In this paper, we assess the quality of the CS-2 Level 2 elevation data over the Antarctic ice sheet and its margins. For this purpose, CS-2 elevations are compared with laser altimetry data obtained from the ICESat mission. Our comparison provides results for two of three modes that the satellite operates in: Synthetic Aperture Interferometric (SARIn) mode is used for steeper slopes close to the ice sheet margins, and low resolution mode (LRM) on the flatter interior of the ice sheet. The LRM data exhibits a mean bias on the order of 1-2m and a residual slope-dependent bias on the order of 1-10m. The values are higher than what would be expected for a conventional radar altimeter. The mean bias may be due to the re-tracking approach used, or result from the choice of tracking point on the leading edge. In SARIn mode, we identify a problem with the across track angle error flag, which prevents pre-filtering of the data. This leads to an increase in outliers in the dataset. As in LRM mode, the slopeindependent bias is higher than expected. Also, we observe unusually high differences between CS-2 and ICESat elevations at higher slopes. We propose filtering the SARIn mode data by applying a cut-off for all data with backscatter values lower than 10dB. This greatly reduces differences and, while only discarding 16.3% of the data, provides a reasonable compromise for the use of the data while the Level-2 processing chain is optimised. Further, CS-2-CS-2 crossover differences are analysed and show good repeatability with a modal value of 7cm, and no spatial patterns or correlations. Large differences come only from locations with suspect elevations.
展开▼
