Sensors 2011, 11, 719-742; doi:10.3390/s110100719 OPEN ACCESS sensors ISSN 1424-8220 www.mdpi.com/journal/sensors Article CAROLS: A New Airborne L-Band Radiometer for Ocean Surface and Land Observations Mehrez Zribi 1,2,*, Mickael Pardé 2, Jacquline Boutin 3, Pascal Fanise 2, Daniele Hauser 2, Monique Dechambre 2, Yann Kerr 1, Marion Leduc-Leballeur 2, Gilles Reverdin 3, Niels Skou 4, Sten Søbjærg 4, Clement Albergel 5, Jean Christophe Calvet 5, Jean Pierre Wigneron 6, Ernesto Lopez-Baeza 7, Antonio Rius 8 and Joseph Tenerelli 9 1 CESBIO (CNRS/IRD/CNES/UPS), 18 avenue Edouard Belin, 31401 Toulouse Cedex 9, France; E-Mail: [email protected] 2 Université Versailles St-Quentin, UPMC, CNRS/INSU, LATMOS-IPSL (UVSQ/CNRS/UPMC), 11 Boulevard d’Alembert, 78280 Guyancourt, France; E-Mails: [email protected] (M.P.); [email protected] (P.F.); [email protected] (D.H.); [email protected] (M.D.); [email protected] (M.L.-L.) 3 LOCEAN, Tour 45-55-5eme étage, Paris, France; E-Mails: [email protected] (J.B.); [email protected] (G.R.) 4 DTU-Space, National Space Institute, DTU, B. 348, DK 2800, Lyngby, Denmark; E-Mails: [email protected] (N.S.); [email protected] (S.S.) 5 CNRM/GAME (Météo-France, CNRS), URA 1357, Toulouse, France; E-Mails: [email protected] (C.A.); [email protected] (J.C.C.) 6 INRA, UR1263 EPHYSE, F-33140 Villenave d'Ornon, France; E-Mail: [email protected] 7 Facultat de Fisica, Universitat de Valencia, C/Dr Moliner, 50, Burjassot, 46100 Valencia, Spain; E-Mail: [email protected] 8 IEEC/ICE-CSIC, Campus UAB/Fac Ciencies, Torre C-5-parell-2a planta, 08193 Bellaterra, Spain; E-Mail: [email protected] 9 National Institute of Marine Research (CLS), Brest, France; E-Mail: [email protected] * Author to whom correspondence should be addressed; E-Mail: [email protected]. Received: 24 November 2010; in revised form: 4 January 2011 / Accepted: 7 January 2011 / Published: 12 January 2011 Abstract: The “Cooperative Airborne Radiometer for Ocean and Land Studies” (CAROLS) L-Band radiometer was designed and built as a copy of the EMIRAD II Sensors 2011, 11 720 radiometer constructed by the Technical University of Denmark team. It is a fully polarimetric and direct sampling correlation radiometer. It is installed on board a dedicated French ATR42 research aircraft, in conjunction with other airborne instruments (C-Band scatterometer—STORM, the GOLD-RTR GPS system, the infrared CIMEL radiometer and a visible wavelength camera). Following initial laboratory qualifications, three airborne campaigns involving 21 flights were carried out over South West France, the Valencia site and the Bay of Biscay (Atlantic Ocean) in 2007, 2008 and 2009, in coordination with in situ field campaigns. In order to validate the CAROLS data, various aircraft flight patterns and maneuvers were implemented, including straight horizontal flights, circular flights, wing and nose wags over the ocean. Analysis of the first two campaigns in 2007 and 2008 leads us to improve the CAROLS radiometer regarding isolation between channels and filter bandwidth. After implementation of these improvements, results show that the instrument is conforming to specification and is a useful tool for Soil Moisture and Ocean Salinity (SMOS) satellite validation as well as for specific studies on surface soil moisture or ocean salinity. Keywords: radiometer; CAROLS; L band; SMOS; ocean salinity; soil moisture 1. Introduction Passive microwave remote sensing of Soil Moisture (SM) and Sea Surface Salinity (SSS) has been at the center of attention of many research programs for several decades. Various airborne and in situ radiometers have been developed, showing the high potential of L-band measurements for the estimation of surface parameters [1-6]. The SMOS satellite mission is based on an aperture synthesis L-band radiometer [7,8], designed and developed by the European Space Agency (ESA). In the context of the validation activity for the SMOS mission, the authors proposed to design, build and operate the CAROLS L-Band radiometer from an aircraft. Because the sensitivity of L-band brightness temperature to salinity is very small (−0.45 K/psu at a physical temperature equal to 105 K), it was necessary to build a very accurate, sensitive, and stable system. The radiometer was installed in the French research ATR-42 aircraft. In coordination with in situ field