New Passive Instruments Developed for Ocean Monitoring at the Remote Sensing Lab—Universitat Politècnica de Catalunya

Lack of frequent and global observations from space is currently a limiting factor in many Earth Observation (EO) missions. Two potential techniques that have been proposed nowadays are: (1) the use of satellite constellations, and (2) the use of Global Navigation Satellite Signals (GNSS) as signals of opportunity (no transmitter required). Reflectometry using GNSS opportunity signals (GNSS-R) was originally proposed in 1993 by Martin-Neira (ESA-ESTEC) for altimetry applications, but later its use for wind speed determination has been proposed, and more recently to perform the sea state correction required in sea surface salinity retrievals by means of L-band microwave radiometry (TB). At present, two EO space-borne missions are currently planned to be launched in the near future: (1) ESA's SMOS mission, using a Y-shaped synthetic aperture radiometer, launch date November 2nd, 2009, and (2) NASA-CONAE AQUARIUS/SAC-D mission, using a three beam push-broom radiometer. In the SMOS mission, the multi-angle observation capabilities allow to simultaneously retrieve not only the surface salinity, but also the surface temperature and an “effective” wind speed that minimizes the differences between observations and models. In AQUARIUS, an L-band scatterometer measuring the radar backscatter (σ0) will be used to perform the necessary sea state corrections. However, none of these approaches are fully satisfactory, since the effective wind speed captures some sea surface roughness effects, at the expense of introducing another variable to be retrieved, and on the other hand the plots (TB-σ0) present a large scattering. In 2003, the Passive Advance Unit for ocean monitoring (PAU) project was proposed to the European Science Foundation in the frame of the EUropean Young Investigator Awards (EURYI) to test the feasibility of GNSS-R over the sea surface to make sea state measurements and perform the correction of the L-band brightness temperature. This paper: (1) provides an overview of the Physics of the L-band radiometric and GNSS reflectometric observations over the ocean, (2) describes the instrumentation that has been (is being) developed in the frame of the EURYI-funded PAU project, (3) the ground-based measurements carried out so far, and their interpretation in view of placing a GNSS-reflectometer as secondary payload in future SMOS follow-on missions.


Introduction: Principles of L-Band Microwave Radiometry and GNSS-R over the Ocean
Sea Surface Salinity (SSS) is a key climatologic and oceanographic parameter. It has a significant influence on ocean currents (thermo-haline circulation), it is related to the evaporation minus precipitation, is responsible for El Niño Southern Oscillation (ENSO) events, and determines the life conditions of some species… Sea surface salinity can be determined by means of microwave radiometry. The largest sensitivity is found at P-band [1], but at L-band (1.400-1.427 MHz) there is a quiet frequency band where the sensitivity to SSS is still relatively "high" (~0.5 K/psu at 15 °C). However, the sea emission in the L-band depends on other parameters, such as the Sea Surface Temperature (SST) and mainly, the sea state (surface roughness) [2]. Furthermore, the relationship between the brightness temperature changes and the sea state is very complex, since the sea emission is not dominated either by capillary waves, nor by the long waves, and even swell effects have an influence.
Experimental activities conducted in the frame of the SMOS mission have shown that regressions in terms of the geophysical variables usually measured such as the wind speed (WS) and/or the significant wave height (SWH), exhibit too high a scattering (Figure 1) to provide a satisfactory correction to perform the sea state correction for the salinity retrieval, and that it is advisable to retrieve an "effective" wind speed as well, that can compensate for modeling errors [3].
Within the frame of the EURYI program, in 2003 the PAU (Passive Advanced Unit for ocean monitoring) project was proposed to the European Science Foundation to test the feasibility of using GNSS-R over the sea surface to make sea state measurements and, jointly with IR observations to obtain the SST, perform the corrections of the L-band brightness temperature [4]. GNSS-R was originally devised for altimetry applications [5], and here it is extended to try to obtain a direct correction for the sea state, without having to rely neither in numerical sea surface spectra models, nor in scattering and emission models.
The concept is simple: when the electromagnetic wave is scattered over the sea surface, the scattered signal changes its polarization (from RHCP to mostly LHCP) and arrives at the receiver mainly from the specular reflection point, determined by the shortest distance between the transmitting GPS satellite and the receiver, but when the sea is roughed, the scattered signals come from a wider region (known as "glistening zone") that enlarges with increasing sea state, in a similar manner as the Sun reflecting over the sea (Figure 2). Figure 1. Derivation of the brightness temperature sensitivity to wind speed. Scatter plots of the brightness temperature change at (a) horizontal (ΔT B h ) and (b) vertical (ΔT B v ) polarizations vs. the 10-m height wind speed (U 10 ). Solid line: linear fit, and dashed lines: percentile 50% as a function of U 10 for incidence angles from 25° to 65°. (c) Derived T B h,v sensitivity to wind speed as a function of (solid line) polarization and incidence angle, associated ±1 σ error bars, and (dashed lines) linear fit. All data points used ( Figure 6 from [3]

