Snow covered sea ice and its impact on radar altimetry

The snow cover is a fundamental component of the polar atmosphere-ice-ocean interaction system that features a high variation on temporal and spatial scale (Eicken, 2008; Sturm et al., 2002; Iacozza and Barber, 1999). It can feature a complex stratigraphy as a result of different conditions. During winter, subsequent depth hoar formation after snowfall occurs due to temperature-gradient metamorphism (Colbeck, 1982; Nicolaus et al., 2009). As a

1.2 Satellite altimetry over sea ice 11

Table 1.1. Synthetic Aperture Interferometric Radar Altimeter (SIRAL) instrument characteristics, modified from Wingham et al. (2006) and Bouzinac (2012).

Measurement mode

LRM SAR SARIn

Carrier frequency (GHz) 13.575

Antenna Gain (dB) 42

Along-track antenna 3 dB width 1.0766^◦

Across-track antenna 3 dB width 1.2016^◦

Transmitted (measured) bandwidth (MHz) 350 (320)

Transmitted power (W) 25

Transmitted (measured) pulse duration (µs) 49 (44.8)

Pulse repetition interval 1971 Hz 18.182 kHz 18.182 kHz

Burst repetition interval (ms) - 11.7 46.7

Samples per echo 128 128 512

Measurement range window (m) 60 60 240

Measurement range gate (m) 0.46875

Interferometer Baseline (m) - - 1.172

Tracking samples per echo 128

Tracking range window (m) 60 60 480

Tracking range gate (m) 0.469 0.469 3.75

Data rate 51 kbps 12 Mbps 24 Mbps

Power consumption (W) 95 130 125

Mass (kg) 61

Altitude (km) 717

Repeat cycles 369 days with 30 day sub-cycle

consequence, low-density horizons of faceted crystals with diameters of up to 1 cm can form, whereas a fresh snow layer consists of small grains of millimetre to sub-millimetre dimension.

Melting events, the downward transport of moisture and subsequent freezing can cause the forming of ice layers and ice lenses. Moreover, changing radiation and wind compaction also contribute to this high variability and inhomogeneity.

The accuracy of sea-ice thickness retrievals derived from radar altimetry depends on the freeboard retrieval. A snow cover establishes an additional uncertainty contributing to the remote sensing signature of sea ice (Sturm and Massom, 2009). Snow and ice lenses within the snow feature physical properties that affect the scattering of the radar signal (Hallikainen and Winebrenner, 1992), particularly at K_u-band frequency as used by CryoSat-2. In current literature it is widely assumed that the peak power of the returning CryoSat-2 radar echo is expected to originate at the snow-ice interface. This

12 Chapter 1 Introduction assumption is based on laboratory experiments by Beaven et al. (1995). They showed that a 13.4GHz radar echo originates at the snow-ice interface under dry and cold conditions with a uniform snow stratigraphy. Even by considering the increased Arctic melt season length, snow is not melting during the winter (Markus et al., 2009), but metamorphic processes and densification can occur. Data from airborneK_u-band radar altimeters and in-situ field measurements from the CryoVEx 2006 and 2008 campaigns were analyzed by Willatt et al. (2011). They reveal that in Spring 2006, at temperatures around the freezing point, the dominant scattering surface in 25 % of the radar returns is located close to the snow-ice interface whereas in 2008, when the temperatures were lower, this percentage rises up to 80 %. As a consequence an accurate estimation of sea-ice freeboard is only possible under dry and cold snow conditions with a known snow load (Makynen and Hallikainen, 2009) and without a distinct metamorphic history.

In order to isolate effects from scattering within the snow layer, it is useful to compare coincident measurements from laser and radar altimetry (Cullen et al., 2006). Whereas the laser ranges to the snow surface, the radar should measure the distance to the snow-ice interface (Kwok et al., 2004) if internal scattering in the snow is neglected. Hence, the difference in elevation between both sensor retrievals potentially indicates the snow depth, if the lower propagation speed of the radar pulse within the snow layer is considered. Those investigations have been accomplished by Giles et al. (2007) with airborne laser and radar altimeter measurements over the Fram Strait. Connor et al. (2009) compared Envisat radar and airborne laser altimeter measurements over Arctic sea ice and found differences in elevation which they associated with the snow layer to some extent. On the other hand, such comparisons between laser and radar may also serve to evaluate the radar retrieval if information about snow depth exist (Ricker et al., 2013, 2014a).

The snow stratigraphy on Antarctic sea ice features another component that contributes to its complexity. Due to the combination of relatively thin ice compared to the Arctic and high precipitation rates on the other hand, sea ice can be depressed beneath the sea level. Flooding of the ice surface can then occur by lateral incursion and/or by percolation vertically through the ice lattice. The resulting layer of slush is a mixture of ice crystals and water. It subsequently refreezes and builds a layer of snow-ice. Flooded snow layers can reach a height of 0.1- 0.2m (Massom et al., 2001). The dielectric constant of snow with water inclusion is about 40 times higher than of dry snow (Hallikainen et al., 1986) and hence a radar echo will be reflected by wet snow. Willatt et al. (2010) carried out the first field measurements to evaluateKu-band scattering effects within the snow layer on Antarctic sea ice during September and October 2007. They conclude that only snow without morphological features or flooding results in a snow-ice interface as the dominant scattering surface. Moreover, 43% of the returns originated at the air/snow interface, 30% from the snow/ice interface and23% from an internal layer. This results in a mean

1.2 Satellite altimetry over sea ice 13 depth of the dominant scattering surface that is about 50% of the mean snow depth (Willatt et al., 2010).

Simulating the interaction of radar waves with snow is a valuable method to get information about the scattering mechanisms at the interfaces within the snow layer. It can also contribute to our understanding of how the scattering horizons depend on the snow properties. Recent Studies about the simulation ofK_u-band altimeter echoes from sea ice have been presented in Tonboe et al. (2006); Makynen and Hallikainen (2009). Tonboe et al. (2006) use a radiative transfer model to simulate the sea ice effective scattering surface variability as a function of snow depth and density. They reveal that a snow cover might have a variable but significant impact on the estimation of the sea-ice thickness with radar altimetry. This model does not consider surface roughness and does not account for antenna gain function or pulse shape (Makynen and Hallikainen, 2009). In general this model is similar to Ridley and Partington (1988). Makynen and Hallikainen (2009) have built a simulator for ASIRAS (airborne radar altimeter, see Section 1.2.4) echoes over snow covered first year ice which is also valid for Cryosat-2 echoes. In contrast to Tonboe et al. (2006) they take antenna gain and pulse shape into account. The results show that for dry snow the leading edge originates at the sea-ice surface and the volume echo is negligible. Under moist snow conditions the snow-surface echo dominates. The power level of the echo decreases highly due to the attenuation of snow (Makynen and Hallikainen, 2009). Kwok (2014) analysed airborne snow and K_u-band radar data (Operation Ice Bridge, see Section 1.2.4) and concluded that scattering at the snow surface and within the snow layer is non-negligible.