The Greenhouse Gases Observing Satellite (GOSAT) monitors carbon dioxide () and methane () globally from space using two instruments. The Thermal and Near Infrared Sensor for Carbon Observation Fourier-Transform Spectrometer (TANSO-FTS) detects gas absorption spectra of the solar short wave infrared (SWIR) reflected on the Earth’s surface as well as of the thermal infrared radiated from the ground and the atmosphere. TANSO-FTS is capable of detecting three narrow bands (0.76, 1.6, and ) and a wide band () with spectral resolution (interval). The TANSO Cloud and Aerosol Imager (TANSO-CAI) is an ultraviolet (UV), visible, near infrared, and SWIR radiometer designed to detect cloud and aerosol interference and to provide the data for their correction. GOSAT is placed in a sun-synchronous orbit at 13:00 local time, with an inclination angle of . A brief overview of the GOSAT project, scientific requirements, instrument designs, hardware performance, on-orbit operation, and data processing is provided.
© 2009 Optical Society of America
1A. Greenhouse Gases Measurements from Space
Carbon dioxide () and methane () are two major greenhouse gases (GHGs) and are long-lived in the Earth’s atmosphere. A very accurate observation is required to detect the long term change and to estimate the sources and sinks of GHGs. Satellite observation is a very powerful method for this purpose, as long term and frequent global observation can be obtained using a single instrument under a stable environment condition. Rayner et al. have demonstrated the utility of remotely sensed concentration data . As the ground observation stations are limited and not equally distributed globally, the satellite observation drastically improves the observation points and data quality.
Chédin et al. and Engelen et al. retrieved the long term increase from conventional IR channel data of the Television and Infrared Operational Satellite-Next Generation (TIROS-N) Operational Vertical Sounder (TOVS) on the National Oceanic and Atmospheric Administration (NOAA) polar meteorological satellites and the Atmospheric Infrared Sounder (AIRS) on the NASA Aqua platform, respectively [2, 3, 4]. The Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY), which was launched in 2002 on board the Environmental Satellite (ENVISAT), is the first instrument to retrieve the and column density from space. Buchwitz et al. and Barkley et al. have presented the seasonal variation of the column density using the short wave infrared (SWIR) data of SCIAMACHY [5, 6]. However, there are no instruments specially designed for GHG measurements from space. The Greenhouse Gases Observing Satellite (GOSAT) is a satellite to monitor and globally from space with high spectral resolution. The Japanese name of GOSAT is “Ibuki,” which means breath in Japanese. The two instruments selected as the onboard instruments on GOSAT are shown in Figs. 1, 2 : the Thermal and Near Infrared Sensor for a Carbon Observation Fourier- Transform Spectrometer (TANSO-FTS) and the TANSO Cloud and Aerosol Imager (TANSO-CAI). TANSO means carbon in Japanese. The Orbital Carbon Observatory (OCO) by NASA/JPL (Jet Propulsion Laboratory), which is also specially designed for column density observation, failed during launch and will not be providing measurements from space [7, 8].
1B. GOSAT Tasks
GOSAT is a joint project of the Japan Aerospace Exploration Agency (JAXA), the Ministry of the Environment (MOE), and the National Institute for Environmental Studies (NIES). JAXA is responsible for the satellite development, launch, and operation. JAXA and MOE share the sensor development. MOE and NIES are responsible for satellite data utilization. GOSAT was launched 23 January 2009 on board the H-IIA rocket from the Tanegashima Space Center. The objectives of the GOSAT mission are to contribute to the environmental administration by monitoring the global distribution of GHGs, estimating the sources and sinks on a subcontinental scale, and verifying the reduction of GHGs emission, which is required by the Kyoto Protocol to advance Earth observation technologies for future missions. The targets of the mission are observation of density in 3 month averages with 0.3%–1% () relative accuracy in spatial resolution during the first commitment period (2008 to 2012) of the Kyoto Protocol and reduction of un certainties by half in identifying the GHG sources and sinks on a subcontinental scale with the data obtained by GOSAT in conjunction with the data gathered by ground instruments. In addition, the mission aims to observe on the same spatial and temporal scales as and with accuracy higher than 2%.
1C. Instrument Development
Two breadboard models of TANSO-FTS were developed: TOKYO (the laboratory model) and TSUKUBA (the airplane model) . The TOKYO model is developed for demonstrating the FTS short wave application such as the modulation efficiency and spectral resolution and for selecting the optimized parameters such as the scan speed and the optical throughput. With the TOKYO model from Mt. Tsukuba, Ibaraki, Japan, high resolution solar-scattered spectra from the ground with and absorption were acquired. Yoshida et al. has demonstrated retrieval accuracy of better than 2% . Based on the TOKYO results, the airborne model TSUKUBA was manufactured. In March and April 2007, TSUKUBA was flown over the cirrus clouds on a high altitude airplane, in Adelaide, Australia. Two models of TANSO were manufactured: the engineering model (EM) for checking the performance on the ground and the protoflight model (PFM) for space. Based on the performance and environment test results using EM, PFM design has been modified. Finally, both TANSO-FTS and CAI PFMs were integrated to the GOSAT satellite in March 2008. In this paper, a brief overview of the GOSAT project, scientific requirements, instrument design, hardware performance, on-orbit operation, and data processing is provided.
2. Scientific Requirement and Data Retrieval
2A. Observation Method and Orbit Selection
In this section, the scientific requirements for the instrument design and the data retrieval plan are discussed. In order to monitor GHGs, it is essential to measure the troposphere where the greenhouse effect occurs. In general, it is more difficult to measure the troposphere from space than the stratosphere because scattering by cloud-and-aerosol and the height of the Earth surface have to be considered. A limb viewing method such as limb emission and solar occultation is not capable of lower troposphere measurements. A down-looking measurement is the only feasible way to measure GHGs by detecting the flux that passes through the lower troposphere. Therefore GOSAT accommodates the down-looking instruments TANSO-FTS and CAI to measure GHGs and to detect cloud and aerosol.
A sun-synchronous orbit is selected to improve GHGs retrieval accuracy using the same observation geometry after the revisit. Table 1 shows the satellite orbit specification. The revisit interval is selected to be 3 days in order to detect weekly variation, which is the typical time scale of global distribution changes. Therefore the GOSAT satellite is introduced on a orbit. Olsen et al. have shown diurnal variation of the vertical profile. Local solar time of GOSAT is selected to be 13:00 when density is not affected much by a boundary layer condition . Eguchi and Yokota and Miller et al. have presented the clear sky occurrence rate of afternoon orbits [12, 13]. Their study shows clear sky regions are approximately 5%. In addition, a smaller solar zenith angle is preferable to measure brighter targets.
