4. IASI Level 1 Products Overview
The Infrared Atmospheric Sounding Interferometer is composed of a Fourier transform spectrometer (IASI) and an associated Integrated Imaging Subsystem (IIS). The Fourier transform spectrometer provides infrared spectra with high resolution between 645 and 2760 cm-1 (3.6 µm to 15.5 µm). The IIS consists of a broad band radiometer with a high spatial resolution. However, the IIS information is only used for co-registration with the Advanced Very High Resolution Radiometer (AVHRR).
The main goal of the IASI mission is to provide atmospheric emission spectra to derive temperature and humidity profiles with high vertical resolution and accuracy. Additionally it is used for the determination of trace gases such as ozone, nitrous oxide, carbon dioxide and methane, as well as land- and sea surface temperature and emissivity and cloud properties.
IASI has 8461 spectral samples, aligned in three bands between 645.0 cm-1 and 2760 cm-1 (15.5µm and 3.63 µm), with a spectral resolution of 0.5 cm-1 (FWMH) after apodisation (L1c spectra). The spectral sampling interval is 0.25 cm-1. The IASI sounder is coupled with the IIS, which consists of a broad band radiometer measuring between 833 cm-1 and 1000 cm-1 (12µm and 10µm) with a high spectral resolution. Table 4.1 summarises the spectral characteristics of IASI.
645.0 – 1210.0
8.26 – 15.50
1210.0 – 2000.0
5.00 – 8.26
2000.0 – 2760.0
3.62 – 5.00
Table 4.1: IASI's three spectral bands
The IIS is used for collocation between IASI and the AVHRR
and is only available during Level 1 processing. The full specification of the
IASI instrument is given in [RD14].
Warning: The radiometric calibration performed during the on-board processing provides three real spectra corresponding to each of the three bands of the instrument presented in Table 4-1. They are then merged to give one full spectrum covering the entire useful band of IASI. In these two interband regions (1145–1190 cm-1 and 1925–1980 cm-1), the measurement quality decreases at the edge of the band because the measurement noise increases due to the decrease in optical transmission and detector sensitivity.
The instrument may be operated in two different modes in which scientific data are acquired:
Considerable data rate reduction is achieved by pre-processing the calibration on board. The on-board processing consists of the Fourier transformation of the measured interferograms, the non-linearity correction for band B1, the radiometric calibration and merging of the three different bands into one spectrum. Processing and quality information are generated to ensure the quality of L0 products. This approach reduces the raw data rate from about 45 Mbit/s to about 1.5 Mbit/s.
On ground the IASI Level 1 processing is performed within the EPS core ground segment (CGS). This includes the radiometric post-calibration of IASI based on additional information from the IIS and the IASI spectral database. The geolocation of the IASI products is performed with the help of AVHRR L1b products. The calibrated IASI L1c spectra are sampled onto an equidistant grid and are apodised.
IASI is an across track scanning system with scan range of ±48° 20´, symmetrically with respect to the nadir direction. A nominal scan line covers 30 scan positions towards the Earth and two calibration views. One calibration view is into deep space, the other is observing the internal black body. The scan starts on the left side with respect to the flight direction of the spacecraft.
The effective field of view (EFOV) is the useful field of view at each scan position. Each EFOV consists of a 2 x 2 matrix of so-called instantaneous fields of view (IFOV). Each IFOV has a diameter of 14.65 mrad, which corresponds to a ground resolution of 12 km at nadir and a satellite altitude of 819 km. The 2 x 2 matrix is centred on the viewing direction. The instrument points spread function (PSF) is defined as the horizontal sensitivity within an IFOV. The IFOV diameter (D = 14.65 mrad) is defined so that the integral of the PSF over this circular area is larger than 95%. The non-uniformity within the inner 80% of the IFOV (D=11.72 mrad) is not larger than ±5%. The IIS field of view is defined by a square area of 59.63 x 59.63 mrad, consisting of 64 x 64 pixels and covering the same area as the IASI EFOV.
The instrument scans in a step and stare modus. Each interferogram is acquired within 151 ms. The 30 Earth interferograms per scan line are taken in equally spaced time intervals every 8/37 s so that a synchronisation with AMSU is achieved. Figure 4.1 summarises the synchronisation of IASI with the ATOVS instruments AMSU and MHS.
