Satellite measurements allow global observations of atmospheric constituents. Passive remote sensing instruments use three basic types of geometry: nadir, limb and occultation. Critical point in the latter two is always the pointing knowledge, i.e. the precise knowledge of the viewing direction and with that the altitude grid of the observations.

SCIAMACHY (SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY) is a passive remote sensing moderate-resolution imaging ultraviolet–visible–near-infrared spectrometer on board the European Space Agency's (ESA) Environmental Satellite (Envisat). Envisat was launched in March 2002 from Kourou, French Guiana, to a sun-synchronous orbit, where it accomplished a successful 10-year mission. In April 2012, the mission ended due to a platform malfunction. SCIAMACHY observes Earth's atmosphere in nadir, limb and solar/lunar occultation geometries and provides column and profile information of atmospheric parameters of relevance to ozone chemistry, air pollution and climate monitoring issues .

SCIAMACHY's solar occultation measurements are done in the Northern Hemisphere, when the sun rises due to the orbital motion of the platform. On the day side of the orbit, alternating limb and nadir measurements are performed. The same air masses are probed first in limb and a few minutes later in nadir. Lunar occultation is made in the Southern Hemisphere, if the waxing moon with a phase larger than 0.5 is in the field of view and the moon rise occurs after sunset. Dark current measurements are performed during eclipse. Various types of monitoring measurements are executed in the daily, weekly and monthly calibration periods.

The pointing information for the instrument is derived from the attitude information of the platform, which itself is calibrated using the three star trackers of Envisat. Here, we use SCIAMACHY's own measurements to further improve the pointing knowledge. In , precise pointing knowledge has been derived for SCIAMACHY's solar occultation measurements. It used the sun over limb port measurements, which are the solar occultation measurements continued above the atmosphere. In this work, we will additionally use the moon over limb port measurements and the sun over sub-solar port observations to derive further pointing information. The viewing direction of the instrument towards the sun or moon is derived from these measurements. From the attitude information of the platform, the viewing direction towards sun or moon is calculated. Systematic differences between the measured and the calculated viewing directions indicate a mispointing. Appropriate mispointing parameters are needed, which improve the attitude information of the platform and the SCIAMACHY line of sight information.

The first set of in-flight mispointing parameters has been derived in 2007 by , used in SCIAMACHY data products (re-)processed since 2010. Priority was minimising the large tangent height offsets of about 1 km observed in the limb measurements. The sun over limb port and over sub-solar port measurements were used for this. In , a remaining systematic offset in tangent height of 280 m, a horizontal offset of 5 km and seasonal pattern in both offsets have been shown for solar occultation data.

In this work, the combination of the pointing measurements from all three measurement types listed above is used for fitting new mispointing parameters, which further improve the pointing accuracy for all SCIAMACHY measurements.

Section  introduces the quantities and the parts of the instrument relevant for this work. The relevant measurement types are described in Sect. . The variability of the lunar disk needs special care as described in Sect. . Finally, all measurements are utilised to derive improved mispointing parameters in Sect. . Appendices A and C add technical details of the algorithm. Appendix B lists the important abbreviations used throughout this paper for the reader's convenience.

Three types of measurements are used to derive additional pointing information: the sun and moon observed through the limb port (the solar/lunar occultation measurements at atmospheric tangent heights) and the sun observed through the sub-solar port. Due to observing geometry defined by the orbit and the viewing ports, sun over limb port is performed above the Northern Hemisphere, moon over limb port above the Southern Hemisphere and the sun over sub-solar port above the equatorial region.

The sun over limb port measurement (Fig. ) follows the rising sun, performing vertical scans over the solar disk. The viewing direction towards the centre of the sun in elevation is determined by fitting the maximum intensity during the scan. In azimuth, the sun is tracked with the SFD. Details are given in . Sun over limb port is the upper part of the solar occultation measurement performed every orbit. Once per day, a shortened solar occultation measurement is performed, which ends with a 2 s period with tracking by the SFD also in elevation. The difference between the two methods is below 1 mdeg .

Three different sequences for sun over sub-solar port measurement are done. In the sequence used here, the sun is tracked in elevation by the SFD (Fig. ). The ASM is not involved in this light path. The sun moves in azimuth direction through the field of view because of the orbital motion of the satellite. At the maximum intensity during the measurement, the sun is at 270∘ azimuth angle (the opening direction of the sub-solar port). This sun over sub-solar port measurement is part of the monthly calibration. After October 2006, it was scheduled every 3 days.