2B. Radiative Transfer and Column Density Retrieval
Kuang et al. have presented space borne measurements of atmospheric by high-resolution NIR spectrometry of reflected sunlight with three spectral bands . FTS measures the scene radiance of solar short wave infrared spectra (SWIR) reflected on the Earth’s surface that pass through troposphere. We have added measurement of thermal infrared spectra (TIR) radiated from the ground and the atmosphere to retrieve a vertical profile by acquiring the absorption lines of various cross sections. The lower tropospheric amount is estimated by extracting the upper atmospheric amount retrieved from the TIR band. Figure 3 shows the radiative transfer of SWIR and TIR in the Earth’s atmosphere and the flux that are observed from space. As indicated in the figure, the scene flux includes both surface reflection and single and multiple scattering of clouds and aerosol in SWIR. Thus an accurate optical path has to be estimated by properly combining the simulation of the radiative transfer and cloud-and-aerosol observation. For TIR, temperature vertical profile information is needed for the vertical profile retrieval. Figure 4 illustrates the spectral coverage and absorption lines of the TANSO-FTS SWIR and TIR observation. From these spectral data, major GHGs such as , , and ozone () are observed. The column density of is mainly retrieved from the region absorption lines, of which intensities are less temperature dependent and not interfered with by other molecules. Especially, the lines such as , , , and are level temperature dependent, which is much lower than other spectral regions in thermal infrared. By using these temperature insensitive lines, accurate retrieval can be obtained by minimizing temperature estimation uncertainty. In addition, can be retrieved in the region. The region can be used as the secondary retrieval band. The oxygen () A band absorption at is used to estimate the effective optical path length, which is a key parameter for the column density retrieval. The effective optical path length is a function of air-mass factor, cloud fraction, height, optical thickness, and pressure. As concentration is constant, well known, and uniformly distributed throughout the atmosphere, it can be used as a reference. Air-mass factor can be calculated from the observation geometry information of the solar zenith angle and viewing angle (pointing angles). From the estimated optical path length using the A-band data, the presence and characteristics of cloud-and-aerosol can be retrieved [15, 16]. For TIR, a model-estimated temperature profile is applied for retrieval.
2C. Performance Requirements and Spectrometer Selection
NIES has studied the sensitivity of and retrieval from space [17, 18, 19, 20]. Their studies show that for SWIR bands, a signal-to-noise ratio (SNR) higher than 300 at 13050, 6200, and with a spectral sampling interval of and a spectral resolution of meets the observation requirement. SNR is defined under the diffusive reflectivity of 30% and the solar zenith angle (SZA) of conditions. The spectral resolution is defined as full width at half-maximum (FWHM) of the measured instrument line shape function (ILSF). As GHGs are retrieved by differential absorption from SWIR spectra, both SNR and spectral resolution are the two key performance items that have to be achieved. Furthermore, the surface albedo and aerosol retrieval requires an accuracy of absolute radiometric calibration better than 5%. GHG column densities are retrieved from SWIR bands, but regarding TIR, the target SNR is defined as 300 or more at with blackbody light input for , , and retrieval. Saito et al. presented a study on vertical profile retrieval . Wave number ranges of bands 1, 2, 3, and 4 are specified as 12900–13200, 5800–6400, 4800–5200, and in order to cover absorption lines of , , , and windows for surface albedo retrieval. A flat spectral response better than within each band is required for bands 1, 2, and 3.
A single instrument that covers a wide spectral range is preferable, considering limited resources for space. GOSAT is a medium size satellite that accommodates one main instrument and one small radiometer. Compared with grating and etalon spectrometers, FTS has the following advantages. The throughput advantage is for high SNR and consequently precise and column density observation from space. The multiplex advantage of FTS enables wide spectral range observation with several detectors. However, major technical challenges exist: (1) FTS application for shorter wavelength, and (2) FTS operation under on-orbit thermal and micro vibration environments. This paper describes how to solve these items for the space application.
2D. Detection and Correction of Cloud and Aerosol
Dufour and Bréon and Bösch et al. presented error analysis for retrieval from space [22, 23]. Their studies show that cloud and aerosol are major uncertainties. There are several aerosol parameters that affect radiative transfer: thickness, type, size, and height. These parameters have to be measured or modeled as the inputs of the radiative transfer model. The error caused by aerosol and cloud has to be much smaller than and observation accuracy. So cloud and aerosol that cause allocated error budget must be detected and screened.
From the TANSO-CAI data, with higher spatial resolution than TANSO-FTS, the cloud fraction in the field of view of FTS should be well estimated. From the A-band, with several lines that have different line strengths, at least two geophysical parameters can be estimated, such as the scattering albedo and the cloud top height. As the measured absorption data depend on pressure, optical thickness, scattering albedo, and height, estimated parameters have uncertainties. Despite these uncertainties, by retrieving these parameters, the estimation accuracy in the effective optical path length is much improved.
In addition, aerosol scattering albedo has spectral characteristics, which have to be measured. At least four spectral regions have to be observed to characterize the aerosol property, such as optical thickness and type. Center wavelength and bandwidth of TANSO-CAI are carefully selected to avoid water vapor () and absorption. The selected bands are (band 1), (band 2), (band 3), and (band 4). Band 1 in the ultraviolet region is for aerosol retrieval over the land where the surface albedo is small. Bands 2 and 3 are selected shorter and longer than the red edge, where reflectance of vegetation changes rapidly. Band 4 is the same spectral region of the TANSO-FTS main band at . Thus the spatial coverage of TANSO-CAI wider than TANSO-FTS helps the retrieval of the type of the aerosol, which has a relatively large scale distribution.
TANSO-CAI must provide the information on both cloud and aerosol properties. For the former purpose, the wide dynamic range is required for large cloud scattering, and the latter requires high SNR for small aerosol scattering. To meet these conflicting requirements, the integration time of the detector has to be fine set to meet at least 200 SNR for all bands. Diffusive reflectances for the SNR definition are 0.152 for band 1, 0.106 for band 2, 0.112 for band 3, and 0.101 for band 4. The maximum ranges that can be observed have to be tuned between the diffusive reflectances of 0.5–0.693, 0.5–1.065, 0.5–1.123, and 0.5–1.014 for bands 1, 2, 3, and 4, respectively, by selecting integration time.
Equation (1) shows simplified radiative transfer in the Earth’s atmosphere of TANSO-FTS observation:24]. By inputting these parameters in Eq. (1), the column density of molecule i can be well retrieved.
Using multiple spectral bands, spectral characteristics of aerosol scattering can be retrieved together with optical thickness. In addition, band 1 () data provide aerosol information over the land. A study by NIES shows that by using TANSO-FTS spectra, TANSO-CAI data, and the retrieval algorithm to remove cloud and aerosol contamination, the and column densities can be retrieved with 1% and 2% accuracy, respectively.
2E. Polarization Measurements
SWIR path radiance is highly polarized, and the surface reflected light is less polarized. On the other hand, the instrument itself has polarization sensitivity. Optical components with polarization sensitivity are a ZnSe beam splitter of the FTS-mechanism and a pointing mirror. High SNR and a wide spectral coverage are given priority over low polarization sensitivity of the instrument. Instead, the polarization of the instrument must be well characterized, and the polarization of the scene flux should be acquired by measuring two linear polarizations [primary (P) and secondary (S)], simultaneously. Thus the Stokes parameters of the scene flux can be compared from P and S measured data of TANSO-FTS. The polarization directions are defined as the direction of the polarization beam splitter. Linearly polarized intensities are acquired in two perpendicular directions. In order to achieve high SNR for both P and S bands, polarization sensitivity of the instrument has to be lower than 0.34, 0.38, and 0.38 for bands 1, 2, and 3, respectively. The sensitivity is defined as the difference of the output maxima and minima divided by the sum of the maxima and minima of linear polarization light input. Similar two linear polarizations have been measured with SCIAMACHY in visible regions and applications of polarization measurements to characterize cloud, and aerosol properties are discussed [16, 25, 26].
The polarization sensitivity of TANSO-FTS before the launch can be well characterized. However, contamination on the optics might change the polarization sensitivity on orbit. P and S responses are calibrated using the solar diffused light on orbit, of which the degree of linear polarization is very small. A detailed polarization model was characterized before the launch and defined as the combination of the Mueller matrixes as presented in Subsection 5B. Therefore linear polarization measured using TANSO-FTS can be accurately compared with I, Q, and U components of the Stokes vector of the outgoing light from the top of the atmosphere simulated with a vector radiative transfer model. We assume the V component is zero. In addition, the simultaneous measurements of the two linear polarizations become redundant if GHGs are retrieved independently at the expense of polarization information.
