1. Firn variability derived from a statistical analysis of airborne ice penetrating radar Thwaites Glacier catchment, West Antarctica -- Cyril Grima, D.M. Schroeder, D.D. Blankenship, D.A. Young 2013 IGS International Symposium on radioglaciology Lawrence, Kansas, USA September 9th, 2013

2. 2 =+=+ Scattering (incoherent) +++ Roughness +++ Non-deterministic structure (volume) + Permittivity Reflectance (coherent) +++ Permittivity +++ Deterministic structure (layering) + Roughness TOTAL POWER

3. = Reflected power (P c ) = Scattered power (P n ) Because of the scattered part, the surface echo is stochastic Amplitude distributions fitted with Homodyned K-envelope (HK) Increasing roughness HK allows within a footprint: few and clustered scatterers + one specular reflector [Dutt & Greenleaf, 1994; Chitroub, 2002; Ward et al., 2006; Destrempes and Cloutier, 2010] where J 0 is the 0 th order Bessel function of the first kind Demonstrated with SHARAD (Mars) [Grima et al., 2012]

4. 4 HiCARS radar [Peters et al, 2005,2007] f = 60 MHz ( = 5 m)  f = 15 MHz Footprint (along/cross-track) 30-50 m / 250-350 m Ice thickness sensitivity 5-10 m Amplitude distributions obtained along-track 1000 consecutive observations each

5. 5 dB REFLECTANCE (P c )SCATTERING (P n ) correlation between amplitudes & HK distribution < 95% i.e. less confident results due to roughness and/or permittivity heterogeneities

6. 6 where U = P c /P n LASER (@ 100 m baseline)RADAR (@ 5-50 m baseline) Small Perturbation Method (SPM) + Nadir approximation + Large correlation length [Grima et al, 2012] [Grima et al., in prep.] ---- SPM 1dB limit

7. 7 Small Perturbation Method + Nadir approximation + Large correlation length where [Grima et al, 2012] 2 2.2 2.4 2.6 2.8 3  = 0.4-0.5 Real dielectic constant for the first 5-10 m of firn

8. 8  = 0.4-0.5 0 0.1 0.2 0.3 0.4 > 0.5 Slopes [°] The anomaly (  > 2.5) is a vein (30-60 km wide) whose northern boundary matches a slope break (0.5°) across the whole dataset coverage (~500 km)

9. 9 What could explain  = 0.4 - 0.5 in a 5-10-m thick slice of ice ? Ice composition/structureMaximum expected range Corresponding  Crystals shape ad size [Achammer and Denoth, 1994; Mätzler, 1996] Random orientation~ 0 Temperature [e.g. Mätzler and Wegmüller, 1987] -40 to 0 °C< 0.04 Density (dry ice) [e.g. Kovacs, 1995] 350 to 917 kg.m -3 ~ 1.5 Wetness [e.g. Frolov and Machoeret, 1999] 0 to 10 %~ 2.1 Neutral impurities [Looyenga, 1965] 0 to 1 vol.ppm< 10 -5 Ionic impurities [e.g. Hallikainen, 1992; Fujita, 2000] 0 to 100 vol.ppm~ 0.05

10. 10 Consistent with a 550-kg.m -3 critical density at < 9 m in this region, as modeled by [Ligtenberg et al., 2011] [Kovacs et al., 1995] Anomly could be + 200 kg.m -3

11. 11 3-layers model [from Mouginot et al, 2009] 11 (1)(1) Solutions for  = 0.4-0.5  0 = 1  1 = ?? (>  2 )  2 = 2.3-2.4 ?? (Atmosphere) An upper high-permittivity layer with a subwavelenght thickness generates constructive interferences 5-10 m

12. 12  Dual frequency analysis (HF/VHF) to solve permittivity/layering ambiguities Reflectance/scattering components extracted from the signal 500 x 30 km permittivity anomaly detected over Thwaites catchment, coincident with a prominent slope break Higher density firn or wet snow layer  Both indicative of a higher densification rate  Implications for surface mass balance at regional scale  Application to bed interface

13. Thank you CreSIS for travel grant !