campaigns, it is used together with a suite of other airborne instruments: a scatterometer—or a C-Band radar—used to measure ocean roughness and winds, a GPS receiver to estimate ocean surface roughness, an infrared radiometer to measure sea surface temperature, and a visible wavelength wide angle camera to provide a qualitative description of the surface. The first campaign was conducted in September 2007, for the qualification and certification of the aforementioned instruments. Following various improvements to the CAROLS instrument, a second campaign was carried out in November 2008, in order to validate the CAROLS data quality. In spring 2009, a scientific campaign was organized, to acquire different types of brightness measurements over oceanic and land surfaces. The main objective of this third campaign was to validate the direct and inversion models used to estimate Soil moisture and Sea Surface Salinity. Sensors 2011, 11 721 This paper presents a detailed description of the CAROLS instrument, of its refined calibration system, and of the first results obtained from the campaigns carried out in 2007, 2008 and 2009. In Section 2, we present the instruments used in our campaigns, in particular the CAROLS radiometer, and in Section 3 we describe the campaigns. The CAROLS data quality is discussed in Section 4. Our conclusions are given in Section 5. 2. Airborne Instruments and Interface with the ATR-42 2.1. CAROLS Radiometer 2.1.1. System Description 2.1.1.1. General System Design CAROLS is a total power radiometer and has a simple structure and high theoretical sensitivity. The receiver was developed as a copy of the EMIRADII radiometer [9,10], in collaboration between the TUD (Technical University of Denmark) and the LATMOS laboratory (Laboratoire Atmosphères, Milieux, Observations Spatiales). It is a fully polarimetric correlation radiometer using direct sampling [9,10]. The antenna system comprises two large, identical Potter horns and two waveguide orthomode transducers (OMT). These provide dual–incidence measurements, useful for the estimation of soil moisture or ocean salinity from brightness temperatures. The CAROLS system periodically performs two-point calibrations, to calibrate the receiver gain and noise temperature by means of noise injection. In such a calibration system, the receiver looks at known sources and makes use of their associated outputs to infer the relationship between the output value provided by the A/D converter and the antenna temperature. A matched load is coupled with a noise source, which can be switched on and off, in order to measure two different brightness temperatures. Laboratory calibration of the instrument consists in determining a set of parameters relating the power received from the noise source to its physical temperature. All of the analog components are placed inside a double insulated and thermally regulated box and their physical temperatures are recorded in real time by means of temperature sensors. The accuracy of the receiver is highly dependent on the stability of the noise source and other components used in the noise injection circuitry. 2.1.1.2. Technical Specifications The main technical specifications are presented in the following table (Table 1) where AFE and DFE refer to the ‘Analog Front End’ and ‘Digital Front End’ electronics respectively: Table 1. Main technical specifications of the CAROLS radiometer. CAROLS: correlation radiometer with direct sampling Receiver type Total power receiver Operating frequency [1.4–1.427 GHz] Bandwidth 24 MHz@ −3 dB Polarization Polarimetric (4 stockes parameters: I, Q, U, V) Sensors 2011, 11 722 Table 1. Cont. CAROLS: correlation radiometer with direct sampling Sampling rate 139.4 MHz Integration time 1 ms and 1.8 s Internal calibration Load and noise diode Sensitivity 0.1 K for 1 s integration time Stability 0.1 K System noise temperature 150 K Physical temperature of AFE 45 °C Physical temperature of DFE 90 °C 2.1.1.3. Detailed System Design A. Receiver Unit a. Analog Front End (AFE) Electronics The Analog Front End (AFE) is composed of several microwave components (switch, director coupler, noise source, filter, isolator). The antenna and receiver unit is illustrated in Figure 1. Figure 1. Block diagram of the antenna and receiver unit. Sensors 2011, 11 723 This unit ensures a gain of 81 dB