Helicoperborne Multi-frequency, Multi- Polarization Scatterometer Measurements during Ark Xix/1

Sea ice type classification based on active microwave remote sensing can be improved using multi-frequency, multi-polarization data. We present results of a first analysis of such data acquired by the helicopterborne multi-frequency (five different frequency bands), multi-polarization (all likeand cross polarizations) scatterometer HELISCAT during ARK XIX/1 around Svalbard. HELISCAT C band data were compared to almost coincident Envisat ASAR imagery yielding a correlation of 0.496 and 0.745 for two selected flight legs and a good agreement in the relative C band backscatter variation along the flight track. Changes in L, (C), and Ku band data are in good agreement with changes in sea ice thickness from electromagnetic thickness sounding, laser altimetry of the sea ice freeboard, and CCD-camera imagery. INTRODUCTION Spaceborne remote sensing sensors such as the synthetic aperture radar (SAR) aboard, e.g., the European Remote Sensing Satellites ERS-1/2, the Canadian RADARSAT-1, and the recently launched Envisat ASAR provide high-resolution information about different surface properties by means of backscatter measurements. Among other application fields, SAR data is used for high-resolution sea ice mapping, and has improved our knowledge about sea ice type distribution. However, the incidence angle dependency of the SAR signature of both sea ice and wind-roughened open water, the camouflaging effect of a snow cover, and the often very small changes in sea-ice backscattering properties of significantly different sea-ice types hamper sea ice type discrimination and lead to ambiguous results of classification methods used for sea ice mapping. Such ambiguities can be effectively reduced and the sea ice type discrimination can be improved if backscatter measurements of different frequencies and different polarizations are combined (Rignot and Drinkwater, 1994; Dierking et al., 2003). We present and discuss first sea ice backscatter measurements done with the helicopter-borne multi-frequency, multi-polarization HELISCAT instrument over 1 University of Hamburg, Centre for Marine and Climate Research, Institute of Oceanography, Bundesstrasse 53, D-20146 Hamburg, Germany Arctic sea ice in winter during the expedition ARK XIX/1 in 2003. The paper is organized as follows. An introduction of the HELISCAT system is given in the next section, followed by a brief description of the two flights considered here. The second section is about the results and is followed by some discussion and conclusion. THE INSTRUMENT The HELISCAT (HELIcopter SCATterometer) was developed and built at the University of Hamburg and is flown on a Messerschmidt-Bölkow-Blohm BO-105 helicopter. It operates at 1.25GHz, 2.4GHz, 5.3GHz, 10.0GHz, and 15.0GHz (L, S, C, X, and Ku band, respectively) and is capable of performing radar backscatter measurements quasi-simultaneously at the four polarization combinations VV, VH, HH, and HV (the first and the second letters denote the polarization of the transmitted and of the received microwave, respectively; V means vertical and H horizontal polarization). HELISCAT uses a single broad-band 96cm parabolic dish antenna both for transmission and reception. The antenna is aft-looking and can be tilted mechanically during the flight in such a way that the nominal incidence angle covers the range between 23° and 65°. The effective incidence angle, particularly at L and S band, horizontal polarization, may be slightly smaller, because of the antenna beam geometry. The system parameters of HELISCAT are given in table 1. Numerous experiments were carried out with the HELISCAT (primarily over open water) to help interpreting spaceborne data as, for example, of SIR-C/X-SAR (Gade et al., 1998; Wismann et al., 1998). However, the measurements shown and discussed in this paper are the first taken with the HELISCAT in its present form over Arctic sea ice and during winter conditions. Table 1: Specifications of the HELISCAT system. All measured data were stored on tape and were digitized after the campaigns. Because of the limited number of recording channels and since the two crosspolarization channels at a single radar band contain the same information, only time series at VV, HH, and HV polarization were recorded. In addition, time series of the Scatterometer Type Superheterodyne Doppler Scatterometer Antenna Type Parabolic Dish, 96 cm Ø Polarization HH, HV, VV, VH Nominal Flight Altitude [m] 150 Nominal Ground Speed [m/s] 50 Nominal Incidence Angle [°] 23–65 Pulse Repetition Frequency [kHz] 40 Radar Band L S C X Ku Frequency [GHz] 1.25 2.4 5.3 10.0 15.0 Output Power [mW] 150 100 40 10 10 Antenna Beamwidth (2-way; 3 dB) [°] 17.0 7.1 3.2 1.7 1.1 Antenna Footprint at 23° incidence angle [m × m] 53° incidence angle [m × m] 53.1×24.4 128.9×37.3 22.0×10.1 51.7×15.5 9.9×4.6 23.2×7.0 5.3×2.4 12.3×3.7 3.4×1.6 8.0×2.4 helicopter’s pitch and roll measured by a gyre were recorded simultaneously. Images acquired by a CCD-camera looking at the footprint of the antenna were also recorded on tape and complete the data set, together with GPS information and digital photographs taken at selected waypoints along each flight track. For