Instrument Developments
In order to test the feasibility of using GNSS-R over the sea surface to make sea state corrections to the radiometric data, the PAU (Passive Advanced Unit for ocean monitoring) project was proposed to the European Science Foundation in 2003, and it was granted in 2004 within the EURYI program. PAU is also a test bed of new technological demonstrators such as real aperture radiometers with digital beamforming and polarization synthesis, and fully-digital synthetic aperture radiometers etc. In the PAU system a suite of three different instruments operate in a synergetic way: (1) PAU/RAD: an L-band radiometer to measure the brightness temperature of the sea surface, A number of PAU system prototypes have been developed: (1) PAU-Real Aperture instrument with a 4 × 4 element array with digital beamforming and polarization synthesis that uses of an innovative pseudo-correlation radiometer topology to avoid the classical input switch in a Dicke radiometer and (2) PAU-Synthetic Aperture instrument, which is also used to test potential new technologically developments and algorithms for future SMOS missions.
In addition to these two, and in order to advance the scientific studies relating the GNSS-R and radiometric observables other PAU demonstrators have been developed: (3) PAU-OR with just one element for ground tests and algorithms development, and griPAU, an improved PAU-OR instruments fully automated, (4) PAU-ORA, a lighter version of PAU-OR for aircraft operations from a remote controlled plane, and (5) MERITXELL (Multi-frequency Experimental Radiometer With Interference Tracking For Experiments Over Land And Littoral) a classical Dicke radiometer, that includes not only L-band, but S-, C-X-, K-, Ka-, and W-bands, plus a multi-spectral camera, in addition to the PAU/IR and PAU/GNSS-R units.
In these technology demonstrators the input signals (PAU/RAD and PAU/GNSS-R) are the same: the L1-GPS band. This is not critical, since, due to the scattering on the sea surface, the scattered GPS signal is at least 23 dB below of the thermal noise signal (to be measured by PAU/RAD). Thanks to the 30.1 dB correlation gain, PAU/GNSS-R can detect the GPS signal when the correct C/A code is applied, and the error introduced in PAU-RAD observables is negligible, and it only occurs when the specular reflection is coming from the antenna boresight. Taking into account that the beam can be electronically steered the chance of having a significant interference is very remote.
Signals' bandwidth are limited to the GPS bandwidth (2.2 MHz) and are demodulated at an intermediate frequency (IF) of 4.309 MHz. Signals are then digitalized at 8 bits at a sampling frequency of 5.745 MHz, to allow the use of digital demodulation, using band-pass sampling techniques.
These different instruments are explained in the following sections.

PAU-Real Aperture
One of the technological goals of the project is to demonstrate the feasibility of combining in a single hardware two types of receivers: the radiometer itself (PAU/RAD) which, for stability reasons cannot be a total power radiometer, and the GNSS-Reflectometer (PAU/GNSSR). In order to be able to use the same receivers for both the radiometer and the GPS-reflectometer, a new radiometer topology has been devised ( Figure 5). Another technological goal, as challenging as the first one, is to provide an L-band radiometer with a digital beamforming. This capability is a very useful novelty in microwave radiometry, since it allows one to obtain simultaneously brightness temperature measurements at different incidence angles without mechanical scan.
As compared to a real aperture radiometer, instead of connecting the antenna output directly to the radiometer receiver, it is connected to the input of a Wilkinson power splitter that divides the signal in two signals that are in phase. However, the 100 ohm resistor of the Wilkinson power splitter that connects the two outputs also adds two noise signals that are 180° out-of-phase.

