Atmosphere and Climate

Data from the Global Ozone Monitoring by Occultation of Stars (GOMOS) instrument on-board the European environmental satellite Envisat have been used for retrieval of ozone and temperature profiles by new algorithms, followed by profile validation. We discuss ozone profiles, determined from GOMOS transmission data, validated with operational GOMOS ozone profiles and ECMWF (European Centre for Medium-Range Weather Forecasts) analysis data. For ozone we developed an optimal estimation retrieval scheme, using sensibly selected channels from the Spectrometer A transmission spectra within 260 nm to 340 nm and 602 nm to 634 nm. Furthermore, profiles of a new GOMOS temperature profile retrieval are compared to CHAMP (CHAllenging Minisatellite Payload) and ECMWF analysis data. GOMOS temperatures are gained by exploiting pointing data of the Steering Front Assembly (SFA) and the Star Acquisition and Tracking Unit (SATU), which provide information on the refraction of the star light in the atmosphere and thus allow to derive refractive bending angle profiles. The bending angle profiles are then converted via refractivity and pressure profiles to temperature profiles. Bending angles were assumed to have errors of 3 μrad. Statistical optimization of observed bending angles with model bending angles was used to provide adequate data quality for the Abel transform from the stratopause region upwards, which led to a significant gain in temperature retrieval accuracy up to 40 km height due to suppressed downward propagation of errors induced by Abel transform and hydrostatic integral. Based on data from year 2002, a set of GOMOS ozone and bending angle derived temperature profiles is validated and discussed.

The Full Spectrum Inversion (FSI) method was developed in the beginning of 2002 in an effort to solve the multipath problem in radio occultation measurements. The physical ideas, which lead to the method, were that the occultation path could be considered as a synthetic aperture and the radio occultation Doppler frequency in a single path was a monotonic function of time. In star occultations, at optical wavelengths, the multipath problem is easily solved by having a lens in front of an array detector separating the beams in space. The lens is performing a spatial Fourier transform i.e., a plane wave is focused into a point displaced from the optical axis an amount given by the direction (the spatial frequency) of the plane wave. The analogy to this space processing method in time, is to have a “time lens”, which can separate multiple temporal frequencies occurring at the same time: Obviously this is what a temporal Fourier transform does. These ideas were implemented in 2001 and tested successfully on simulations of radio occultation signals, which had circular satellite orbits. However, for non-circular orbits the plain Fourier method turned out to give a not fully correct result and the work on the FSI emerged realizing that some preprocessing steps were necessary in order to eliminate the impact of non-radial orbits. This involves pre-calculation of phases, which, multiplied on the occultation signal, reduces the impact of the non-circular orbits on the resulting Fourier transform of the preprocessed signal. In 2003 the phasematching method was developed, where the impact of the non-circular orbits was totally solved, but with the cost that the processing could not be implemented with a fast Fourier transform. Both the FSI and the phasematching methods will be discussed in detail in this paper. The present development on the FSI method includes its practical implementation and making the method robust for mass processing of radio occultation signals. Filtering of signals in the FSI method is important and different filtering methods will be discussed in this paper.

During the past decade, researchers have been working to develop methods for the assimilation of Global Positioning System (GPS) radio occultation data into numerical weather prediction (NWP) models. Until recently, two strategies have received the most attention: 1) assimilation of the retrieved bending angle profiles, and 2) assimilation of the retrieved refractivity profiles. The assimilation of retrieved bending angle profiles, e.g., via a ray tracing model, has the advantage that horizontal refractivity gradients, affecting the observations, can be accounted for. However, accurate computation of ray paths through a NWP model is very time consuming in an operational system, and approximations to the ray tracing have to be made. In contrast, the assimilation of retrieved refractivity profiles, interpreted as being representative of the local vertical structure in the atmosphere (also known as assimilation of local refractivity), is simple and fast, but this approach does not account for the influence of the horizontal gradients. Recently, some new observation operators have been developed which are both fast and, to a large degree, capable of taking into account the influence of the horizontal gradients. These new observation operators rely on predefined ray trajectories, and could also become useful for future assimilation of other kinds of occultation data, e.g., ionospheric GPS occultation data, absorption data from low earth orbit constellations, or data from solar/stellar occultation experiments. In this review, a brief account of past and present efforts to develop efficient observation operators for the assimilation of GPS occultation data into atmospheric models will be given, ranging from early suggestions to the recently developed observation operators.