SCIAMACHY Product Handbook

# Calibration Monitoring

 CHAPTER 4 4. Calibration & Monitoring Spaceborne spectral measurements over long time periods require to translate the measured signals into physical quantities and to maintain this process with high precision. Therefore calibration and monitoring of the instrument is a crucial prerequisite for any successful retrieval of atmospheric geophysical parameters. Calibration of the instrument should be valid during any point in the mission. For calibration measurements which can not be performed in flight it is necessary to perform these before launch. Once in orbit the instrument can be expected to change, and needs to be calibrated in flight where possible, and monitored where in-flight calibration is not feasible. 4.1 Calibration 4.1.1 The General Calibration Equation The goal of the calibration is to convert electronic signals of detectors (Binary Units – BU) into physical units (e.g. W/m2/nm). This is achieved by applying a complex sequence of individual calibration steps to measurement data. For a detailed description of each step see Slijkhuis (2000a) and Lichtenberg et al (2005). Following the light from the earth through the telescope, slit, spectrometer, detector, and on-board processing we arrive at the following equations for a scanning instrument which measures only the spatially integrated light of one ground pixel at a time: where I is the polarised intensity at wavelength λ and geographic position lat and lon, polarisation is represented in Stokes notation with a vector, and Pt is the point spread function of the telescope which may affect the polarisation state of the light and is represented here as a matrix. Optical distortion due to the telescope is implicitly included in Pt which thus may depend on α and β. The coordinates α and β are defined in the focal plane of the telescope at the entrance slit to the spectrometer part of the instrument. Convolution is denoted with *. After having passed through the telescope with optical properties represented by the point spread function with possible polarisation sensitivity, the light passes through the entrance slit where the intensity masked out by the slit is removed: where F is a function of the focal plane coordinates α and β and is zero where the light is blocked and unity where the light passes unhindered through the slit. The optics after the slit project the image of the slit on the detector and add wavelength dispersion in order to obtain a spectrum. For instruments with an integrated instantaneous field of view, like SCIAMACHY, the detector can be 1-dimensional. For instruments with a spatially resolved slit, the detector must be 2-dimensional. We consider only the 1-dimensional case here: where Po is the wavelength and polarisation dependent point spread function of the optics, including the dispersion introduced by the spectrometer. The coordinate x is the index for the detector pixels, and integration is performed over each pixel individually. Possible self-emission of the instrument (thermal background radiation) is covered by B which may be position dependent. Note that polarisation information is lost once the signal has been projected on the detector. Once integrated over the detector pixels, the signal can be treated electronically. Up to this point the signal has been considered purely linear in I, but the detector material and electronics may introduce non-linearity and hysteresis:

where E is the function describing the transfer of I into the digitally sampled detector signal Sdet, taking into account the exposure time texp. The function E may depend on the history of I, represented here as I(t).
On-board processing may modify the digitally sampled signal Sdet, in the case of SCIAMACHY only in the form of co-adding multiple read-outs of the detector. The electronic sampling function E can be simplified into a linear term in I, a constant offset in I, a constant offset in Sdet, and additive terms describing remaining effects.
Combining all of the equations above, considering a dedicated correction for polarisation sensitivity (see below), describing the effects of the point spread function of the optics Po on the recorded spectrum as an additive component, and introducing explicit temperature dependence of the detector quantum efficiency, the equations can be simplified to (equ. 4-1b):

where Γinst is the total transmission of the instrument, QE the detector temperature dependent quantum efficiency, Sstray the stray light, DC the total dark signal and Selec electronic effects such as non-linearity. This equation must be solved for every detector pixel. In order to obtain the spectrum as a function of wavelength λ for each pixel, the wavelength has to be determined and the equation has to be inverted to calculate the intensity I. Generally, the transmission of the instrument is dependent on the polarisation of the incoming light.

The experience gained from GOME flying on-board the ERS-2 satellite, where various air-vacuum effects led to calibration problems, showed that spectrometers should ideally be calibrated under thermal vacuum conditions. In the case of SCIAMACHY a range of incidence angles on the mirror(s) and mirror-diffuser combinations had to be covered in the calibration, requiring a rotation of the instrument. The available vacuum chamber hardware did not allow rotation of the instrument itself, and only provided views in a limited angular range centred on nadir and limb in the flight direction. Therefore, a combination of thermal vacuum (TV) and ambient measurements was used.