Remote Sens
Therefore difference o s proportion and sampled between the That is, the chopped, so  vertical-polarization signals are cross-correlated, the antenna temperature at vertical polarization is measured. If the horizontal-polarization signals are cross-correlated, the antenna temperature at horizontal polarization is measured. And if the vertical-polarization signal is correlated with the horizontal-polarization one, the third and the fourth Stokes elements are measured, as twice the real and imaginary parts of the complex cross-correlation. Note that, since the vertical and horizontal channels do not share a Wilkinson power splitter, there is no "Dicke effect" (output proportional to the difference between the antenna temperature and the physical temperature of a reference) for the third and fourth Stokes parameters. Figure 6 shows the schematic of the whole PAU-Real Aperture instrument. In addition, two calibration signals are required for calibration purposes: • uncorrelated noise, generated by a matched load at each input channel, to compensate for instrumental biases (cross-correlations must be zero), and • two different levels of correlated noise generated by a common noise source, to compensate for phase and amplitude mismatches among receivers. Details on the calibration techniques and beamforming can be found in [10][11][12]. Figure 7 shows the PAU-RA instrument: (a) in the UPC anechoic chamber for testing, (b) front-view without cover, and (c) inside view without cover and with foldable sides opened.

PAU-Synthetic Aperture
The receivers and frequency plan of PAU-Synthetic Aperture is the same as for PAU-Real Aperture, but since the number of receiving elements (Y-shaped array with 8 elements per arm plus a central one) is much larger than in the PAU-Real Aperture case (16 elements), limitations in the FPGAs processing capabilities, forced to retain only one receiver per element and polarization, and the Wilkinson input switch was removed (topology in Figure 5 could not be followed).
Another of the objectives of PAU-Synthetic Aperture was to test new techniques, technologies and algorithms that could be eventually used in future SMOS missions. To better understand this, Table 1 summarizes the main parameters of the MIRAS instrument aboard SMOS mission [13] and the PAU-Synthetic Aperture ones. The rightmost column lists the rationale behind each technological decision. The PAU-Synthetic Aperture is composed by a Y-shaped array of eight antennas per arm plus the one in the center, and additional dummy antenna at the end of each arm. Moreover, the four central antennas plus three additional ones (seven in total) are used to create a steerable array for PAU/GNSS-R to point to the specular reflection points (total = 31 dual-polarization antennas). Figure 8 shows the PAU-Synthetic Aperture topology, and two figures taken during the integration. Hardware and simulation details can be found in [16,17].

PAU-One Receiver and griPAU
In order to develop the science behind the GNSS-R observables and their relationship with the brightness temperatures, two simplified instruments have been developed, which consist of just one LHCP down-looking antenna, a seven LHCP path hexagonal array, and a RHCP up-looking antenna.
Two of these instruments have been built, one for ground based operations (PAU-One Receiver-rev 1- [18] and griPAU-rev 2- [19]), and another one for airborne operations (PAU-One Receiver Airborne, see Section 2.4).   (instead of f 0 = 1575 MHz, B = 2.2 MHz), and the second one for PAU/GNSS-R. The smaller up-looking patch antenna in the center is used to track the delay of the direct signal, and fed it to the reflectometer. Figure 10 shows the griPAU (PAU/GNSS-R) block diagram. This instrument includes an automatic tracking of the specular reflection point of the pre-selected GPS satellite to simplify instrument's operation, while at the same time ensures observations collocated in time and space, exactly in the 1.400-1.427 MHz band used for passive observations. Figure 11a shows the PAU-One Receiver Airborne on the bay of a remote controlled aircraft. Details on the control, telemetry, data links and data storage can be found in [20][21][22]. Figure 11b shows the DDMs measured when the direct (RHCP) and reflected (LHCP) signals are collected simultaneously using two different antennas connected to the inputs of a non-resistive 2-way power combiner. The left-hand side peak corresponds to the direct signal, which has a larger amplitude, while the right-hand side one corresponds to the reflected one, which is attenuated in the scattering and in the longer signal path. The separation between peaks is 21 samples, which correspond to ~770 m, since in this implementation of the instrument, the sampling frequency is 8.1838 MHz. Therefore, since the antenna was pointing to the nadir direction and the GPS satellite was close to the zenith, the estimated height is ~385 m, which is very close to the flight height (379 m). This design offers several advantages over the previous ones (just measuring the reflected signal) since it intrinsically provides absolute calibration of the scattering coefficient (ratio of peaks between direct and reflected DDM), if offer altimetry capabilities, and sea state determination using the full DDM, as shown in Section 3. PAU-ORA PAU-ORA Figure 11. Cont. c) d)

The Multifrequency Experimental Radiometer with Interference Tracking for Experiments over Land and Littoral (MERITXELL)
The MERITXELL radiometer is a step forward advancing our understanding of the potentials of combining data from different sensors: microwave radiometers, multi-spectral and TIR cameras, and GNSS Reflectometers. It will also be used in testing radio frequency interference (RFI) detection and mitigation algorithms for microwave radiometry.
The MERITXELL microwave radiometer is a multi-band dual-polarization Dicke radiometer covering eight protected bands used for passive remote sensing: L, S, C, X, K, Ka, and W (see Table 2). To add flexibility and simplify the design, a spectrum analyzer is used as IF stage, for filtering and power detection for all bands. This allows an easy reconfiguration of the band and/or frequency response shape, since the antennas and amplifiers response exceed those indicated in table 2. Antennas are 4 × 4 dual-polarization path arrays at L, S, and C bands, and horn antennas with a lens in the aperture to provide a quasi Gaussian beam for the other bands.