3. TANSO Instrument
Specification of TANSO-FTS is summarized in Table 2. There are three units in TANSO-FTS: an optics unit, a control unit, and an electronics unit, as shown in Fig. 5. The optics unit consists of a pointing mechanism, a monitor camera, relay optics, a FTS mechanism, detectors, and analog-signal processors. The optical configuration including the band separation and detector optics is illustrated in Fig. 6. All the optical components are mounted on the optical bench, of which temperature is controlled at and kept between during the entire 5 year mission. Thermal radiation from a blackbody is at and small enough, whereas typical input radiance from the scene is . The optical bench is mounted on the satellite structure with three kinematic mounts to isolate the thermal distortion. Optical components are both radiatively and conductively isolated from their environments with thermal blankets and thermal isolators to minimize the short-term temperature variation. Mounted on the optics unit structure are a cooler for a TIR detector, a blackbody, a diode laser for ILS calibration, a diffuser plate and its driving mechanism, a hood to minimize the stray lights, and analog electronics. The control unit consists of a FTS controller, a pointing mechanism controller, a detector temperature controller, and a cooler controller. The electronics unit has functions of heater controlling, data processing, and a data-telemetry/command-transmission interface with the satellite bus system. All the components of the electronics unit have redundancy.
3A1. Pointing System and Monitor Camera
A two axis pointing mechanism has pointing and image motion compensation functions for the cross track (CT) and along track (AT) directions. By mainly rotating the CT stepper motor, the pointing mechanism views the Earth’s surface, deep space, the blackbody, and the diffuser. To point any target on the Earth’s surface within a 3 day revisit cycle, a scan angle covers from the nadir. An AT angular motor is mounted on the CT mechanism and has a limited traveling range: physical motion and line of sight. It is controlled with an angular rate sensor (ARS) with accuracy better than . These two axis motors and satellite-yaw-steering motion to compensate the Earth-rotation enable zigzag raster motion for an observation grid (Fig. 7). Pointing patterns are registered on the onboard tables, which have up to 3000 pairs of CT and AT angles. The motion of the pointing mechanism is synchronized with the FTS mechanism. Both the AT and CT motors are actuated and settled down within rotary arm turnaround of the FTS mechanism, which is between 0.4 and . Also, by mainly rotating the AT motor, TANSO-FTS can view the same field to compensate the satellite motion during one way interferogram acquisition time between turnarounds, which is typically . The secondary mechanism for redundancy is also mounted on the optical bench and has the same function as the primary one (see Fig. 6). As during the acquisition of the interferogram, jitter of the pointing mechanism may cause a ghost signal larger than noise in observed spectra; the pointing is designed to settle down within turnaround time. At the end of the turnaround when the interferogram acquisition starts, the resolver data with a format of AT and CT angle are acquired, digitally converted, and inserted in the header of the measured interferogram together with a time stamp. Accuracy of TANSO time is .
A small two dimensional CMOS camera is installed to register the FTS instantaneous field of view (IFOV). Careful alignment and rigid mounting of the monitor camera on the optical bench before launch stabilize registration between the centers of FTS IFOV and the monitor camera before and after the launch, which will be kept over the GOSAT mission. By comparing on-orbit TANSO-CAI and monitoring camera images, the registration between CAI and FTS can be estimated. The registration has to be estimated with an accuracy of . This estimated registration can be validated by viewing the ground control point (GCP). The monitor camera has a wider field of view and higher spatial resolution than the FTS single pixel. Its specification is summarized in Table 3. The size of the image and acquisition interval can be selected by the command from the ground. By setting the same sampling interval as the interferogram acquisition, such as every , a scene image of each FTS interferogram can be acquired. The camera data are transmitted to the mission data processor (MDP).
When inserting the fold mirror on the optical path as shown in Fig. 6, the scene flux from the redundant system can be introduced. All the mirrors on the pointing mechanisms and fold mirrors are similarly silver coated, which has high reflectivity over the SWIR region higher than 97.5% at and 99% at 1.6 and for both linear polarizations and incident angles of (measured value).
3A2. Fourier Transform Spectrometer Mechanism
The FTS mechanism is a double pendulum type interferometer with two cube-corner reflectors, which creates a optical path difference (OPD) four times as long as the mechanical motion. Both sides of the optical path difference () are acquired to make real and imaginary spectra for proper phase correction. The two cube-corner reflectors are mounted at the edges of a rotary arm (Fig. 6). They are carefully aligned with each other to maximize modulation efficiency with the minimum optical shear permanently. The arm is mounted on the beam splitter holder with a flexible blade and is actuated with a rotary motor. The two cube-corner reflectors and the motor are statically balanced in all three axes and are robust against vibration. The rotary arm mechanism moves smoothly with better than 1% speed stability, creating uniform frequency of the output modulated signal. In order to obtain high SNR, sampling electrical bandwidth has to be small enough; in other words, the scan speed has to be slow enough. At least is needed to achieve the required SNR, and the nominal time for one way interferogram acquisition is set at . The FTS mechanism also has faster scan modes for denser grid observation at the expense of SNR, and . TANSO-FTS has a fully redundant sampling system with two distributed-feedback (DFB) lasers and two InGaAs detectors. Each diode laser has much longer life time than a conventionally used He–Ne laser, and its size is small enough to have redundancy. A lifetime test over the equivalent of 10 years has been completed. As the diode laser wavelength has temperature dependency, its temperature is well controlled with stability of with a thermoelectric controller. So the wavelength stability () is better than , which is equivalent to typical He–Ne laser stability.
The cube-corner reflectors are the most critical optical component. The available clear aperture is and the pupil of the TANSO-FTS optics is centered on the cube-corner reflector. The structure of the FTS mechanics is made of aluminum, and the two cube-corner reflectors are made from three gold coated Zerodur plates on an invar structure for minimizing the thermal expansion. Surfaces of the cube-corner reflectors are better than (RMS) at . A maximum beam divergence smaller than shows excellent orthogonality. Back sides of the cube-corner reflectors are gold coated in order to minimize thermal radiation coupling with the environment. Thickness of the beam splitter is selected to be larger than the maximum OPD to avoid channeling. Its surface has high quality (better than at ) and no AR coating to maintain high optical efficiency over a wide spectral range. Bare ZnSe material has a spectrally flat index of refraction and transmittance higher than 65% over the wide spectral region from 0.76 to .
Modulation efficiency is a key parameter to characterize the FTS performance, especially at SWIR. In general, modulation at shorter wavelengths is lower and FTS application becomes more difficult. With careful manufacturing and screening of the optical components, the modulation efficiencies of at , at , and at were obtained (measured value with PFM).
3A3. After-Optics and Detector
Modulated light from the FTS mechanism is then collected into a circular stop. Geometric registrations between seven spectral bands of TANSO-FTS are well aligned by sharing this common field stop. It defines IFOV of , which is equal to the footprint for nadir pointing. To maximize the optical throughput, IFOV corresponds to maximum divergence allowed at the most important band 2 for spectral resolution (interval). The scene flux is collimated and divided into four spectral bands with dichroic filters. The three dichroic filters reflect the shortest wavelength band light and transmit the longer wavelength bands light. These filters are carefully designed to minimize the polarization sensitivity of SWIR bands and to minimize background radiation of TIR. Light in each SWIR band passes through a narrow bandpass filter and is divided into two detectors by a polarization beam splitter. The narrow bandpass filters are wedged to avoid channeling and reject shorter wavelength light to avoid aliasing and minimize out-of-the-band stray light.
