The Inverse Barometer Effect on Antarctic Ice Shelves

Short-term height variations of Antarctic ice shelves must be removed before satellite-derived surface height can be used to assess climate trends, and satellite radar data can be used to determine ice motion. The dominant source of high-frequency vertical motion of ice shelves is their response to tides in the ocean cavities under the shelves. The tide-induced range (low tide to high tide) can exceed 7 m in the southern part of the Ronne Ice Shelf in the Weddell Sea. However, tides are now relatively predictable, and so we now focus on a second source of ice shelf vertical motion, which is due to the response of the underlying ocean to changes in atmospheric pressure (P_air). This response is known as the inverse barometer effect (IBE). Open-ocean measurements of P_air and sea level agree with the theoretical response of ~1 cm per millibar for low frequency variability of P_air. By looking at data from ice shelves, we find that that shelves experience a response of similar magnitude. A simple correction for the IBE is justified for ice shelf response to low-frequency (<0.5 cycles per day) of P_air. At higher frequencies the IBE becomes weaker. The IBE contribution to the ice shelf surface height, h_IS, can exceed 50 cm, with typical magnitudes of ~10-20 cm. Although the IBE is usually smaller than the tidal contribution to h_IS, the tide can be removed with current Antarctic tide models with an accuracy similar to the IBE. Global atmospheric models, however, do not presently predict P_air with sufficient accuracy to be used to correct measured variability of h_IS. Thus, in the absence of concurrent in situ P_air data, the IBE is a major source of error in correcting ice shelf heights for the tasks we mentioned above.

An Example: Amery Ice Shelf

In the figure at below we show:

(a) Top Left: In blue, the time series of hourly height (cm) from a GPS record from the Amery Ice Shelf (AIS) after removal of the mean. The dominant signal here is the tide, which has not yet been removed. The red line shows the prediction of the IBE, i.e. -P_air.

(b) Lower Left: In blue, the same time series after the mean and tides have been removed. The scale of the y-axis is much smaller than in the previous plot. Again, the red line in the IBE estimate from P_air.

(c) Upper right: In blue, the spectrum of detided residual height from the GPS measurements and in red, the spectrum of P_air. Note that the tide removal is not perfect: some energy remains near ~1 and ~2 cycles per day.

(d) Lower Right: The coherence-squared between the two spectra in the previous plot (detided h_IS and P_air). This plot shows that coherence is high at low frequencies (>0.8 up to 0.5 cycles per day), but low at higher frequencies. We interpret this to imply that the ice shelf responds to the IBE at low frequencies (which includes most of the energy in the spectrum of P_air>), but that the ice shelf does not respond to rapid changes in P_air>, such as when weather fronts pass through.