two days with SMMR; see the SMMR 48 Hour Temporal Resolution Consequences discussion for more information. The SSM/I and SSMIS input data both come from DMSP SSM/I-SSMIS Daily Polar Gridded Brightness Temperatures, while the AMSR-E input data come from the AMSR-E/Aqua L2A Global Swath Spatially-Resampled Brightness Temperatures. Passive microwave inputs also include the Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I-SSMIS Passive Microwave Data data set, which is used to distinguish ice-covered and ice-free ocean.

Table 14 lists the channels used by each passive microwave sensor to derive input sea ice motion vectors. In general, the passive microwave sensors provide all-sky coverage, whereas the AVHRR visible and infrared channels are limited by cloud cover. For more information on the passive microwave instruments, please refer to: 

• AMSR-E Instrument Description
• SMMR, SSM/I, and SSMIS Sensors

Processing

The following steps were used to create this sea ice motion product.

1. Compute the Sea Ice Motion Fields
   A sea ice mask, derived from Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I-SSMIS Passive Microwave Data using a threshold of 15%, is used to restrict all independent sea ice motion estimates to ice-covered ocean. Detailed information about the methods used to compute sea ice motion for each input data source can be found on the Measuring Sea Ice Motion web page.
    

2. Grid the Input Data to the 25 km EASE-Grid
   Each of the input sea ice motion estimates are mapped to the output grid (e.g. Northern Hemisphere 25 km EASE-Grid).

3. Merge the Sea Ice Motion Fields
   ​Input sea ice motion estimates are combined to produce the daily sea ice motion product. Daily sea ice motion is a source- and distance-weighted average of the 15 highest-weighted input sea ice motion vectors. Each input data set is weighted according to the expected accuracy of the source data. For example, estimates derived from nearby buoys are weighted higher than NCEP/NCAR-derived estimates. Note that where data are sparse, the input data will be widely separated; when data are dense, only the very nearest estimates are considered. If the input sea ice motion estimates vary significantly from each other, this method can result in daily sea ice motion vectors that do not vary smoothly from one grid cell to the next.

4. Calculate Daily Error Values
   The input vectors from the individual input sources (NCEP/NCAR, SSM/I, SSMIS, SMMR, AMSR-E, and AVHRR) are weighted separately based upon cross-correlations with buoy vectors. The optimal interpolation uses these weights, along with their distances from the location being estimated, to obtain the final error variance. If the closest input vector was greater than 1250 km, then a value of 1000 is added to this variable. Because interpolation was applied to a surface map from passive microwave data, coastlines may contain false ice. In this case, the third variable was assigned a negative value to allow users to remove these vectors near coastlines (within 25 km). For example, a value of -1035 indicates all of the following conditions: the vector was near a coastline, the nearest sampled vector was further than 1250 km, and the vector had a standard deviation (σ) of 3.5 and an estimated error variance (σ²) of 12.25.

5. Compute Weekly Fields
   Weekly sea ice motion was computed from the daily gridded sea ice motion data for both the northern and southern polar regions. Weekly sea ice motion is an average of all the daily sea ice motions calculated for that week. At least four out of seven days were needed to compute the weekly mean.
   Weekly means for each year begin on 01 January for consistency. The last day of each year (or last two days if in a leap year) were excluded. In other words, the first week is always 01 January through 07 January and the last week of a year is either 24 December through 30 December or 23 December through 29 December (if in a leap year).

6. Encode Data and Associated Metadata in netCDF Files
   Beginning with Version 4, data are provided in netCDF format.

Quality, Errors, and Limitations

Quality Assessment - Input Data

The Measuring Sea Ice Motion web page provides information on the accuracy of sea ice motion estimates calculated from each input sensor.

Beginning with Version 3 of this data set, IABP buoy sea ice motion estimates above 70 cm/s were deemed physically unrealistic; thus, velocities that exceed this threshold were excluded from this data set. In addition, beginning with Version 4 of this data set, buoys which were outside the mask of valid sea ice motion (away from the coast, where sea ice concentration was greater than 15%) were excluded.

Quality Assessment - Daily and Weekly Sea Ice Motion Product

The icemotion_error_variance (daily data) and number_of_observations (weekly data) fields both provide a means of characterizing data quality. For the daily files, the error variance can indicate a "near coastline" check. For the weekly data, the more days that contributed to calculating the sea ice motion vector in a given grid cell, the higher the data quality. 

Missing Data

Beginning with Version 3 of this data set, there are some missing data in the Southern Hemisphere because there was not enough data from SSMI and AVHRR sensors to yield sea ice motion vectors. Missing data fields are left blank.