for each channel and an isolation of approximately 80 dB between the H (Horizontal) and V (Vertical) channels. Special attention was paid to the performance of the bandpass filter design. This device, based on cavity properties, provides a center frequency of 1,413 MHz and a 3 dB bandwidth of 24 MHz. The filter was specially tuned to attenuate interfering signals lying outside the bandwidth of interest. The performance of the analog front end is strongly dependent on its physical temperature. Therefore, to achieve the desired radiometric sensitivity, the measurement of inside temperatures is very important for the calibration process. The receiver is fitted with approximately ten thermal sensors, placed at various locations. The microwave section is thermally controlled using a PI (Proportional Integral) regulation algorithm. The AFE demonstrated its excellent ability to stabilize the internal temperature at 45 degree: during a 4-hour flight this varies by only 0.1 degree. b. Digital Front End (DFE) Electronics The analog signal output from the analog receiver is directly digitized by two A/D converters, with a sampling frequency of 139.4 MHz. The unit is designed for sub-harmonic sampling of the analog signals, i.e., direct sampling of the L-band signal with no analog down conversion. The signals produced by the converter are fed into a FPGA (Field Programmable Gate Array), in which 1.8 µs and 1 ms integrations are computed. The 1.8 µS integration period is useful during post processing, for the removal of Radio Frequency Interferences (RFI). B. Antenna Unit The antenna unit comprises a horn and an orthomode transducer (OMT). Each horn has an identical pattern and a 37.6° half-power beam width (HPBW), for both horizontal and vertical polarizations. This large beam and corresponding antenna dimensions were specified in order to fit the lower fuselage apertures of the aircraft. The antenna system (OMT and Potter horns) is characterized by almost ideal radiation patterns with practically no side-lobes and low losses. The performance of the Potter horns was tested at the CNES’ (Centre National d’Etudes Spatiales) anechoic chamber, and has been judged to be very satisfactory, with its side and back lobes being suppressed by more than 30 dB, and an even lower cross-polarization level. The performance of each OMT was also measured. Isolation between ports was greater than 49 dB for all frequencies between 1.4 and 1.427 GHz. C. Antenna and Cable Losses Each attenuator not only attenuates the incoming signal, but also introduces an associated thermal noise [1]. In the aircraft, each antenna was insulated to minimize temperature variations within its structure. Indeed, it was our aim to maintain identical temperatures at each port. The temperature of each OMT and horn was continuously monitored by means of a set of thermistors. Because coaxial cable losses are the main sources of signal error, a significant effort was made to very accurately determine its influence. Four 4.5 m long cables are needed to connect each antenna to the receiver unit, as shown in Figure 2. These cables transport signals (corresponding to the vertical and horizontal polarizations), which require predictable insertion phase and insertion loss characteristics. For the Sensors 2011, 11 724 present application, it was decided to use SUCOFEED 1¼ coaxial cable from HuberSuhner, which has very low losses (0.19 at 30 °C) and a very low drift coefficient (0.055 K/C). Because the temperature in the aircraft cabin is not stabilized, the coaxial cables are insulated and equipped with five temperature sensors mounted inside their insulating cladding. Insertion phase is also a very important characteristic for RF cables. Although phase matched cables are not mandatory for radiometric applications, phase-stabilized cables must be used. The relative phase matching requirement is satisfied by matching the phase, as opposed to the absolute length, of the two cables linked to each antenna relative to each other. Figure 2. The CAROLS radiometer inside the ATR42 research aircraft, (a) illustration of nadir antenna inside the aircraft, (b) illustration of CAROLS system (receiver and antennas) inside the aircraft. 