the investigations presented herein only the variations in the backscattered radar power from different ice (and snow) surfaces are of interest. Therefore, HELISCAT was not absolutely calibrated (thus yielding the relative backscattered power, RBP, instead of the normalized radar cross section, NRCS). For each radar band time series at the three polarization combinations, HH, HV, and VV were sampled at a frequency of 10kHz. From these time series radar Doppler spectra (of length 4000) were calculated, and the integral of the spectral Doppler peak (within its 6dB limits) was used for the computation of the RBP time series with a sample rate of 2.5Hz (time steps of 0.4s). Note that, because of the wider beamwidths at L and S band, at a nominal ground speed of 50m/s and at a nominal flight altitude of 150m consecutive data points in the RBP time series at L and S band represent the backscattered radar power from overlapping surface areas. Only data with a signalto-noise ratio exceeding 3dB have been used, in order to avoid any influence of the instrumental noise on the obtained results. For this reason, we restricted our investigation on HH and VV polarization data. THE EXPERIMENT The presented data were acquired during two (April 15 and 19, 2003) of a total of 14 flights which took place during the cruise of the research icebreaker Polarstern ARK XIX/1 in the framework of the Winter ARctic Polynya Study (WARPS) and CRYOsat Validation EXperiment (CRYOVEX) between Feb. 28 and Apr. 24, 2003, near Svalbard. On April 15, the Polarstern was on drift station anchored to a large multiyear ice floe at about 81.81°N/10.22°E. The helicopter flight took place between 9:20 and 9:50 UTC along a triangle of about 60nm length, heading 180°, 60°, and 300°. Flight altitude was about 80m; the speed above ground was 50m/s. The sky was clear, wind speeds were around 5m/s, and air temperatures were around –18°C. Prevailing ice was multiyear ice with overfrozen leads. On April 19, the Polarstern was slowly heading south. The flight started at 81.26°N/ 10.60°E, took place between 18:30 and 19:10 UTC along a triangle of about 80nm length, heading 270°, 150°, and 30°. This flight was a tandem flight together with the helicopter-borne electromagnetic ice thickness sounder (EM-Bird) of the AlfredWegener Institute (AWI) mounted on a second helicopter flying about 100m ahead of the helicopter carrying the HELISCAT. Therefore, flight track and speed above ground (60m/s) are essentially the same. Flight altitude was again 80m. Weather conditions: little high cloud, wind speeds around 5m/s, and air temperatures around -15°C. There was no change in the prevailing ice type to April 15. RESULTS April 15 Data of this flight are presented here, because of the ability to compare relative variations in HELISCAT C band HH polarization backscatter with a quasi-coincident Envisat ASAR wideswath image (C band, HH polarization) acquired at 12:30 UTC, same day. A comparison was made for two specifically, i.e. by means of GPS data and digital images, selected legs along the flight track of the helicopter. Figure 1 shows a subset of the Envisat ASAR wideswath image superimposed by the two legs. Bright areas in this image correspond to high backscatter values as caused, e.g., by a rough surface (in case of surface scattering), or by porous or comparable ice/snow properties (in case of volume scattering). This can be rough and/or deformed first-year ice and/or multiyear ice. Dark areas correspond to low backscatter values as caused, e.g., by a smooth surface – which can be level and/or young ice or a calm open water area. A correction was applied to the flight track taking into account a possible drift of the Polarstern attached to the multiyear ice floe within the three hours between the HELISCAT flight and acquisition of the ASAR image. Here it was assumed that the movement of this ice floe corresponds to the drift of the Polarstern during the three hours period mentioned, and that this movement is representative for the drift of the entire sea ice region covered by the helicopter flight. The Polarstern drifted by about 50m towards north and by about 700m towards east between HELISCAT flight and ASAR image acquistion. Moreover, the ASAR image was corrected for the incidence angle variation across the subarea shown in figure 1 a), assuming that the sea ice incidence angle dependency can be neglected over this lateral distance (50km). Figure 1: a) Zoom of a near-range (25° incidence angle) subarea of an Envisat ASAR wideswath image acquired at 12:30UTC on April 15, 2003, superimposed with two legs (white lines) of the flight track of the helicopter (flight direction is indicated by the arrows); pixel size is 75m x 75m; b) Comparison of HELISCAT (thick line) and Envisat ASAR (thin line) C band HH polarization backscatter along leg 1, averaged over 500m; c) Scatterplot of backscatter values shown in b); correlation of both data sets is 0.745; d) same as c) but for leg2; correlation of both data sets is 0.495. The good agreement between HELISCAT and Envisat ASAR data puts some confidence to the HELISCAT measurements of which one more example is shown in figure 2. This figure 2 shows a comparison of C and Ku band data: profiles of the leg 1 leg 2 a) b)