Two Si detectors are applied for bands 1P and 1S. The four InGaAs detectors for bands 2P, 2S and 3P, 3S are nonbiased and cooled at with three-stage thermoelectric coolers to minimize dark currents. Packages of the InGaAs detectors are custom designed for TANSO-FTS by Hamamatsu Photonics K.K., and low shunt-capacitance detectors are carefully screened for high frequency operation of the FTS application. The TIR light is collected on a single photoconductive (PC) HgCdTe (MCT) detector, which is mounted on the dewar and cooled at with a pulse tube cooler. The pulse tube cooler has a very low vibration level with no effect on the interferogram acquisition, and it also has a variable motor frequency to minimize unexpected vibration interference on orbit.
3A4. Signal Processing and Data Transmission
Each detector signal current is converted to voltage and amplified with operational amplifiers. To minimize high frequency noise, multistage amplifiers with low pass filters are integrated. Especially noise at the second and third orders of the signal frequency is carefully reduced. Then analog signals are converted in parallel to digital data with one AD converter per detector. Therefore seven interferograms are acquired simultaneously. For bands 1, 2, and 3, up-and-down zero crossings of laser fringes are utilized for sampling triggers, which result in 76,336 sampling points per interferogram with the half laser wavelength () interval. Band 4 samples are 38,168 points per interferogram with λ interval. For band 1, to avoid aliasing of noise, only a modulated portion (AC) is acquired. For bands 2 and 3, AC plus offset data (DC) are acquired. For band 4, both modulated portion (AC) and nonmodulated portion (DC) are acquired and amplified separately. The DC portion is used to correct the nonlinearity effect of the PC-MCT detector. To minimize electric and magnetic interference, the analog amplifiers and AD converters are mounted close to the detectors in the optics unit. Digital signals are processed in the electronics unit. In addition, telemetry data used for level 1 and 2 processing are added to the data header. The telemetry data include the satellite position, GPS time, amplifier gain levels, AT and CT angles of the pointing mechanism, and temperature of the blackbody and the detectors. Finally, acquired data are transmitted to MDP of the GOSAT satellite bus system in Consultative Committee for Space Data Systems (CCSDS) format.
3A5. Onboard Calibration Sources
A space grade Spectralon diffuser was manufactured from well baked Teflon powder. It was removed from a clean storage and mounted on the optical unit structure at the Tanegashima launch site just before the launch. The diffuser plate is mounted on the mechanism, which can be rotated by . Both sides of the diffuser plate are used for degradation comparison with different exposure time. It views the sun directly near the North Pole during sunrise each orbit. During the day side of the orbit, a shade prevents the direct solar illumination from entering the inner optics. Diffuser reflectivity provided by the manufacturer is better than 0.99, 0.98, and 0.96 for bands 1, 2, and 3, respectively, for both sides. The bidirectional reflection distribution function (BRDF) and diffusive reflectivity of both sides were characterized after the launch during the initial checkout phase (3 months after the launch) by maneuvering the satellite.
ILSF symmetry represents spectral resolution and alignment of the optics. A monochromatic diode laser is mounted on the outer structure of the optics unit and illuminates the diffuser plate. It provides ILSF calibration only for bands 2P and 2S. As all TANSO-FTS bands share the common circular stop and the aft optics are rigidly mounted, the on-orbit ILSF of bands 1, 3, and 4 can be estimated.
One blackbody for the TIR radiometric calibration and a deep space view window are placed on the side wall of the optics unit in such way that both primary and secondary pointing mechanics view the same surface and window. The blackbody surface has emissivity higher than 0.95 with small black-anodized pyramid like shapes. Its temperature is kept as the surroundings () without active temperature control to improve the effective emissivity and monitored with platinum temperature sensors. The TANSO-FTS-TIR band is calibrated nominally 4 times per orbit by viewing deep space and the blackbody.
3A6. Contamination Control
GHGs are retrieved by differential absorption. Outgassing from the structure, glues, tapes, wires, and wrapping materials may overlap GHGs absorption lines and degrade the GHGs retrieval accuracy. Urabe et al. has measured absorption spectra of the materials that are used for TANSO and GOSAT satellite buses [27, 28]. From these test results, some adhesives and wires to fold cables have been replaced by less contaminant materials.
3A7. Micro Vibration Effect Control
Sinusoidal micro vibration on the FTS mechanism during interferogram acquisition can create sidelobes (ghost signal) on the inverse-Fourier transformed spectra . Since GHGs are retrieved from differential-absorption spectra, these sidelobes might significantly degrade accuracy and must be minimized. Three possible phenomena affect the TANSO-FTS performance under micro vibration conditions: (1) One is amplitude modulation due to shear between two beams. Input vibration from the satellite structure has to be minimized. Jitters from a reaction wheel, gyro, Earth sensor, and paddle drives of the satellite solar panels are carefully damped, especially near the resonant frequency of the FTS mechanism. (2) Another is frequency modulation due to sampling jitter. Delay mismatch between the sampling laser fringe and scene flux fringe is caused by electrical phase delay of the amplifiers and has to be minimized. Frequency modulation is the dominant sidelobe source. By applying a delay matching circuit between the laser fringe and the sampling trigger for the AD converter, the level of the sidelobe becomes much lower than the noise level. (3) Last is intensity modulation due to a scene oscillation of the pointing mechanism. AT and CT motors have to be well controlled, and input vibration from the satellite has to be isolated. Pointing stability is better than against IFOV of .
TANSO-CAI is a radiometer to retrieve clouds and aerosol characteristics for more accurate TANSO-FTS data retrieval. Table 4 shows the TANSO-CAI specification. The center wavelength is determined from a moment’s analysis. TANSO-CAI has a continuous spatial coverage, wider field of view, and higher spatial resolution than TANSO-FTS in order to detect aerosol spatial distribution and cloud coverage. As in general aerosol is widely distributed and much larger than the TANSO-FTS IFOV, a wide swath is required. The swath is in such a way that there is no gap between 44 orbits in 3 days. TANSO-CAI has two units, an optics unit and an electronics unit. The optics unit has three telescopes and analog signal processors. The electronics unit has a function of data processing and a telemetry-and-command interface with the satellite. All the components of the electronics unit have redundancy.
3B1. Optical Design
TANSO-CAI has three telescopes: one for band 1, one for bands 2 and 3, and one for band 4. The spectral responses of the four spectral bands including the optics efficiency and the detector response are illustrated in Fig. 8. Band 4 blocks a cutoff wavelength region of the InGaAs detector, which has large temperature dependency. Bands 1, 2, and 3 have a wide field of view (FOV) and the swaths are wider than orbit path spacing on the equator (). Band 4 uses a 512 pixel detector and is limited to a wide FOV, which is a swath on the ground.
3B2. Signal Processing
Bands 1, 2, and 3 use one-dimensional Si CCD array detectors with 2048 pixels without temperature control. Estimated temperature variation on orbit is smaller than . Band 4 use a one-dimensional InGaAs array detector with 512 pixels, and its temperature is controlled at with a one-stage thermoelectric cooler to minimize dark current. Each pixel of the InGaAs detector array has an individual amplifier. A dark current level can be monitored both prelaunch and on orbit by acquiring data at the night side. For band 4, two feedback capacitances can be selected by the software command on orbit. Signals of each band are multiplexed and analog-to-digital converted. Then the telemetry data such as GPS time, satellite position, and integration time are added and transmitted to MDP. For all TANSO-CAI bands, longer integration time improves SNR but reduces its dynamic range. Therefore 32 levels of integration time with 19% time increment have to be carefully selected by commands to optimize SNR and its dynamic range depending on season and latitude.