Limitations

The passive-microwave-derived sea ice motion estimates are based on changes in brightness temperature over consecutive days. The methods used to generate these input vectors requires fairly large areas of open ocean. As a result, ice motion cannot be calculated in regions of mixed land and ocean coverage, such as the Canadian Archipelago. The absence of sea ice motion estimates in such locations does not imply the absence of ice in these locations.

Version History

Related Data Sets

AMSR-E/Aqua L2A Global Swath Spatially-Resampled Brightness Temperatures

AVHRR Polar Pathfinder Twice-Daily 5 km EASE-Grid Composites

AVHRR Polar Pathfinder Twice-Daily 25 km EASE-Grid Composites

DMSP SSM/I-SSMIS Daily Polar Gridded Brightness Temperatures

NIMBUS-7 SMMR Pathfinder Brightness Temperatures

EASE-Grid Sea Ice Age

Related Websites

SMMR and SSM/I-SSMIS | Overview

Contacts and Acknowledgments

Mark Tschudi
University of Colorado
Colorado Center for Astrodynamics Research (CCAR)
Department of Aerospace Engineering Sciences
University of Colorado, Boulder
431 UCB
Boulder, Colorado USA 80309-0431

Walter N. Meier
University of Colorado Boulder
449 UCB
Boulder, CO 80309-0449

J. Scott Stewart
University of Colorado Boulder
Boulder, CO 80309-0449

Chuck Fowler 
University of Colorado Boulder
Boulder, CO 80309

Jim Maslanik
University of Colorado Boulder
Boulder, CO 80309

References

Brown, W. P. and E. G. Kerut. 1978. Air droppable RAMS (ADRAMS) buoys. AIDJEX Bulletin, 40, 21-29.

Cavalieri, D. J., C. L. Parkinson, P. Gloersen, J. C. Comiso, and H. J. Zwally. 1999. Deriving long-term time series of sea ice cover from satellite passive-microwave multisensor data sets, J. Geophys. Res., 104(C7), 15803-15814. doi:10.1029/1999JC900081.

Cavalieri, D., C. Parkinson, P. Gloersen, and H. J. Zwally. 1996, updated yearly. Sea Ice Concentrations from Nimbus-7 SMMR and DMSP SSM/I-SSMIS Passive Microwave Data, Version 1. [indicate subset used]. Boulder, Colorado USA: NASA DAAC at the National Snow and Ice Data Center. doi:10.5067/8GQ8LZQVL0VL.

Cracknell, A. 1997. The Advanced Very High Resolution Radiometer. London: Taylor and Francis.

Emery, W., C. Fowler, and J. Maslanik. 1995. Satellite Remote Sensing of Ice Motion, in Oceanographic Applications of Remote Sensing, ed. Motoyoshi Ikeda and Frederic W. Dobson. CRC Press, Boca Raton.

Haumann, F. Alexander, Nicolas Gruber, Matthias Münnich, Ivy Frenger and Stefan Kern. 2016. Sea-ice transport driving Southern Ocean salinity and its recent trends. Nature 537:89-92. doi:10.1038/nature19101.

Isaaks, E., and R. M. Srivastava. 1989. An Introduction to Applied Geostatistics. New York: Oxford University Press.

Kalnay, E., M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L. Gandin, M. Iredell, S. Saha, G. White, J. Woollen, Y. Zhu, A. Leetmaa, R. Reynolds, M. Chelliah, W. Ebisuzaki, W. Higgins, J. Janowiak, K. Mo, C. Ropelewski, J. Wang, R. Jenne, and D. Joseph. 1996. The NCEP/NCAR 40-Year Reanalysis Project. Bulletin of the American Meteorological Society, 77, 437–471.

Kidwell, K. 1995. NOAA Polar Orbiter Data User's Guide. U.S. Dept. of Commerce, National Oceanic and Atmospheric Administration, NESDIS.

Kistler, R., William C., Suranjana S., Glenn W., John W., Eugenia K., Muthuvel C., et al. 2001. The NCEP–NCAR 50–Year Reanalysis: Monthly Means CD–ROM and Documentation. Bulletin of the American Meteorological Society 82 (2): 247–67. doi:10.1175/1520-0477(2001)082<0247:TNNYRM>2.3.CO;2.

Krasnopolsky, V. M., L. C. Breaker, and W. H. Gemmill. 1995. A Neural Network as a Nonlinear Transfer Function Model for Retrieving Surface Wind Speeds from the Special Sensor Microwave Imager. Journal of Geophysical Research, 100(C6), 11,033–11,045.

