Start Page
Contact
Research
Teaching
CV
Publications




Lund Observatory
Research
Education
Library
Staff
Local events
Seminars
Resources
Links
Lund University

Research Interests

Current research is heavily focused on the Gaia space astrometry mission where I am involved in the core data reduction for Gaia, known as the Astrometric Global Iterative Solution (AGIS). This uses an astrometrically well-behaved subset of all observed objects called ‘primary stars’ which will number at least 100 million. The core solution simultaneously determines the astrometric parameters (positions, parallaxes and proper motions), the accurate spacecraft attitude and the geometric instrument calibration. In addition, a small number of global parameters will be estimated, one of these being the Parameterized Post Newtonian (PPN) parameter gamma (γ) due to rest mass. Of interest here is that some alternative theories of gravity predict deviations of these parameters from their values in general relativity (γ = β = 1). For PPN γ this deviation could be of the order of 10-5 to 10-8. It is, therefore, highly interesting to estimate the accuracy by which this parameter could be determined from Gaia observations.
The direct measurement of large angles is fundamental to Gaia's ability to construct a globally consistent reference system as well as to determine stellar parallaxes and PPN γ. Gravitational light deflection by the Sun causes an apparent shift of a distant object, by a few milliarcseconds, in the direction away from the Sun (green arrows in the figure).

General Relativity and Gaia

By contrast, stellar parallax causes an apparent shift towards the Sun (orange arrows in the figure). To first order, the two effects are highly correlated and difficult to distinguish. This correlation slows down the convergence of the iterative solution, so an artificial parameter has been introduced to estimate and eliminate this correlation without changing the final solution for PPN γ in any way. The PPN formalism was developed to make sense of the large number of alternative theories of gravitation, to elucidate their similarities and differences, and compare predictions with experimental results in a systematic way. The formalism characterizes the slow motion, weak field limit of gravitation by a set of 10 real-valued parameters, and is ideally suited to the analysis of solar system gravitational experiments. Measurements of PPN parameters can be compared with their values in Einstein’s general theory and can place constraints on which, if any, alternative theory of gravity is correct.  Of the 10 parameters used in the formalism, Gaia is expected to provide useful constraints on β, measuring the degree of nonlinearity in the superposition law of gravity, and γ, measuring the curvature of space-time (and, hence, the degree of gravitational light deflection). PPN γ large numbers of small-scale astrometric solutions are performed, taking into account the simultaneous determination of stellar astrometric parameters, the satellite attitude and PPN γ. Accurate determination of PPN γ requires that as many relatively bright (G < 16) stars as possible are used in the solution as these give the largest contribution due to lower observation noise. Extrapolating the results of these small-scale simulations to the full-scale solution for Gaia, assuming some 100 million primary stars, demonstrates that PPN γ could be obtained to ~10–6 which is significantly better than today's best estimate from the Cassini mission with an accuracy of 2×10-5 and also within the range predicted for some alternative theories to deviate from general relativity. This also assumes that remaining minute instrumental or modelling effects are not significantly correlated with the γ-signature in the astrometric signal.

While new research is focused on Gaia I have previously been involved in implementing numerical algorithms for the attitude and orbit control of spacecraft, and in developing precise orbit determination methods based on GPS techniques. In this context I have worked on some of Europe's largest satellite development projects, including the XMM-Newton, Integral, Herschel and Planck satellites. An important part of any satellite is the onboard computer and the attitude and orbit control algorithms contained within. I have been involved in implementing these algorithms and testing the flight critical software. My main area of specialisation concerns such algorithmic elements as the attitude determinator, the controllers, sensor data processing and actuator commanding. The algorithms use a variety of numerical techniques such as Householder transforms, eigenvalues, least squares methods and numerical integration. The objective was to retrieve sensor data and compute the spacecraft's attitude. Corrections are then applied to the spacecraft by means of the thrusters and momentum-based reaction wheels. Particularly complex algorithms were required for the Herschel and Planck missions due to the very high degree of pointing accuracy.

I have also helped develop Precise Orbit Determination techniques, which are one of the key aspects of most Low Earth Orbiting satellites. This becomes critical if additional timeliness requirements are imposed leading to a near-real time processing scenario. Such is the case for many new meteorological satellites in the processing of GPS data used in the generation of atmospheric sounding profiles. The processing of sounding data is based on the Doppler shift experienced by the signal emitted by an occulting GPS satellite while traversing the atmosphere. The high sensitivity of the signal to small perturbations in the atmosphere requires knowledge, with a very high degree of accuracy, of all contributing error sources. In particular, the error in the computation of the satellite's velocity has to be limited. The accuracy requirements impose a target of 1m in position and 0.1mm/s in velocity to be achieved within a few minutes of processing. I have conducted a number of studies for the European Space Agency to demonstrate the feasibility of such missions. I have, together with colleges in Terma and in collaboration with the University of Bern, developed a numerical processing scheme involving a least squares solution that can achieve the above accuracy (in fact ~ 4 cm position accuracy) under the required timing constraints.

Space Research and Development

Gaia and the Global Iterative Solution

Precise Orbit Determination

The Herschel and Planck Missions

The XMM-Newton and Integral Observatories

Computational Physics

Noncollinear Magnetism

Optical Properties of Amorphous Materials


Lund Observatory, Box 43, SE-221 00 Lund, Sweden
Visiting address: Sölvegatan 27
Phone: +46 46 22 27 300, Fax: +46 46 22 24614
E-mail: david@astro.lu.se
Last updated: August 29, 2010

About the astronomy house