Spatially and temporally continuous reconstruction of Antarctic Amundsen Sea sector ice sheet surface velocities: 1996-2018

• Main Author: Gardner, Alex
• Format: Other/Unknown Material
• Language: English
• Published: Zenodo 2023
• Subjects: ice sheet; velocity; cryosphere; Antarctic; Amundsen Sea; Antarc*; Antarctica; Ice Sheet
• Online Access: https://doi.org/10.5281/zenodo.7809354

Spatially and temporally continuous reconstruction of ice sheet surface velocities for the Amundson Sea Sector of the Antarctica. The reconstruction is derived from the synthesis of annual published InSAR (R14: Rignot et al. 2014) and optical (G18: Gardner et al., 2018 & Gardner et al., 2022) surface velocities. Data are posted on a uniform 240 m by 240 m grid in Antarctic Polar Stereographic (EPSG:3031) coordinates. The temporal posting is every 2.4 months or 1/5 of a year. R14 and G18 annual velocity data have large errors and data gaps in both space and time that make the data challenging to work with. For this reason, a Spatially and temporally continuous reconstruction was made. These are the preprocessing steps that were applied to create the reconstruction: R14 component velocities [vx/vy] are mapped to the same 240-m grid as G18 for the Amundson Sea sector. Velocities falling outside of mapped ice extents (see Paolo et al., 2022) are set to no data values. A reference velocity is defined as the 1996 velocity field or the earliest valid measurement thereafter. The average of both velocities is taken if multiple observations exist for the first year of data. For areas moving faster than 200 m/yr., the percentage anomalies are calculated for all years relative the reference velocity. This was done for both G18 and R14 velocities separately. Annual velocity anomalies are then filter with a 5-km windowed moving median. G18 and R14 filtered anomalies are merge by taking the mean of each year. Years with less than 30% coverage for fast moving ice (>= 200 m/yr.) were discarded. If missing annual values were within 25 km of a valid datapoint they are filled using natural neighbor interpolation, otherwise anomalies were set to zero. Outside of fast-moving areas, annual anomalies are tapered to zero using a 10-km cosine taper. Merged and filled annual anomalies are then smoothed one last time using a 5-km windowed moving mean. To create a continuous record of velocity, annual anomalies are interpolated in ...