Maslanik, J., C. Fowler, J. Key, T. Scambos, T. Hutchinson, and W. Emery. 1998. AVHRR-based Polar Pathfinder Products for Modeling Applications. Annals of Glaciology 25:388-392

Rosborough, G., D. Baldwin, and W. Emery. 1994. Precise AVHRR Image Navigation. IEEE Transactions in Geosciences and Remote Sensing 32(3):644-657.

Schweiger, A., C. Fowler, J. Key, J. Maslanik, J. Francis, R. Armstrong, M. J. Brodzik, T. Scambos, T. Haran, M. Ortmeyer, S. Khalsa, D. Rothrock, and R. Weaver. 1999. P-Cube: A Multisensor Data Set for Polar Climate Research. Proceedings on the 5th Conference on Polar Meteorology and Oceanography, American Meteorological Society, Dallas, TX, 15-20 Jan., 136-141.

Thorndike, A. S., and R. Colony. 1982. Sea Ice Motion in Response to Geostrophic Winds. J. Geophys. Res. 87(C8):5845–5852. doi:10.1029/JC087iC08p05845.

Tschudi, M. A., Meier, W. N., and Stewart, J. S. 2019. An enhancement of sea ice motion and age products. The Cryosphere Discussion, in review. doi: 10.5194/tc-2019-40.

Appendix - Merged Daily Gridded Vectors

Methods

Optimal interpolation was used to create daily gridded ice motion fields from SMMR, SSM/I-SSMIS, AMSR-E, AVHRR, IABP buoy data, and NCEP/NCAR wind data. A cokriging estimation method described by Isaaks and Srivastava (1989) was employed for the interpolation. This method utilizes the cross-correlation between several variables—in this case, the vectors derived from SMMR, SSM/I-SSMIS,AMSR-E, AVHRR, IABP buoy data, and NCEP/NCAR winds—to minimize the estimation error variance. See Figure A1 for more details.

AVHRR autocorrelation

The plot shows discontinuity near 50 km distance. If discontinuity is ignored, the autocorrelation line would cross the zero distance line at about 0.8. This value, which is related to measurement error, is often referred to as the "nugget effect" in geostatistics. The shape of the curve is similar for both u and v vector components and for all of the auto-correlation and cross-correlation functions among all of the data sets. The major difference is the starting point at zero km distance. The autocorrelations for the ice motions for AMSR-E were very similar to AVHRR, and the vectors from the wind data were very close to the 37 GHz vectors. See Table A1 for more details. 

A simple exponential function was used to model the correlation (dotted line) in the above figure. This function provided the best model to approximate the vector data. Cross-correlations between the u and v components were performed over the entire Arctic and Antarctic regions, and it was found that they were nearly independent. The u and v components were thus treated independently and were interpolated separately for simplicity. Using the modeled correlation functions, vectors for each ice-covered area on the 25 km EASE-Grid werre interpolated. See Figure A2 for an example.

Sample daily-averaged vector plot

The number of vectors from the buoy, passive microwave, and AVHRR data varies. A plot of monthly average percent coverage of ice-covered areas shows that summer weather conditions affect the ability of satellites to track ice. Figure A3 shows the percentage of sea ice grid cells that contain vectors.

Figure A4 shows the average number of vectors available from each source.

Similarly, Figure A5 shows the percentage of available vectors from each source.

Note that the vectors from the 85 GHz channels began in 1987 with the launch of the SSM/I instrument.

Final Summary

Gridded vector fields are not the optimal data set for everyone. They are best used for evaluating climatological patterns and changes in sea ice motion over the last 25 years. Various assumptions were factored into the production of gridded vector fields. For optimal interpolation, the data should be stationary, homogenous, and isotropic. Sea ice does not have these characteristics. Changes in ice concentration and thickness may negate the assumptions. In an ideal case, two-dimensional correlation functions are found at every grid location on each day to help alleviate some of these problems.

Another limitation of the data is cloud cover, which can affect a satellite's ability to discern sea ice, and thus, the number and location of available vectors on each day.

Data users should use the summer vector plots with caution due to the low number of vectors resulting from surface melt and increased cloud cover in the summer. As individual summer daily plots contain significant noise, weekly- and monthly-mean plots are more reliable.

Accuracy

Several years of vectors were interpolated to the same grid for comparison, but without using the buoy data. The mean difference between the interpolated u components and the buoy vectors was 0.1 cm/sec with a Root Mean Square (RMS) error of 3.36 cm/sec. For the v component, the mean was 0.4 cm/sec with an RMS error of 3.40 cm/sec.