2.1.2. Performance of the Instrument 2.1.2.1. Sensitivity The radiometric sensitivity (or radiometric resolution), referred to as NEDT, is the minimum detectable change in the radiometric antenna temperature of the observed scene. This sensitivity is directly related to the rapid fluctuations of the measured radiometric temperature. The radiometric sensitivity, related to the thermal noise NEDT of the radiometer, is given by: T T in R NEDT  (2) B where T is the temperature at the input of the receiver, T is the noise temperature of the receiver,  in R the integration time and B the bandwidth of the receiver. In Figure 3, we illustrate the theoretical sensitivity and the CAROLS sensitivity measured in the laboratory, as a function of integration time. It can be seen that for an integration time longer than 300 ms, the sensitivity is better than 0.1 K. 2.1.2.2. Linearity The linearity of the radiometer receiver, over a range of temperature covering natural ocean and land emissions, is crucial to obtain consistent data. A proven technique for linearity validation involves the use of a cryogenic load providing a very cold brightness temperature at the input, a variable Sensors 2011, 11 725 attenuator, and a data collection system at the system output. For each attenuation value, the receiver output is determined with the noise diode being switched on and off. This experiment is run when the system is correctly regulated, and it is assumed that the noise diode adds a constant noise, at the receiver input. The linearity measurements indicate worst-case scale variations of the order of 0.4% (Figure 4). Figure 3. Radiometric resolution of the CAROLS radiometer. Figure 4. Illustration of CAROLS receiver linearity. 2.1.2.3. Stability Figure 5 provides two plots of calibrated temperatures, measured by CAROLS in the laboratory, when exposed to a target simulated by a cryogenic load (with a stable temperature of 80.1 K). The black points correspond to measurements calibrated using only one initial calibration period, before the cryogenic load measurement, whereas the circles indicate measurements calibrated using two calibration times (before and after the cryogenic measurements), separated by an interval of 30 minutes. For each calibration, the gain and the receiver noise temperature are estimated with help of two calibrated points (load and load + noise diode). Using two calibration phases, the measured gain and receiver noise temperature are interpolated. We show that, whereas for the first calibration point Sensors 2011, 11 726 only, if the temperature drift is approximately 0.1 K (between 16.1 H and 16.4 H), after 20 mn, the use of two calibration times leads to a variation in measured temperature of less than 0.08 K, after 30 mn. Figure 5. Stability of the CAROLS measurements. Non corrected data correspond to one calibration point; corrected data correspond to two calibration points before and after data acquisition. 2.2. Auxiliary Airborne Instruments STORM Radar: STORM, developed at LATMOS [11,12], is a C-Band radar (5.35 GHz) based on an FM/CW (Frequency-Modulated/Continuous Wave) transmitted waveform. It transmits alternatively in the H and V polarizations, and simultaneously receives in both of these channels. The antenna is mounted so as to point towards the ground with a 20° incidence angle (measured with respect to nadir), when the aircraft’s attitude is horizontal. STORM data are acquired over the ocean to provide information concerning the wind and waves, to provide complementary data to the CAROLS radiometric measurements, which are sensitive to wind, sea surface temperature and sea surface salinity. GPS GOLD-RTR System: The GPS system was supplied by the IEEC (Institut d'Estudis Espacials de Catalunya or Institute for Space Studies of Catalonia) [13,14]. It is composed of three antennas and a GPS Open Loop Differential Real Time Receiver (GOLD-RT). The aim is to retrieve information concerning sea surface roughness, in the form of the effective Probability Density Function (PDF) of the sea surface slopes. Infrared Radiometer (IRR): The infrared radiometer is part of the standard equipment of the research ATR42. It is used to provide surface temperature estimations, simultaneously with the CAROLS measurements. Visible Wide Angle Camera (VWAC): It is located slightly forward of the slant radiometer antenna and allows the areas covered by the CAROLS nadir antenna measurements to be identified. Sensors 2011, 11 727 2.3. Description