The radiometric sensitivity and the polarisation sensitivity of the instrument were measured under TV conditions for one reference angle α0 and all necessary instrument modes (limb, nadir and irradiance). In order to be able to calibrate all incidence angles on the mirrors (or diffusers), component level measurements of all possible mirror combinations and the mirror/ESM diffuser combination were made under ambient conditions. The detectors used in the ambient calibration were different from the detectors used on-board SCIAMACHY. These ambient measurements occurred for a set of angles – including the reference angle measured under TV conditions – and a set of selected wavelengths. From such ambient measurements the so-called scan-angle correction is calculated.

The reference angle measurement is used to transfer the results from the ambient measurement to the TV conditions. Measurements included both unpolarised and linearly polarised light. Combining TV measurements with the ambient measurements gives ideally the correct instrument response for all incidence angles at Begin-of-Life (BOL) of the instrument. The implicit assumptions for the combination of the TV and ambient measurements are that the polarisation dependence of the mirrors and diffusers are the same in air and in vacuum and that there is no temperature dependence. Both assumptions are reasonable for SCIAMACHY, since uncoated mirrors are used.

Critical points in the transfer of ambient and TV measurements are the geometry (incidence angles on the mirrors or diffusers), the illumination conditions and the detector used for the component measurements. Obviously, errors in the geometry lead to an incorrect angle dependence for the calibration quantity to be measured. Light levels during instrument measurements and during component measurements will always be different. While the footprint of the light source on the component can be matched to the footprint during the instrument measurements, it is impossible to recreate the exact illumination conditions. This may introduce systematic errors into the calibration. Finally, care has to be taken that the detector from the measurements under ambient conditions does not introduce artefacts.

In order to minimise potential errors from the measurements performed under ambient conditions, only ratios of measurements were used for the calibration where possible. The individual calibration parameters derived from the on-ground measurements are combined into a set of data files, the so-called Key Data files. These Key Data are applied by the data processor to derive calibrated spectra. (fig. 4-1)

click to enlarge

 fig. 4-1: Calibration concept for SCIAMACHY. The final calibrated Earth radiance spectra are obtained by applying several calibration steps to the measured Earthshine signals. They include in-flight calibration measurements (red), on-ground measurements performed under thermal vacuum conditions (green) and component measurements from on-ground ambient tests (blue). The optical performance monitoring (red) provides additional corrections. (Graphics: SRON)

The TV on-ground calibration was performed during several campaigns using the OPTEC facility. SCIAMACHY had been placed inside the vacuum chamber with the thermal hardware being replaced by a system based on liquid nitrogen and heaters to reach and maintain the correct temperature. Optical windows in the tank allowed the light from external optical stimuli to enter. In the Key Data the effect of the optical window has been compensated for. The OPTEC facility was first used for requirement verification tests, i.e. a check to see if the instrument met its requirements. Later, calibration measurements were performed in OPTEC. Mainly due to major hardware changes in the instrument, several OPTEC campaigns had to be scheduled and executed.

After the OPTEC-1 campaign it was discovered that the mounting of the optics was unreliable at low temperatures. The refurbishment implied a re-testing, and thus an OPTEC-2 campaign. Here it was found that the SWIR channels 7 & 8 were out of focus so that the OPTEC-3 campaign, executed after repositioning of channels 7 & 8, verified the required performance. In the OPTEC-4 campaign the instrument was relocated inside the facility to get representative illumination for the on-board diffuser. This configuration yielded improved measurements quantifying the instrument’s radiometric properties. Finally, the stray light in the UV channels was reduced by some hardware changes which required the OPTEC-5 campaign. In all, the various OPTEC campaigns ran from summer 1997 till spring 2000.
The ambient calibration was executed between December 1997 and April 1998 in a dedicated set-up, the so-called ARCF (Absolute Radiometric Calibration Facility). In order to allow a rotation of the mirror(s) or the mirror/diffuser combination to any required position, they were placed on a special optical bench. Having two mirrors or a mirror plus a diffuser on the optical bench permitted direct measurement of the combined response of both optical elements and calibration of all SCIAMACHY instrument modes at the appropriate angles. A monochromator and polarisers were used to obtain the response for different wavelengths and polarisations. All measurements took place in a class 100 cleanroom with a controlled temperature of 20° C and 50% air humidity.