4. Observation Strategy
4A. Operational Mode, Pointing, and Sampling
TANSO-FTS has the following operational modes: nominal observation, sun-glint observation, special target, calibration, and standby. Figure 9 shows the concept of TANSO-FTS and CAI operations. During days both SWIR and TIR of TANSO-FTS and TANSO-CAI data are acquired, and at night only TANSO-FTS TIR data are acquired. At sunrise, direct sunlight is introduced into TANSO-FTS through the Spectralon calibration diffuser plate for solar irradiance data acquisition and SWIR radiance calibration. In addition, diode laser light illuminates the same diffuser plate for the ILSF onboard calibration. The pointing mechanism views deep space and the inner blackbody periodically for a zero level and TIR radiance calibration. TANSO-FTS gain and TANSO-CAI integration time are selected to optimize SNR and dynamic range. The levels can be set orbit by orbit and as a function of the latitude.
As shown in Fig. 7, TANSO FTS scans a zigzag raster pattern. As the orbit period changes very slowly in nominal observation, by receiving GPS time of the crossing ascending node from the satellite in every orbit, the TANSO onboard processor calculates the most recent orbit period and allocates a sampling interval for the following orbit observation. The exact observation time is tuned on orbit by adjusting the turnaround time, which includes variable additional scans of the rotary arm beyond OPD and stopping and restarting of the rotary arm motion. TANSO-FTS has a 3-day revisit time. By measuring the same observation points repeatedly, uncertainties of the surface-albedo estimation are reduced. The spatial interval of the grid observation can be selected from to . Nominally, interferograms are acquired every for five points in the CT direction, which results in spacing between grid observations and a total of 56,000 points globally per 3-day period. For high SNR observation, the pointing mechanism can stare at the same FOV for three interferograms acquisition. Each interferogram is Fourier transformed and then three spectra can be coadded. For more spatially dense observation, seven or nine points may be acquired for each cross track scan by reducing observation time per FOV.
Over the ocean where surface reflectance is small, the pointing mechanism views a widely spread sun-glint area, where specular reflection occurs and reflectance is high, without image motion compensation. As the AT viewing angle range is , sun-glint observation is limited to low and middle latitudes. Furthermore, occasionally by uploading the observation start time and set of pointing angles from the ground, TANSO-FTS can view special targets between AT angle of and CT angle of for validation, vicarious calibration, emissions from mega cities, and leakage detection from pipelines.
4B. Temperature Control
The temperature of both TANSO-FTS and CAI optics is controlled at during the 5 year mission. As the modulation efficiency of the FTS mechanism is temperature dependent, stability within a orbit period, which is the solar irradiance calibration interval, is important for a consistent response within an orbit period. Expected temperature variation within based on the prelaunch thermal- vacuum test is about , and the response instability is negligibly small.
5. Calibration and Characterization
5A. Prelaunch Calibration
Before the launch, TANSO-FTS and CAI radiomet ric calibrations were performed using integrating spheres as illustrated in Fig. 10. There are two integrating spheres: One is a diameter gold-coated sphere that has spectrally flat high reflectivity over TANSO-FTS bands 2 () and 3 () regions. The other is a diameter -coated sphere that has excellent diffusive reflection and therefore good spatial uniformity of the radiance. For TANSO-FTS band 1 and CAI calibration, the sphere is utilized because uniformity and diffuse reflectivity are better than those of the gold sphere. For TANSO-FTS bands 2 and 3, the diameter gold-coated sphere is used because reflectivity and spectral flatness of the gold surface are much better than the diameter -coated sphere at the SWIR region. Prelaunch calibrations were completed before TANSO-FTS and CAI were integrated on the GOSAT satellite on March 2008. These integrating spheres have the output spectral radiance equivalent with the Earth’s surface albedo from 0% to 300% for TANSO-FTS bands 1, 2, and 3 and CAI bands 2, 3, and 4 and from 0% to 15% for CAI band 1. Long term absolute stability of the integrating sphere is better than 2%. The absolute spectral radiances of the integrating spheres are calibrated by comparing spectral radiance of the fixed-point blackbody furnace with a double grating monochromator equipped with two input ports. One of the input ports is for an integrating sphere and the other is for fixed-point blackbodies. A copper (Cu) fixed-point blackbody furnace at for TANSO FTS band 1 and zinc (Zn) at for bands 2 and 3 are used, respectively. Last, output radiance and spatial uniformity within the aperture of the integrating spheres are measured with the three GOSAT transfer standard (TS) radiometers [30, 31]. Estimated accuracy of the TS radiometers is . Therefore, with the three GOSAT TS radiometers, GOSAT prelaunch calibration can be certificated with an accuracy of , and also three TS radiometers were utilized for the cross- calibration campaigns with OCO. TS outputs are traceable to the fixed-point blackbody furnace maintained at National Metrology Institute of Japan/ Advanced Industrial Science and Technology (NMIJ/AIST).
5B. Performance and Environment Test on the Ground
The tests are given below. SNR: SNR tests were performed with the same configuration as radiometric calibration experiments. Both TANSO-FTS and CAI detect different radiance values for given illumination levels of the integrating sphere. SNR is calculated from the standard deviation and the average of 32 continuous spectra measurements. The measured SNR values of TANSO-FTS are 345, 246, 322, 257, 412, 287, and 283 for bands 1P, 1S, 2P, 2S, 3P, 3S, and 4, respectively, with the input radiances of equivalent surface albedo of 30% and a solar zenith angle of for SWIR and for TIR. Defined input spectral radiances are , , , and (94, 20, 9.5, and ) at for band 1, for band 2, for band 3, and for band 4, respectively. All the TANSO-CAI bands have SNR better than 200 by selecting proper integration times such as 55, 19, 46, and for bands 1, 2, 3, and 4, respectively. The defined input spectral radiances are 47, 45, 29, and for bands 1, 2, 3, and 4, respectively.
Spectral resolution and ILSF of TANSO-FTS: Monochromatic light of 13076, 6283, 5012, 1174, and from tunable diode lasers is introduced into TANSO-FTS through a laboratory diffuser plate to illuminate the entire IFOV. The laser power is larger than , and bandwidth is smaller than (), which is higher than FTS spectral resolution. Figure 11 shows the measured ILSF of the PFM band 2P together with model simulation, which considers the IFOV size and offset due to as-built alignment uncertainty. The model simulation assumes uniform modulation efficiency over the IFOV. The measured FWHM values of bands 1P, 1S, 2P, 2S, 3P, and 3S are 0.367, 0.356, 0.258, 0.257, 0.262, and , respectively. These results show good agreement in width and symmetry with the theoretical models and meet the scientific requirements.
Modulation transfer function (MTF) of TANSO-CAI: MTF represents the spatial resolution and is measured by introducing collimated light with black-and-white patterns by the Nyquist frequency stripe slits. Measured values are 0.62, 0.30, 0.19, and 0.26 for bands 1, 2, 3, and 4, in the worst case, respectively, which are higher than 0.2 (design specification) for bands 1, 2, and 4 and higher than 0.15 (design specification) for band 3.