of the Interfaces with the Research ATR-42 Aircraft The research ATR42 has been available for research experiments since early 2006 (http://www.safire.fr). It is an original ATR42-320, specifically modified for scientific use. Because of limitations in terms of the existing lower fuselage apertures in the aircraft, i.e., only two openings (one forward aperture: 650 × 450; one aft aperture: 700 × 450 mm), two configurations were proposed for the CAROLS campaign flights, as shown in Figure 6. In the first configuration, providing dual-incidence data, the CAROLS radiometer was operated with two antennas, one pointing to nadir for vertical measurements, and the other pointing to the right side of the aircraft (slant antenna), at an approximately 30° incidence angle. The second configuration combines the CAROLS slant antenna, placed at the aft of the aircraft, with the STORM instrument placed at the front (at the location of the nadir antenna used in the first configuration). This second configuration is particularly useful for ocean measurements. The true incidence angles corresponding to each brightness temperature measurement were calculated accounting for the attitude of the aircraft (roll, pitch), with an accuracy of 0.1 degrees. Figure 6. a. (top) CAROLS instrument with one slant antenna and the STORM instrument, b. (bottom) CAROLS instrument with two antennas (one slant and one nadir). Sensors 2011, 11 728 3. Description of the Carols Campaigns 3.1. Flight Descriptions The CAROLS radiometer was operated with the two configurations described above (with one or two antennas), during all of the CAROLS flights. Table 2 provides details of the three CAROLS campaign flights. All measurements were made from altitudes ranging between 600 m and 3,000 m. Different flight patterns were carried out during the oceanic flights, in addition to straight horizontal flight portions: wing wags with a ±25° roll; nose wags with a ±5° pitch; and circular flights with 15° positive and negative roll angles. These movements are useful for the qualification of the instrument’s behavior and validation. The wing wags provide measurements over a large range of incidence angles, which are particularly useful for the validation of the emissivity models. They allow inter-comparisons between measurements recorded with the two antennas, thus making it possible to validate the loss term estimations and antenna alignment. Finally, they also allow the two polarization channels to be phase matched. The circular flights were defined in order to measure the full azimuthal variation of the Stokes vector, within the shortest possible time. Table 2. Description of the flights undertaken during the CAROLS campaigns. CAROLS’2007 CAROLS’2008 CAROLS’2009 Period 24/09/2007–28/09/2007 12/11/2008–24/11/2008 23/04/2009–28/05/2009 Number of flights 4 flights 4 flights 13 flights Gulf of Biscay, South Gulf of Biscay, South Gulf of Biscay, South West Studied sites West of France West of France of France, Valencia site Departure Toulouse (France) Toulouse (France) Toulouse (France) 3.2. Measurement Sites We selected three different test sites for the CAROLS data acquisitions, over land and ocean surfaces. Simultaneously to the CAROLS measurements, different in situ measurements were acquired in order to qualify the CAROLS data, the direct emissivity models, and finally the inversion algorithms. The ocean measurements were made over the Bay of Biscay [Figure 7(a)]. In order to support the Airborne CAROLS experiment, ship campaigns were organized by the LOCEAN team for each CAROLS campaign, along the same track as that used by the aircraft. Measurements of SSS, sea surface temperature (SST), wind speed, foam, wave spectrum, heat and momentum flux were collected on the ship. Different drifters measuring SSS and SST were deployed. All flights passed over or close to the Gascogne Buoy (45.2°N, 5°W), which is an operational buoy operated by the French Meteorological Office providing SST, pressure, wind, wave spectrum. This first land site, characterized by a temperate climate, is situated in the South West of France [Figure 7(b)]. The Soil Moisture Observing System—Meteorological Automatic Network Integrated Application (SMOSMANIA) determines soil moisture values, based on a portion of automatic ground station network (the RADOME network) [15].