[15-Apr-2013] Skou, N., Kristensen, S., and Soebjaerg, S.
Presented at the 2013 SMOS-Aquarius Science WorkshopBackgroundESA's SMOS mission is faced with the challenging task of measuring the salinity of the oceans as well as the soil moisture over the continents, based on radiometric measurements of the natural emission from Earth. This is done using a microwave radiometer system operating in the protected radio astronomy band near 1.4 GHz (L-band). Especially the salinity measurements impose severe requirements to the brightness temperature measurement fidelity concerning stability and absolute calibration accuracy. Thus the SMOS instrument features several internal calibration loops, but the important antenna system is outside these loops, and in the end, checks of overall calibration, measuring known external targets, is a necessity throughout the lifetime of SMOS. The problem is that such calibration targets have to be not only stable and well known, but also of very extended size, hundreds of kilometers, due to the imaging properties of sensors like SMOS. One of the few feasible earthly targets that might live up to all expectations is the area around Dome-C in Antarctica. This will provide SMOS with a hot calibration point. Analysis of existing radiometric measurements (at higher frequencies) from a range of space missions (SMMR, SSM-I, AMSR-E) shows good stability and spatial homogeneity. An analysis conducted on SMOS data over a two years period confirms the good spatial and temporal stability not only for the area around Dome-C but over a large area of the East Antarctic Plateau. Moreover, the results obtained from ground radiometric measurements (i.e. DOMEX- 2 experiment), provides the final confirmation of the good temporal stability. What remains to be finally proven is good spatial homogeneity. This can only be done by launching a suitable airborne campaign. Another natural, extended calibration target is free space, and indeed SMOS measures this on a regular basis through suitable pitch maneuvers. Free space, when corrected for galactic radiation, provides a cold calibration point. Seen from a radiometer designer's point of view this is an ideal situation: the sensor in question regularly measures both a cold and a hot calibration target, meaning that instabilities can be monitored. Since the cold sky can only be measured occasionally (once each fortnight) the Dome-C hot target becomes extremely important as it is measured several times per day! Moreover, and very important: by having both a hot and cold calibration point, problems with possible simultaneous offsets and multiplicative calibration errors can be disentangled. The temporal stability of the area has already been verified by operating a ground based L-band radiometer from a tower at Dome-C for a full year, which demonstrated that especially at V polarization the brightness temperature is very stable showing an annual standard deviation lower than 0.5 K whilst at H polarization it is larger than 1 K because of the effect of snow layering as also demonstrated in several papers.Irrespective of this promising result, an outstanding issue to be investigated is the spatial variation of TB at SMOS sub-pixel scale, which is expected to be very small around Dome-C, but has, however, never been measured at L-band. Also, a potential azimuth variation (due to for example a pre-dominant wind direction) must be investigated. These issues are deemed important uncertainties and problems in relying on Dome-C as a major SMOS calibration site.In order to fill this gap a measuring campaign is carried out in the Dome-C area using a suitable airplane equipped with the EMIRAD L-band radiometer system. This radiometer system was designed for use in a range of airborne campaigns supporting the development of geophysical algorithms for SMOS, as well as being an important instrument in the Cal/Val campaigns. Detailed measurements are made over a 350 x 350 km area centered at the Concordia Station, but also important are the measurements that can be performed during the transits to and from Dome-C. There is a high interest in transects out to the coast, and also in a transect between Dome-C and the South Pole. EMIRAD-2 L-band Radiometer SystemThe EMIRAD-2 L-band radiometer system has been developed by DTU, and operated by DTU in a range of campaigns, known as the CoSMOS campaigns, in support of SMOS. It is a fully polarimetric (4 Stokes parameters) system with advanced RFI detection features (kurtosis and polarimetry). The system has operated successfully on different aircraft (C-130, Aero Commander, Skyvan) in Denmark, Norway, Finland, Germany, France, Spain, Australia). The main features of the system are:

• Correlation radiometer with direct sampling.
• Fully polarimetric (i.e. 4 Stokes parameters).
• Frequency: 1400.5 - 1426.5 MHz (-3 dB BW) 1392 - 1433 (-60 dB BW).
• Digital radiometer with 139.4 MHz sampling.
• DigitaI I/Q demodulation and correlation for accurate estimation of 3rd and 4th Stokes.
• Advanced analog filter for RFI suppression.
• Additional digital filter bank: 4 sub-bands.
• RFI flagging by kurtosis and polarimetry.
• Data integrated to 1 msec recorded on primary storage PC.
• "Fast data" pre-integrated to 14.4 ìsec. recorded on dedicated PC.
• Sensitivity: 0.1K for 1 sec. integration time.
• Stability: better than 0.1 K over 15 min. before internal calibration.
• Calibration: internal load, noise diode, and Active Cold Load (ACL).
• 2 antennas - one nadir pointing, one side looking at 45 deg. incidence angle.
• Antennas are Potter horns with 37.6 deg & 30.6 deg HPBW.
• Nadir horn has 415 m footprint from 2000 ft flight altitude.
• Tilted horn has 490 m by 640 m footprint, again from 2000 ft flight altitude.
• Each data package time stamped using GPS 1 PPS signal with 100 ns accuracy.
• Minimum operating altitude: 250 m above terrain @ 140 knots.

Further information about EMIRAD-2 is found in [Skou et al, 2006] and [Skou et al, 2010]. For the current campaign, EMIRAD is installed a Basler BT-67 aircraft owned and operated by AWI. The radiometer system features 2 large Potter horns as antennas. One antenna is mounted nadir looking, the other is tilted nominally 40 deg. The antennas are mounted in existing apertures in the fuselage. The radiometer itself is installed very close to the antennas to ensure short antenna cables (low loss). Also the inertial navigation unit is mounted close to the antennas in order to measure attitude correctly. Flight PatternsIn order to evaluate the homogeneity of the area around Dome-C a raster pattern is flown. The area around Concordia is covered by a grid of long lines, such that an area of 350 x 350 km is covered. The area around the tower, where the DOMEX experiment took place, is covered more intensely by a star pattern. The star pattern is centered on the observation tower at Concordia, and one arm of the star is in the view direction of the tower. In addition to the raster pattern and the star pattern, circle flights are carried out to examine a potential azimuthal signature. In order to obtain sufficient radiometric resolution, several circles must be flown and integrated during data processing. In order to take proper precautions against a possible Sun effect, the circle patterns must be carried out 2 times, many hours apart - for example early morning and late afternoon. This way it is possible to separate a potential azimuth signature from the ice surface and from the Sun intrusion. The Sun azimuth signature can be used also to correct the raster pattern measurements if necessary.When investigating azimuth signatures, an alternative to the circle flight pattern is the star pattern already described. Here fewer azimuth angles are sampled, but the crucial TB data are gathered during relatively short and straight flight lines making attitude stability easier to obtain. Thus the star pattern supports the circle pattern when azimuthal issues are investigated.Data ProcessingThe data processing takes place at DTU following the campaign, and it includes the following tasks:

• Calibration using the internal calibration loops and external liquid nitrogen cold target calibration carried out in the field before each mission.
• RFI detection and mitigation using kurtosis and polarimetry.
• Production of TB map of the area covered by the grid pattern, including estimates of uncertainties in the measurements and the final map product. Main descriptive statistics could be mean, variance, histograms,
• Analysis of the homogeneity of this map. Spatial statistics like spatial covariance functions, power spectra, estimate of correlation lengths.
• Production of the circle flight and signatures.
• Production of star pattern signatures.
• Analysis of azimuthal signatures in the morning and the afternoon circle flights. Extract Sun signature.
• Correct raster pattern measurements for Sun effects.
• Analysis of azimuthal signatures in the star pattern.
• Estimates of biases between tower measurements and airborne measurements.
• Cross comparison of airborne measurements with ancillary geophysical information.
• Comparison of airborne data with SMOS data.

Skou, N., S. S. Sobaerg, J. Balling, and S. S. Kristensen: "A Second Generation L-band Digital Radiometer for Sea Salinity Campaigns", Proceedings of IGARSS '06, 4p., July 2006.Skou, N., S. S. Sobjaerg, J. Balling, S. S. Kristensen, and A. Kusk: "EMIRAD-2 and its use in the SMOS Cal/Val Campaign", DTU Space, AR 502, pp. 1-59, June 2010.