Dual-frequency Altimeter of Envisat above the Antarctica Ice Sheet

The dual-frequency altimeter of Topex allows since few years the observations of earth from 62°N to 62°S both in Ku and in C band. Previous works has been already performed above continental areas by [1, 2, 3, 4] and above the southern part of the Greenland ice sheet [5, 6]. These works show the great interest of a dual-frequency above both for the retrieval of some geophysical parameters and for a better understanding of the physics of the measure. The RA2 altimeter on board ENVISAT allows for the first time such observations over the polar regions, up to 82°N and 82°S. The low frequency of the RA2 is lower than the one of Topex (S band, 3.6 GHz) also allowing a complementary observation of the continental area. This low frequency induces a greater penetration capability and a greater robustness of the waveform than the ku and C bands. These new characteristics are very promising for the study of polar regions, especially ice sheets. The paper is devoted to this new observation of both Ku band and S band over the Antarctica ice sheet. 1. METHODOLOGY AND MODEL The waveform data are retracked with the ice2 algorithm [7, 4] that is detailed in this issue [8]. The waveform parameter extracted from the altimetric waveform for both bands are showed on Fig. 1. We then have at our disposal the altimetric height, the radar backscattering, and two waveform shape parameters, the leading edge width and the trailing edge slope for the two frequencies. The editing, the atmospheric and geophysical corrections are performed, tested and validated in [4, 8]. The waveform is affected by the topography (slope and undulations), by the surface characteristics (roughness, snow density) and by the surface characteristics (internal stratification, ice grain size). The topography effect is modulated by the antenna pattern gain that depends on the square on the frequency, so that the S band is strongly less affected than the ku band. The subsurface signal is mostly due to internal stratification [9], even for the ku band and is controlled by the penetration depth so that the S band is expected to penetrate more deeply within the snowpack than the ku band. Fig. 1 Waveform parameters as deduced from the ice2 algorithm 2. STATISTICAL ANALYSIS OF THE ALTIMETRIC PARAMETER 2.1 Backscattering coefficient The dynamic, around 10-15 dB, and the pattern of the backscattering are very similar for both bands. The correlation coefficient between both is 0.965, so that up to 93% of the variances are similar. The effect of the surface slope is to decrease the backscattering only for ku band while we expected more subsurface echo for S band. Indeed, in average the backscattering in S band is slightly greater than for ku band one, but a large stripe crossing over the East Antarctica roughly between the 2000 m and the 3000 m isocontours and two large sectors in East Antarctica exhibit a ku backscatter values, larger than the S values of up to 2 dB. _____________________________________________________ Proc. of the 2004 Envisat & ERS Symposium, Salzburg, Austria 6-10 September 2004 (ESA SP-572, April 2005) Fig 2. Difference between both backscattering coefficients. (S-ku, expressed in dB) 2.2 Trailing edge slope The nominal value, e.g. for a flat surface without volume signal is 0.22 s for S band and 3.9 s for ku band. For ku band, as already observed with the ERS1 data, few areas show values letter than the nominal value associated with strong backscattering coefficient and a short leading edge suggesting a "pseudo-specular" signal due to a flat and smooth surface. The areas in this case for S band, are really more numerous. About 10% of the Antarctica show "pseudo-specular" waveforms in ku band while up to 30% show pseudo-specular S waveforms. This can be explained by the effect of the roughness that is perceived, with respect to the radar wavelength, letter for S band than for ku band. However, these areas can be associated with a large range of backscattering coefficient and of leading edge that contradicts the classical explanation for the pseudo-specular ku waveform already observed with the ERS-1 altimeter. The effect of the surface slope is to increase the trailing edge only for ku band while we expected more subsurface echo for S band, thus a larger trailing edge. However, the difference between both with respect to the nominal differences shows that the ku trailing edge is frequently greater than the S one, except in the central part. First, the histogram of both trailing edges shows that each distribution is Gaussian. The ku's one is shifted of 1 s with respect to the S one. This shift cannot be explained by the surface slope that increases the ku trailing edge: a mean surface slope of 4.6 m/km should be needed. It can thus be also explained by the volume echo that indeed increases the trailing edge. The weak value of the S trailing edge is due to the roughness effect that decreases the trailing edge. Fig.3 Trailing edge slope for S band (in s). The value of the encircle areas are letter than the nominal values (0.22 s) suggesting pseudospecular echo. Indeed, when the trailing edge is under its nominal value, the correlation between the backscattering and the trailing edge, usually slightly positive, is largely positive for the two bands. This confirms that the pseudo-specular waveforms are generated by a smooth surface and also explain why the S band values are letter than the ku band's one. This effect on S band is greater than for ku band because the radar wavelength is greater, but this effect also affects ku band, whose trailing edge slope is slightly compensated by surface slope effect. Nevertheless, he correlation coefficient between both is 0.91, so that up to 83% of the variances are similar, the dynamic of the trailing edge being less important for S band.