Polarization sensitivity: The polarization sensitivities of both TANSO-FTS and CAI are characterized from the instrument outputs against linear polarized light input with a polarizer and a rotating phase plate. TANSO-FTS has polarization sensitivity smaller than 0.34 for band 1 and 0.38 for bands 2 and 3 (design specification), and CAI has a value smaller than 0.03 (design specification). The measured vales are 0.27, 0.28, and 0.30 for TANSO-FTS bands 1, 2, and 3 and 0.026, 0.027, 0.030, and 0.019 for CAI bands 1, 2, 3, and 4 in the worst case, respectively.
Polarization extinction ratio and polarization model: Extinction ratio is defined as the ratio of the output against the perpendicular linear polarization light to the one against parallel linear polarization light. The measured value is smaller than 6% (in the worst case) over the cross-track pointing range, although the design goal is smaller than 3%. This value represents a phase difference between the P and S light. The phase difference exists both in the input light source in the laboratory and the pointing mirror reflection. The phase at the pointing mirror coating provides a function of the incident angle and wave number. Based on these prelaunch data, the detailed polarization of the TANSO-FTS optics is modeled by Stokes vectors and Mueller matrixes as expressed in Eqs. (2, 3, 4, 5, 6, 7, 8):
FTS stray light measurement: Stray light in the modulated light introduced from an improper optical path increases error of the GHG retrieval. Brighter scattered light from clouds outside the IFOV is a major stray light source. Stray light can be measured prelaunch using the saturated absorption lines in a long gas cell within IFOV and a brighter light source outside IFOV. A large Spectralon diffuser plate illuminated with a halogen lamp simulates bright cloud. It was moved step by step vertically and horizontally to cover the entire aperture of the TANSO-FTS hood and the line of sight of the onboard diffuser. Stray light could be detected as a ghost signal at the fully absorbed lines. The light level has to be zero at the core of the saturated absorption lines. An experiment tests whether light from outside IFOV adds any observable signal in the saturated line cores. Results indicated that no detectable stray light above noise level was observed.
CAI stray light measurement: Stray light from outside the FOV is tested by introducing bright light. Stray light inside the FOV can be detected by illuminating knife edge light from the collimator. No stray light above noise level was detected.
Environment tests: Sinusoidal vibration, acoustic, and thermal vacuum tests were performed. Test levels covered the estimated launch and on-orbit conditions.
Life time tests: With spare parts, the life time of the diode laser, the flex-blade of the FTS mechanism, and bearings with lubricant in the pointing mechanism have been tested for the equivalent of 10 years, which is twice as long as the GOSAT mission life.
Radiation hardness test on the detectors, AD converter, and optical filters: Si and InGaAs detectors were tested against proton, heavy ion, and gamma ray exposures expected on orbit for the GOSAT mission life. After the radiation exposure, it was confirmed that the dark-current increase, which might degrade the SNR performance over the entire mission, was small enough. Latch-up protection of the AD converter was also tested against heavy ions. The optical filters were tested against gamma rays. These radiation tests were performed with spare parts. No detectable degradation was observed.
UV and atomic oxygen radiation tests on the calibration diffuser: Both UV radiation and atomic oxygen exposure tests covering the total exposure time of the GOSAT mission life on-orbit were completed. Reflectivity of the UV-radiated calibration diffuser shows larger degradation at wavelengths shorter than TANSO-FTS band 1 (). Degradation is a function of contaminant amount on the diffuser surface and radiation time. Thus the flight diffuser plate must be kept in a purged container and mounted just before the launch. Atomic oxygen removes contaminants on the diffuser surface, which becomes brown with UV radiation, and no significant degradation was observed. These results show that a mixture of UV radiation and atomic oxygen on-orbit does not degrade diffuser performance significantly.
5C. Prelaunch Cross Calibration with Orbital Carbon Observatory
OCO by NASA and GOSAT by JAXA were planned to be launched and operated on-orbit in the same period observing common spectral bands for detection . Thus these sensors provide independent measurements of the global distribution from space. OCO is a grating spectrometer, and TANSO-FTS is a Fourier transform spectrometer. For the direct comparison of their data, it is desirable that both instruments are calibrated to common standards in their 0.76, 1.6, and channels. Cross comparison increases confidence in the results from both instruments.
By introducing unpolarized light from the integrating spheres, the response of both TANSO-FTS and OCO can be compared despite the separate polarization channels observed by GOSAT in TANSO-FTS bands 1, 2, and 3. The cross-calibration experiments were performed in April 2008 at JPL and December 2008 at JAXA . The OCO and GOSAT instrument absolute radiance calibrations agree to 3% for all bands during both experiments. The spectral radiance conversion table is created prelaunch with the calibrated integrating sphere. By applying this table to on-orbit data and viewing the same target from space with both instruments, consistency of radiometric calibration before and after the launch can be confirmed.
5D. Onboard Calibration
Onboard calibration items and frequency are summarized in Table 5.
Solar irradiance calibration: For calibration against the sunlight in the initial checkout phase of the onboard operation, yaw maneuvers are performed to obtain the initial diffuser BRDF. This involves rotating the satellite by an angle corresponding to fluctuations in the β angle (the angle made by the direction of the sun and the plane of the orbit) around the yaw axis. The purpose of this step is to determine whether an apparent change with time detected during calibration against sunlight is in fact a change with time or is actually due to a dependency on the β angle within the calibration system. After the initial checkout phase, both sides of the diffuser with different exposure time are utilized to correct the long term diffuser degradation. The front side of the calibration diffuser is used for the regular per- orbit calibrations. The back side is used once a month to provide a reference calibration of slower degradation. Thus both the degradation of the TANSO- FTS optics and the front side of the diffuser can be estimated.
Lunar calibration: The lunar surface has diffusive and temporally stable reflectance in SWIR . The Robotic Lunar Observatory (ROLO) spectral data from 0.35 to is available as the lunar surface reflectance. By pitch maneuvering, the FOV of the GOSAT instruments is pointed to the center of the full moon during night. At first, TANSO-FTS views the moon disk of width, which is smaller than the IFOV of , such that spatially averaged albedo can be measured in all SWIR bands as a secondary calibration standard. Then by slightly rotating the satellite, TANSO-CAI can vertically scan the full disk of the moon surface. This lunar calibration is scheduled during the initial checkout phase and then approximately once a year. The lunar calibration is the only onboard calibration method for TANSO-CAI.
ILSF calibration: ILSF symmetry is monitored once a month by introducing diffusive monochromatic light in order to confirm that the optical axis of the most important band 2 is well aligned and stable on-orbit. As the diode laser is mounted at the outer structure and it is not temperature controlled, wavelength calibration cannot be performed. The measured data at the launch site after replacing Spectralon diffuser with the space-grade one are shown in Fig. 12. ILS calibration laser wavelength is (), which is shorter than the specified band 2 region (), but the laser has enough power to be calibrated at the out-of-band low optical-efficiency region. As bands 1, 2, 3, and 4 share the common field stop and aft-optics are rigidly mounted, on-orbit ILS symmetry of all bands can be monitored.
5E. Vicarious Calibration
Several ground sites, where surface albedo is spatially uniform and aerosol optical thickness is small, were selected for the vicarious calibration. The ground surface albedo and its spectral charac teristics are measured on the ground with the spectrometer and Spectralon reference when GOSAT passes over the site. Other geophysical parameters such as BRDF, optical thickness, and vertical profile of temperature, pressure, and relative humidity are measured at the same time with a radiometer, a sun photometer, and a radiosonde. The spectral radiance at the top of the atmosphere is calculated by the radiative transfer model with input of the measured albedo and compared with the calibrated satellite radiance data. One of the candidate sites is Railroad Valley, Nevada, USA. .