Figure 4.1: Synchronisation of IASI and ATOVS instruments on
|Scan type||step and stare||–|
|Number of Earth scans / line - EFOV||30||–|
|IFOV - shape at nadir||circular||–|
|IFOV - size at nadir||12||km|
|IFOV - size at edge of scan line across track||39||km|
|IFOV - size at edge of scan line along track||20||km|
Table 4.2: IASI scanning characteristics
The collocation between IASI and the ATOVS instruments is shown in the following figures. It can be recognised that IASI starts scanning shortly before AMSU (Figure 4.1), therefore the IASI EFOV is not fully centred within the AMSU field of view at the beginning and end of the scan line (just discernible in Figure 4.4).
Figure 4.2: Collocation of IASI (yellow) and AMSU (red). The distance between adjacent pixels is given for one scan line in km.
Figure 4.3: The collocation between IASI (yellow), AMSU (red), MHS (green) and HIRS (blue) is shown for four scan lines around nadir. The distance along and across track is given in km.
Figure 4.4: The collocation between IASI (yellow), AMSU (red), MHS (green) and HIRS (blue) is shown for four scan lines at the end of the IASI scan line. The distance along and across track is given in km.
The Level 1 ground processing of IASI is illustrated in Figure 4.5, below.
Figure 4.5: Level 1 processing chain including IASI L0, AVHRR L1b and the IASI spectral database
The IASI Level 1 is processed in three different chains, which have to be executed consecutively. It is important to note that the ISFRM chain has to be executed on a complete granule (22 lines) to obtain the spectral calibration and apodisation functions needed as input for the IASI chain. As the ISFRM chain uses the calibrated images from the IIS, the IIS chain must first be executed over a complete granule as well.
The IIS is calibrated using the deep space and the internal black body views. From these two readings the slope and the offset are generated and filtered. Finally the images are calibrated by applying these on a pixel-by-pixel basis.
The purpose of the ISRFEM chain is to supply the spectral calibration function, which is used to perform the spectral calibration within the IASI chain. Additionally the apodisation function is generated. Both the spectral calibration and the apodisation function depend directly on the position of the interferometer axis in the detector plane.
The actual interferometer axis is therefore calculated within this chain and the corresponding apodisation and calibration functions extracted and interpolated from the pre-calculated IASI spectral database.
188.8.131.52 Level 1a
The spectra have been radiometrically calibrated during on-board (L0) processing. At this stage an approximation with respect to wavenumbers used for the calculation of the Planck function is applied. Using the spectral calibration function from the ISRFEM chain, new Planck functions are calculated and an appropriate correction applied to the spectra.
As the emissivity of the internal black body is not unity, the contributions from the reflected radiance based on a model of radiant surfaces seen by the detector are taken into account.
The impact of the scanning mirror at different angles and its temperature dependency is accounted for.
Finally, the geolocation of IASI is estimated based on the results from the correlation of AVHRR Level 1b data and the calibrated IIS image.
184.108.40.206 Level 1b
To perform the resampling, the IASI Level 1a spectra are over-sampled by a factor of 5. The over-sampled spectra are finally interpolated on a new equidistant spectral grid by using a cubic spline interpolation.
220.127.116.11 Level 1c
The IASI Level 1b spectra are apodised using the apodisation function estimated within the ISRFEM chain.
The Level 1 processing finishes with the generation of the radiance cluster analysis based on AVHRR within the IASI IFOV using the IASI point spread function (IPSF). The offset between the IASI sounder and IIS and the IIS-AVHRR offset is used to produce the IASI-AVHHR offset.
An overall quality flag is generated from information supplied by the on-board and Level 1 processing (Figure 4.4). The quality flag and the performance indicators of the IASI Level 1 products are described in detail in section 4.3.2.
18.104.22.168 On-board processor
Quality information and performance measurements gained during the on-board processing are ingested into the Level 1 processing and used for the generation of the Level 1 quality flag and the performance indicators.
22.214.171.124 Level 1 processor
The quality control chain as part of the Level 1 processor generates instrument performance information and the overall quality flag. In this chain not only L1 processing information is taken into account but quality information from on-board processing is incorporated as well.
Detailed information about the instrument performance will be added when available.
The Boolean flags DEGRADED_INST_MDR and DEGRADED_RPOC_MDR give general quality information with respect to degradation due to instrument and/or processing. These variables are the generic quality indicators.
A further quality flag and instrument noise indicators are generated during the Level 1 processing based on information gained during on-board and Level 1 processing.
The Boolean quality flag (GQisFlagQual) indicates the quality of each of the three IASI spectral bands (see Table 4.1) for every IASI spectrum (named "GS1cSpect" in the case of the L1C product). GQisFlagQual is established by evaluating the formerly derived quality flags and quality information from on-board and Level 1 processing, e.g. quality of the internal calibration black body temperature measurement.
