1236 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 45, NO. 5, MAY 2007 Use of C-Band Ground Penetrating Radar to Determine Backscatter Sources Within Glaciers Kirsty Langley, Svein-Erik Hamran, Kjell Arild Høgda, Rune Storvold, Member, IEEE, Ola Brandt, Jon Ove Hagen, and Jack Kohler Abstract—The question of penetration of synthetic aperture radar (SAR) signals at C-band frequency into polar glaciers is addressed by comparing ground penetrating radar (GPR) and SAR backscatter signatures. Profiles of the Kongsvegen glacier, Svalbard, were obtained with a C-band GPR. The received signal is converted to the equivalent radar cross section using the stan- dard radar equation, thus mapping the effective scattering sources within the glacier at this frequency. The depth of the observed scattering sources is greatest in the superimposed ice where layers are clearly seen to a depth of approximately 14 m. The very high scattering properties of the upper firn layers preclude layers deeper than approximately 6 m from being imaged. Integrating the radar cross sections over the depth gives a single backscatter value that we compare with the backscatter coefficient (scattering cross section per unit area) of the processed SAR data for the same profile. The comparison indicates that for coincidentally acquired GPR and SAR data, the radar cross section measured by the GPR does represent the features that contribute to the SAR signal. Index Terms—C-band ground penetrating radar (GPR), glacier, scatter cross section, synthetic aperture radar (SAR). I. I NTRODUCTION T HE SIGNALS measured by synthetic aperture radar (SAR) over glaciers penetrate some depth into snow and ice facies. This has implications for the use of SAR for the monitoring and mapping of glaciated areas. Since dry snow is highly transparent at microwave frequencies (as long as λ is much greater than the snow grain size) [1], the SAR signal originates from the ice and firn below the winter snow pack, from a range of depths, which are determined both by attenuation of the signal and the effectiveness of the scattering sources. These two factors are closely related because of the low loss of EM waves in ice. The firn area of the Greenland ice sheet is one of the brightest radar targets on Earth [2]. The presence of ice lenses Manuscript received January 5, 2006; revised September 18, 2006. This work was supported in part by the Norwegian Research Council (NFR) under Contract 155834, and by the Norwegian Polar Institute, Arctic Research Grant. Envisat ASAR data provided by the European Space Agency. K. Langley and J. O. Hagen are with the Department of Geosciences, University of Oslo, Blindern, 0316 Oslo, Norway (e-mail: kirstyl@geo.uio.no; j.o.m.hagen@geo.uio.no). S.-E. Hamran is with the Department of Geosciences, University of Oslo and the Norwegian Defence Research Establishment—FFI, 2027 Kjeller, Norway (e-mail: Svein-Erik.Hamran@ffi.no). K. A. Høgda and R. Storvold are with the Norut Information Technol- ogy, Forskningsparken, 9294 Tromsø, Norway (e-mail: kjell.arild.hogda@itek. norut.no; rune.storvold@itek.norut.no). O. Brandt and J. Kohler are with the Norwegian Polar Institute, Polarmiljøsenteret, 9296 Tromsø, Norway. Digital Object Identifier 10.1109/TGRS.2007.892600 and pipes at depths ranging from 0.4 to 2 m was originally considered to be the main cause of backscatter in this region at C-band frequencies [3], [4], but more recent studies suggest greater penetration depths. The effective penetration measured directly as the difference between interferometric heights and reference heights obtained with the GPS and laser altimetry measures 10 m [5], while modeling based on the decorrela- tion of interferograms gives an estimate of 7.5–17.5 m [6]. Rignot et al. [7] consider the phase center of the interferometric SAR (InSAR) data and estimate approximately 9-m penetration in the dry snow on the summit of Greenland, but only 4 m (±3 m) at high elevations on temperate Alaskan glaciers. In Antarctica, Rott et al. [8] measure penetration depths of the order of 20 m in the dry firn. There is limited discussion on the penetration depths in the superimposed ice (SI), except that it would be comparatively low, on the order of 1-2 m [7]. SI displays large variations in ice texture and air bubble content over depth representing the seasonal cycles of formation [9], [10]. Shallow cores show it to have a layered structure, with alternate white ice with small air bubbles (less than 1 mm in size) and clear ice with large and often elongate air bubbles (often more than 1 cm in size) [11]. Backscatter in this region is mainly attributed to volume scattering by the air bubbles. The literature indicates that there is a limited understanding of the penetration of C-band SAR signals in ice. However, con- sidering the fact that possible inclusions that cause volumetric scattering, such as air bubbles, ice and firn grains and clusters, are generally small and of low dielectric contrast, it seems reasonable to expect that the penetration could be substantial. Indeed, the penetration in pure ice at L-band has been shown to be greater than 60 m [12], even up to several hundred meters [13] and at least 2 m at X-band [14]. In order that the backscatter measured by SAR can be more fully understood, the type of interfaces and inclusions which are causing the backscatter, and their depth within the target medium, need to be determined. We propose a C-band ground penetrating radar (GPR) solu- tion that is capable of resolving the interfaces and inclusions that produce the backscatter signatures measured by the SAR. Since the two systems operate at the same center frequency, the backscatter response of the two instruments should be similar. Our objective was to identify the 5.3-GHz backscatter sources within a glacier that contribute to the SAR signal using the GPR. To do this, we designed and built a 5.3-GHz step- frequency GPR [15] and acquired simultaneously the Envisat C-band advanced SAR (ASAR) and GPR data. By converting 0196-2892/$25.00 © 2007 IEEE