Field Experiments
During May-June 2008 the first PAU-One Receiver was deployed at El Mirador del Balcón, La Aldea de San Nicolás, in the North-West coast of Gran Canaria in the Canary Islands, and gathered for the first time ever L-band radiometric and GNSS-R data, together with oceanographic data (sea surface temperature + sea surface directional spectrum buoys). The field experiment was repeated during the same period of time in 2009 with an improved version of the instrument (griPAU) that collected radiometric and GNSS reflectometric data collocated both in time and space using two different antennas with the same 22° beamwidth (Figure 14). Now, we start to understand the relationship between the sea state and the GNSS-R observables (DDMs) and the changes in the brightness temperature. Figure 15 shows the scatter plot of the measured DDM volume (in arbitrary units) vs. the SWH for different threshold values. It can be appreciated that increasing the threshold decreases the sensitivity to SWH since a lower volume is being considered. However, this threshold cannot be arbitrarily small, since it has to be above the noise threshold to provide meaningful observations ( Figure 4). The correlation of the instantaneous brightness temperature changes and the instantaneous DDM volumes observed during ALBATROSS 2009 is shown in Figure 16 for incidence angles larger than 55°, since the cliff already imposed a 45° mask, and incidence angles between 45° and 50° were affected by multi-path.
As it can be appreciated, the behavior is very similar to that shown in Figure 1, including a zero crossing around 55° degrees at vertical polarization, which suggests that both descriptors (WS in Figure 1 and the volume of the DDM, V DDM , in Fig 16) are measuring the same phenomena.
Despite these encouraging results, there is still a long way to go until meaningful physical quantities that can be successfully extracted from satellite data to be used by the oceanographic communities, and they can be used to perform the sea state correction in sea surface salinity retrievals. More extensive data sets need to be gathered and processed. Figure 16. Estimated brightness temperature sensitivity to changes in the normalized DDM volume at vertical (red) and horizontal (blue) polarizations respecively [9].
The limited GNSS-R data gathered by the UK-DMC satellite and publicly available [23] shows the potential of this technique, and supported the proposal of a PAU secondary payload in SeoSat/INGENIO (Spanish Earth Observation Satellite) [24]. This proposal went through phase A, but did not succeed to pass into phase B due to the accommodation issues with the primary payload raised after a configuration change. Simplified, lighter and less power consuming payloads are currently under development in cooperation with industry and they will be available for future launches of opportunity in micro-or even in pico-satellites.

Conclusions
This paper has presented the basic principles of GNSS reflectometry and how these observables can be eventually used to correct for the sea state induced changes in the brightness temperature at L-band. These corrections [2] are needed to improve the quality of the sea surface salinity retrievals using L-band microwave radiometry. The suite of PAU instruments that has been (is being) developed at the Remote Sensing Lab of the Universitat Politècnica de Catalunya has been presented. These include: (1) PAU-Real Aperture with and array of 4 × 4 dual-polarization elements with digital beamforming and polarization synthesis, (2) PAU-Synthetic Aperture with a Y-shaped array of eight elements per arm to test digital technologies and new calibration techniques that can be eventually be used in future SMOS follow-on missions, (3) PAU-One Receiver and griPAU, one receiver only versions to demostrate the novel PAU radiometer topology concept ( Figure 5) and perform round-based field experiments, (4) PAU-One Receiver Airborne, one receiver only version to demostrate the concept from remote controlled aircraft, and (5) MERITXELL a version with multi-frequency Dicke radiometers, multi-spectral and thermal infrared cameras.
These instruments have been developed for two purposes: (1) to analyze the nature of reflectometric observables and their relationship with the brightness temperature at L-band, and (2) to be technological demostrators of improvements to be applied in future space-borne missions (SMOS follow-on missions), or secondary payloads that can help in the sea state correction.
The promising results of two field experiments carried out in Gran Canaria, during 2008 and 2009 have been presented. However, these techniques are not restricted to the observation of the ocean, but they can also be applied over land to infer soil moisture and vegetation height [25,26].