6. Data Processing and Products
Table 6 shows overall data processing flow and the data products from TANSO-FTS. Acquired raw interferograms and telemetry data are transmitted to the ground 9 times per day as level 0 data. The downlink sites for GOSAT are Kongsberg Satellite Services (KSAT) in Svalbard, Norway, and Earth Observation Center (EOC) in Hatoyama, Japan. Then interferograms are inverse Fourier transformed to spectra. From spectra, effective optical path and GHG absorption are retrieved and then density is calculated: From SWIR data, the column densities of and data are acquired, and from TIR data, vertical profile are retrieved. Every three days, global GHG distributions are created, and, finally, by inversion with a chemical transfer model, sources and sinks of GHGs are retrieved. JAXA is responsible for providing TANSO-FTS level 1A (raw interferogram) and 1B (geolocated spectra) data and TANSO- CAI level 1A (raw output data). NIES is in charge of FTS levels 2 (column density), 3 (global distribution), and 4 (sources and sinks inversion), and levels 1B (interband registration corrected) and 2 (spectral radiance) of TANSO-CAI .
TANSO-FTS level 1 processing flow is presented in Fig. 13. Level 1A is raw interferogram data before calibration. Then the spike caused by cosmic rays is removed, and nonlinearity is corrected for band 4. The zero-path-difference (ZPD) position at center burst is searched as the peak output of the interferogram. After ZPD detection, the interferogram is inverse Fourier transformed by integrating with OPD. For SWIR, the phase retrieved from complex spectra is corrected for every interferogram. As the sampling laser wavelength of the FTS mechanism is temperature dependent and alignment of the laser optical path might be changed after the launch, wave number is roughly calibrated using the A-band ab sorption line in band 1P. This calibration factor is applied to the other bands. For SWIR, no apodization is applied, and raw inversed-Fourier-transformed spectra with phase correction are stored as level 1B in units of . Together with the radiometric calibration data, level 1B raw spectra can be converted to spectral radiance in units of . The most probable conversion table to spectral radiance is provided to users separately from level 1B data. When degradation is detected on orbit, this table will be updated based on the prelaunch radiometric data and onboard solar- irradiance calibration data.
For TIR, spectral radiances are directly calculated from complex calibration with blackbody and deep-space calibration spectra as follows:
For TANSO-CAI, together with level 1A raw output data, radiance conversion and dark level tables are provided. By multiplying this conversion values with level 1A data and dividing them by integration time in milliseconds, the spectral radiance in units of can be calculated. The radiance conversion table is based on the prelaunch radiometric calibration data and will be updated if the degradation is detected after onboard lunar and vicarious calibrations. A dark level is a function of integration time based on the prelaunch data and its table will be modified using periodically measured night data on-orbit.
In parallel, geolocation of TANSO-FTS and CAI can be calculated using GPS time stamp, satellite position data, and resolver outputs of the pointing mechanism in telemetries.
The design, calibration, preflight performance, and data processing plan of TANSO-FTS and CAI have been described. The throughput advantage of FTS demonstrated high SNR and spectral resolution observation for GHG retrieval from space. We have presented FTS application for near, short wave, and thermal infrared regions. TANSO-FTS has very high modulation efficiency over a wide spectral range. TANSO-CAI has a wide spatial coverage and moderate spatial resolution. Their function and performance have been well tested and characterized before the launch. These results provide useful information for calibrating onboard data.
This research was conducted by Japan Aerospace Exploration Agency. We thank Tatsuya Yokota, Yukio Yoshida, Hiroyuki Oguma, and Isamu Morino of NIES, Fumihiro Sakuma of NMIJ, and the GOSAT Science team led by Gen Inoue for their useful suggestions, as well as the members of Ministry of the Environment, NEC TOSHIBA Space Systems, Ltd., ABB BOMEM, Inc., Hamamatsu Photonics K.K., Fujinon Corp., and JAXA aerospace research and development directorate for their cooperation.
1. P. J. Rayner and D. M. O’Brien, “The utility of remotely sensed concentration data in surface source inversions,” Geophys. Res. Lett. 28, 175–178 (2001). [CrossRef]
2. A. Chédin, A. Hollingsworth, N. A. Scott, S. Serrar, C. Crevoisier, and R. Amante, “Annual and seasonal variations of atmospheric , and CO concentrations retrieved from NOAA/TOVS satellite observations,” Geophys. Res. Lett. 29, 110–114 (2002). [CrossRef]
3. A. Chédin, S. Serrar, N. A. Scott, C. Crevoisier, and R. Amante, “First global measurement of midtropospheric from NOAA polar satellites: tropical zone,” J. Geophys. Res. 108, 4581 (2003). [CrossRef]
4. R. J. Engelen, E. Andersson, F. Chevallier, A. Hollingsworth, M. Matricardi, A. P. Mc-Nally, J.-N. Thépaut, and P. D. Watts, “Estimating atmospheric from advanced infrared satellite radiances within an operational 4D-Var data assimilation system: methodology and first results,” J. Geophys. Res. 109, D19309 (2004). [CrossRef]
5. M. Buchwitz, M. R. de Beek, S. Noël, J. P. Burrows, H. Bovensmann, H. Bremer, P. Bergamaschi, S. Körner, and M. Heimann, “Carbon monoxide, methane and carbon dioxide columns retrieved from SCIAMACHY by WFM-DOAS: year 2003 initial data set,” Atmos. Chem. Phys. 5, 3313–3329 (2005). [CrossRef]
6. M. P. Barkley, U. Frieß, and P. S. Monks, “Measuring atmospheric from space using full spectral initiation (FSI) WFM-DOAS,” Atmos. Chem. Phys. 6, 3517–3534 (2006). [CrossRef]
7. D. Crisp, R. M. Atlas, F.-M. Breon, L. R. Brown, J. P. Burrows, P. Ciais, B. J. Connor, S. C. Doney, I. Y. Fung, D. J. Jacob, C. E. Miller, D. O’Brien, S. Pawson, J. T. Randerson, P. Rayner, R. J. Salawitch, S. P. Sander, B. Sen, G. L. Stephens, P. P. Tans, G. C. Toon, P. O. Wennberg, S. C. Wofsy, Y. L. Yung, Z. Kuang, B. Chudasama, G. Sprague, B. Weiss, R. Pollock, D. Kenyon, and S. Schroll, “The Orbiting Carbon Observatory (OCO) Mission,” Adv. Space Res. 34, 700–709 (2004). [CrossRef]
8. D. Crisp, C. E. Miller, and P. L. DeCola, “NASA Orbiting Carbon Observatory: measuring the column averaged carbon dioxide mole fraction from space,” J. Appl. Remote Sens. 2, 023508 (2008). [CrossRef]
9. A. Kuze and H. Suto, “TOKYO and TSUKUBA models: TANSO precursor experiments,” Trans. JSASS Space Technol. Japan 7(26), Po_4_1–Po_4_6 (2009).
10. Y. Yoshida, H. Oguma, I. Morino, H. Suto, T. Yokota, G. Inoue, and A. Kuze, “Measurement of absorption spectra at an altitude of 800 meters by using the FTS (GOSAT-BBM) in SWIR spectral region,” presented at the 2nd Asian-Pacific Radiation Symposium, Kanazawa, Japan, July 2006.