The moon over limb port measurement (Fig. ) used here is a monitoring measurement, which is performed well above the atmosphere. In its first phase, the moon is tracked in azimuth and elevation by the SFD. This phase is used here. The SFD adjusts the viewing direction towards the IC of the actual lunar disk. The location of the IC is discussed in Sect. . This monitoring measurement is performed once per day, if the moon appears in the field of view. This is the case for about 1 week per month during the phase of the waxing moon with a lunar phase larger than 0.5. Note: the regular lunar occultation measurements are performed every orbit with the same restrictions. Additionally, the tangent point has to be at local night time, limiting the usable period to about 6 months per year.

Mispointing of SCIAMACHY can have two main causes. First, the assumed mispointing angles to determine the SRAR coordinate system may not be correct, either because the SCIAMACHY misalignment is not well described or because the platform attitude knowledge is inaccurate. The SRAR coordinate system can be aligned by appropriate roll, pitch and yaw mispointing angles.

Second, the scan mirror positions might have an offset. As the ASM rotates in the yaw axis, an ASM offset corresponds to a yaw mispointing if the ASM mirror is used in the measurement. An ESM offset corresponds to a roll mispointing. However, if the ASM is also used (limb light path), an ESM offset corresponds to a mispointing in pitch; see Figs.  and .

In summary, we have five parameters to minimise the mispointing: the three mispointing angles and two offsets for the mirror positions.

Sub-solar measurements are available once per month, and therefore we selected roundabout one measurement per type and month during the mission. Excluded is the year 2002, because the first sub-solar measurements failed due to a timing problem. In total 96 moon over limb port, 96 sun over limb port and 90 sun over sub-solar measurements are used.

A non-linear Levenberg–Marquardt fit is used to minimise elevation and azimuth angle offsets by varying the five mispointing parameters. Initial guesses are the currently used final values from 2007. The newly fitted mispointing parameters are listed in Table . For completeness, the previously used mispointing parameters are also given as explained in the table caption.

Figure  shows the improvement. For the measurements used in the fit, the observed elevation/azimuth angle offset (dλ/dφ) is plotted, once using the mispointing parameters from 2007 and once using the new parameters (marked as 2016). The elevation and azimuth angle offsets for all three measurement types are clearly reduced for the full time series. In elevation, the offsets with opposite signs for the lunar and solar measurements over limb port have been corrected. In azimuth, the large horizontal offset for the measurements over the limb light path has been removed.

The new parameters briefly have the following impact on the pointing:

The ASM mirror offset fixes the large azimuth angle offset.

We provide a method to determine improved mispointing parameters for SCIAMACHY on board ENVISAT from in-flight measurements. The viewing direction towards sun or moon is derived for three types of measurements and compared to the expected viewing direction. These types are sun and moon over limb port and sun over sub-solar port measurements.

SCIAMACHY's sun follower device adjusts the azimuth direction in sun over limb port, the elevation direction in sun over sub-solar port and both directions in moon over limb port observations. Sun over limb port elevation offset are derived from the scans over the solar disk: at maximum intensity the instrument points to the solar centre. Similarly, the solar disk moves in azimuth over the field of view of the sub-solar port by the orbital motion; again the maximum intensity defines the direction towards the solar centre.

The moon is an inhomogeneous and variable target, and the SFD points to the IC of the moon instead of the lunar centre. We investigated reference images from the USGS ROLO and derived a parameterisation for the location of the IC from the observation geometry only.

The viewing directions are compared to the ones according to the attitude information of the platform to derive azimuth and elevation angle offsets. Mispointing angle in pitch, roll and yaw for the platform can correct the assumed attitude. Zero offsets for the positions of the scan mirrors can improve the pointing of the instrument.

With a suitable subset of the available pointing measurements, we minimise the mispointing for the whole mission by varying the five mispointing parameters in a non-linear Levenberg–Marquardt fit.

The azimuth and elevation angle offsets observed with the previous set of mispointing parameters are clearly reduced by the newly derived set. For sun over limb port measurements, the mean offset in tangent height is reduced from 276 m to negligible 29 m. The amplitude of the seasonal cycle of the tangent height offset is reduced from 122 to 48 m. The previously observed horizontal offset of 5 km is reduced to 71 m.

The tangent height altitudes of SCIAMACHY's solar occultation measurements are changed by -130 to -330 m, lunar occultation is altered by 0 to +130 m. The limb measurements are changed by -50 to +60 m, including the removal of an east–west asymmetry in the limb tangent height offsets. The horizontal location of the tangent point is for all three cases changed by about 5 km.

The new mispointing angles and the mirror position offsets according to Table  will be used in the next release of the SCIAMACHY data processor (Level 1b version 9 and Level 2 version 7) and improve the geolocation for all measurements. Most important is the better knowledge of the tangent heights for SCIAMACHY's limb and occultation measurements.