The new IASI L1 quality indicator (GQisFlagQualDetailed) provides further information if quality is degraded and provides the Boolean flag as a summary of all three bands (which is identical to the previous definition of the GQisFlagQual flag, for convenience). Please see the GQisFlagQualDetailed description table for a more detailed description.
The noise performance of the IASI sounder and the IIS is given by the performance indicators GQisQualIndexSpect and GQisQualIndexRad, GQisQualIndex and GQisQualIndexIIS. The contributions from the spectral calibration to the instrument noise relative to the nominal instrument noise in given by GQisQualIndexSpect. GQisQualIndexRad indicates the impact of the radiometric calibration of the noise level of the product. Contributions from both radiometric and spectral calibration on the instrument noise are given in GQisQualIndex, again relative to the nominal instrument noise. The performance of the IIS is indicated by GQisQualIndexIIS, which again measures the performances as the ratio between actual IIS noise and nominal IIS noise. These performance indicators are calculated on the basis of a complete scan line.
Performance indicators are meant to be used for updating the nominal instrument noise to its actual value rather than indicating the usability of the measurements itself. The usabilty of a IASI spectrum is indicated by the boolean flag GQisFlagQual for each band of every IASI spectrum.
The performance of the co-registration between IASI and AVHRR measurements is given by GQisQualIndexLoc. This value is the semi-major axis of the ellipse error in AVHRR pixel units (estimated by the Level 1 processing for each individual field of regards). GQisQualIndexLoc indicates the co-registration error and therefore small values indicate good quality.
A list of the quality flags is given in Table 4.3 and Section 10.
|Level 1 quality flag||Description||Boolean||Occurrence||MDR|
|DEGRADED_INST_MDR||Quality of MDR, degradation due to instrument||Yes
|line||1A, 1B, 1C|
|DEGRADED_RPOC_MDR||Quality of MDR, degradation due to processing||Yes
|line||1A, 1B, 1C|
|GQisFlagQual||Individual IASI-System quality flag||Yes
|Each spectral band of IFOV/spectrum||1A, 1B, 1C|
|GQisFlagQualDetailed||Detailed IASI-System quality flag||Yes
|IFOV||1A, 1B, 1C|
|GQisQualIndex||Indicator for instrument noise performance (contributions from spectral and radiometric)||No||line||1A, 1B, 1C|
|GQisQualIndexSpect||Indicator for instrument noise performance (contributions from spectral calibration)||No||line||1A, 1B, 1C|
|GQisQualIndexRad||Indicator for instrument noise performance (contributions from radiometric calibration)||No||line||1A, 1B, 1C|
|GQisQualIndexLoc||Indicator geometric quality index||No||line||1A, 1B, 1C|
|GQisQualIndexIIS||Indicator for IIS imager noise performance||No||line||1A, 1B, 1C|
Table 4.3: IASI Level 1 quality flags and performance indicators
Since July 2007, interpixel radiance differences at Level 1c have been observed in some parts of the spectra. These radiance differences were exceeding 0.1 K, though it should be noted that this is less than 0.1% in terms of radiance. This was not a non-conformance at instrument level because interpixel radiometry is specified on black body targets and atmospheric spectra have been analysed in this study. In any case, some applications, in particular L2 inversions using channels above 2000 cm-1, suffered from this effect. It was therefore decided to analyse the problem and try to find a solution to fix it. This point is also worth addressing in the frame of L1c reprocessing in order to produce hyperspectral climatological data records. The Centre National d’Études Spatiales (CNES) has been conducting the study.
After some investigation, it was found that the apodisation performed in the algorithm S1C described in [RD16] is not able to fully correct the impact of the cube corner constant offset. The two main differences between the cube corner constant shear effect and others are the introduction of:
The solution was to symmetrise the self-apodisation function used to apodise the spectra in the algorithm S1C and to resample by taking into account the true xZPD, even if this last point does not have a strong effect. In order to not modify the operational ground processing software, CNES has implemented the symmetrisation of the self-apodisation function and xZPD resampling through the Spectral Data Base (ODB), the one used since 7 February 2011.
The interpixel signature has been largely reduced, especially in band 3. This has a positive impact on the retrieved CO in the IASI L2 products. Note that a very small residual interpixel effect can still be observed, mostly in the difference between pixels 1/2 and pixels 3/4. This is mostly due to the ghost effect which cannot be corrected by the operational ground processing. The residual signatures are mostly seen in band 1 and band 3. The main reason for not correcting this effect is because its phase at ZPD is different for all spectra and cannot be simply and accurately estimated in flight.
The main purpose of the IASI Level 1 products are the assimilation of calibrated Level 1 radiance spectra at NWP centres and the input for the Level 2 processing chain for the retrieval of temperature and humidity profiles.