8 Satellite Microwave Radar Observations of Antarctic Sea Ice
179
Ice Station Weddell SAR Ice Drift Analysis
40
70
Day of Year, 1992
Fig.16. a-c. Time series ofERS-l SAR-tracked ISW ice motion results. a Measured surface wind speed
(dotted) and 3-day smoothed windspeed, b SAR- (dotted) and GPS-derived (solid) u component of
velocity (east is positive), c SAR (dotted) and GPS (solid,J-day running mean) v component of velocity (north is positive). For further details, see text
I6a. Additionally, though not in "free-drift;' SAR-reported drift velocity components
(dotted line in Fig. 16b,c) clearly respond to the 3-day filtered wind forcing. The largest
discrepancy between SAR velocity and 3-day smoothed ISW velocity occurs as a result
of the wind burst from day 63 to day 67, when ISW is pushed to the west at -15 cm S-1. In
the corresponding SAR image record, the closest pair of image acquisitions was at 6 days,
and thus the minimum in ISW u velocity (solid line) is missed. A higher temporal resolution in SAR imaging is required to capture such deviations in ice motion.
Generally, examples such as Fig. 16 indicate that the sea-ice drift field adjusts to the
wind. The temporal frequency of fluctuations in the field of ice velocity vectors closely matches those in wind velocity. Temporal sampling at 3-day intervals or less is concluded to be necessary if the small-scale response to synoptic storms is to be studied
in more detail in an open basin such as the Weddell Sea. The shorter timescale examples such as the I-day repeat orbits plotted at days 46,52, and 71 capture more variance
in ice velocity (indicated by the vertical bars for u and v velocity components). Spectral results in buoy data in Sect. 8.5.2.2 confirm this phenomenon to be attributable to
the aliasing of velocity components induced by tidal motions. Importantly, therefore,
exact 3-day repeat orbits do not alias diurnal or semidiurnal (which are close to 12 h
period) tidal motions, whereas nonmultiples of 12 h contain significant tidal variance
in this location. Ultimately, a trade-off is necessary between resolution in temporal sam-
179
Ice Station Weddell SAR Ice Drift Analysis
40
70
Day of Year, 1992
Fig.16. a-c. Time series ofERS-l SAR-tracked ISW ice motion results. a Measured surface wind speed
(dotted) and 3-day smoothed windspeed, b SAR- (dotted) and GPS-derived (solid) u component of
velocity (east is positive), c SAR (dotted) and GPS (solid,J-day running mean) v component of velocity (north is positive). For further details, see text
I6a. Additionally, though not in "free-drift;' SAR-reported drift velocity components
(dotted line in Fig. 16b,c) clearly respond to the 3-day filtered wind forcing. The largest
discrepancy between SAR velocity and 3-day smoothed ISW velocity occurs as a result
of the wind burst from day 63 to day 67, when ISW is pushed to the west at -15 cm S-1. In
the corresponding SAR image record, the closest pair of image acquisitions was at 6 days,
and thus the minimum in ISW u velocity (solid line) is missed. A higher temporal resolution in SAR imaging is required to capture such deviations in ice motion.
Generally, examples such as Fig. 16 indicate that the sea-ice drift field adjusts to the
wind. The temporal frequency of fluctuations in the field of ice velocity vectors closely matches those in wind velocity. Temporal sampling at 3-day intervals or less is concluded to be necessary if the small-scale response to synoptic storms is to be studied
in more detail in an open basin such as the Weddell Sea. The shorter timescale examples such as the I-day repeat orbits plotted at days 46,52, and 71 capture more variance
in ice velocity (indicated by the vertical bars for u and v velocity components). Spectral results in buoy data in Sect. 8.5.2.2 confirm this phenomenon to be attributable to
the aliasing of velocity components induced by tidal motions. Importantly, therefore,
exact 3-day repeat orbits do not alias diurnal or semidiurnal (which are close to 12 h
period) tidal motions, whereas nonmultiples of 12 h contain significant tidal variance
in this location. Ultimately, a trade-off is necessary between resolution in temporal sam-