11. S. C. Olsen and J. T. Randerson, “Differences between surface and column atmospheric CO2 and implications for carbon cycle research,” J. Geophys. Res. 109, D02301 (2004). [CrossRef]
12. N. Eguchi and T. Yokota, “Investigation of clear-sky occurrence rate estimated from CALIOP and MODIS observations,” Geophys. Res. Lett. 35, L23816 (2008). [CrossRef]
13. C. E. Miller, D. Crisp, P. L. DeCola, S. C. Olsen, J. T. Randerson, A. M. Michalak, A. Alkhaled, P. Rayner, D. J. Jacob, P. Suntharalingam, D. B. A. Jones, A. S. Denning, M. E. Nicholls, S. C. Doney, S. Pawson, H. Bösch, B. J. Connor, I. Y. Fung, D. O’Brien, R. J. Salawitch, S. P. Sander, B. Sen, P. Tans, G. C. Toon, P. O. Wennberg, S. C. Wofsy, Y. L. Yung, and R. M. Law, “Precision requirements for space-based data,” J. Geophys. Res. 112, D10314 (2007). [CrossRef]
14. Z. Kuang, J. Margolis, G. Toon, D. Crisp, and Y. Yung, “Spaceborne measurements of atmospheric by high-resolution NIR spectrometry of reflected sunlight: an introductory study,” Geophys. Res. Lett. 29, 1716 (2002). [CrossRef]
15. A. Kuze and K. V. Chance, “Analysis of cloud top height and cloud coverage from satellites using A and B bands,” J. Geophys. Res. 99, 14481–14491 (1994). [CrossRef]
16. E. Boesche, P. Stammes, R. Preusker, R. Bennartz, W. H. Knap, and J. Fischer, “Polarization of skylight in the A band: effects of aerosol properties,” Appl. Opt. 47, 3467–3480 (2008). [CrossRef] [PubMed]
17. T. Yokota, H. Oguma, I. Morino, and G. Inoue, “A nadir looking “SWIR” sensor to monitor column density for Japanese “GOSAT” project,” in Proceedings of the Twenty-fourth International Symposium on Space Technology and Science (Selected Papers) (Japan Society for Aeronautical and Space Sciences, 2004), pp. 887–889.
18. G. Inoue, T. Yokota, H. Oguma, A. Higurashi, I. Morino, and T. Aoki, “Overview of Greenhouse Gases Observing Satellite (GOSAT) of Japan,” presented at the AGU 2004 Fall Meeting, San Francisco, California, USA (13–17 December 2004), abstract A51C-0790.
19. T. Yokota, H. Oguma, A. Higurashi, I. Morino, T. Aoki, and G. Inoue, “Sensor specification demand of a nadir looking SWIR FTS aboard GOSAT to monitor column density in the clear sky condition,” presented at the AGU 2004 Fall Meeting, San Francisco, California, USA (13–17 December 2004), abstract A51C-0788.
20. H. Ishihara, N. Uemura, K. Nobura, T. Yokota, A. Higurashi, I. Morino, T. Aoki, and G. Inoue, “Retrieval precision tests of column amount from simulated data of the GOSAT SWIR FTS by applying Rodgers’ method,” presented at the AGU 2004 Fall Meeting, San Francisco, California, USA (13–17 December 2004), abstract A51C-0789.
21. N. Saitoh, R. Imasu, Y. Ota, and Y. Niwa, “ retrieval algorithm for the thermal infrared spectra of the Greenhouse Gases Observing Satellite: potential of retrieving CO2 vertical profile from high-resolution FTS sensor,” J. Geophys. Res. 114, D17305 (2009). [CrossRef]
23. H. Bösch, G. C. Toon, B. Sen, R. A. Washenfelder, P. O. Wennberg, M. Buchwitz, R. de Beek, J. P. Burrows, D. Crisp, M. Christi, B. J. Connor, V. Natraj, and Y. L. Yung, “Space-based near-infrared measurements: Testing the Orbiting Carbon Observatory retrieval algorithm and validation concept using SCIAMACHY observations over Park Falls, Wisconsin,” J. Geophys. Res. 111, D23302 (2006). [CrossRef]
24. L. S. Rothman, D. Jacquemart, A. Barbe, D. Chris Benner, M. Birk, L. R. Brown, M. R. Carleer, C. Chackerian Jr., K. Chance, L. H. Coudert, V. Dana, V. M. Devi, J.-M. Flaud, R. R. Gamache, A. Goldman, J.-M. Hartmann, K. W. Jucks, A. G. Maki, J.-Y. Mandi, S. T. Massie, J. Orphal, A. Perrin, C. P. Rinsland, M. A. H. Smith, J. Tennyson, R. N. Tolchenov, R. A. Toth, J. Vander Auwera, P. Varanasi, and G. Wagner, “The HITRAN 2004 molecular spectroscopic database,” J. Quant. Spectrosc. Radiat. Transfer 96, 139–204 (2005). [CrossRef]
25. N. A. J. Schutgens, L. G. Tilstra, P. Stammes, and F.-M. Bréon, “On the relationship between Stokes parameters Q, and U of atmospheric ultraviolet/visible/near-infrared radiation,” J. Geophys. Res. 109, D09205 (2004). [CrossRef]
26. L. G. Tilstra and P. Stammes, “Earth reflectance and polarization intercomparison between SCIAMACHY onboard Envisat and POLDER onboard ADEOS-2,” J. Geophys. Res. 112, D11304 (2007). [CrossRef]
27. T. Urabe, A. Kuze, T. Hamazaki, N. Baba, S. Minami, and H. Saruwatari, “The instrumentation and the contamination control activity of thermal and near-infrared sensor for carbon observation (TANSO) on GOSAT,” Proc. SPIE 6291, 629101 (2006). [CrossRef]
28. T. Urabe, A. Kuze, H. Suto, and T. Hamazaki, “Overview of GOSAT contamination control activity and test results summary,” Proc. SPIE 6678, 667817 (2007). [CrossRef]
30. F. Sakuma, “New development of InGaAs radiometer for preflight cross-calibration measurement,” in Proceedings of IEEE 2007 International Geoscience and Remote Sensing Symposium (IEEE, 2007), pp. 4136–4139. [CrossRef]
31. F. Sakuma, S. Kawakami, and A. Kuze, “Development of InGaAs radiometer for preflight cross-calibration measurement,” in Proceedings of IEEE 2008 International Geoscience and Remote Sensing Symposium (IEEE, 2008), pp. IV-1356–IV-1359
32. F. Sakuma, C. Bruegge, D. Rider, D. Brown, S. Geier, S. Kawakami, and A. Kuze, “OCO-GOSAT Preflight Cross Calibration Experiment,” IEEE Trans. Geosci. Remote Sens. (to be published). [CrossRef]
33. T. C. Stone, “Radiometric calibration stability and inter- calibration of solar-band instruments in orbit using the moon,” Proc. SPIE 7081, 70810X (2008). [CrossRef]
34. C. J. Bruegge, N. L. Chrien, R. R. Ando, D. J. Diner, W. A. Abdou, M. C. Helmlinger, S. H. Pilorz, and K. J. Thome, “Early validation of the multi-angle imaging spectroradiometer (MISR) radiometric scale,” IEEE Trans. Geosci. Remote Sens. 40, 1477–1492 (2002). [CrossRef]
35. T. Yokota, H. Watanabe, O. Uchino, I. Morino, Y. Yoshida, and S. Maksyutov, “Current status of the GOSAT data handling facility, data retrieval and inverse model algorithms, validation plan, and GOSAT research announcement,” presented at the AGU 2008 Fall Meeting, San Francisco, California, USA (15–19 December 2008), abstract A32B-02.