The acceleration approach – an alternative way of processing GRACE data
What is the acceleration approach?
The acceleration approach is based on Newton’s equation of motion, i.e. the kinematic observations are linked to the dynamics = forces. Practically it means that we either need to observe accelerations directly, e.g. by accelerometers, or we need to derive them from other type of observations e.g. by numerical differentiation from positions. Since we are interested in the recovery of one particular force, namely the gravity field and its temporal variations, all other forces need to be subtracted by using e.g. models.
Hold on – I am confused. I’ve heard that accelerometers onboard GRACE measure non-gravitational forces.
This is the particular setup of GRACE. The accelerometers are placed in the center of mass which is in free-fall, i.e. they sense no gravity but only non-gravitational forces such as solar radiation pressure and air drag. The accelerations related to the gravity field is observed by observing the motion of the satellite itself, i.e. the satellite is the test mass or “accelerometer” but the accelerometer inside the satellite measures non-gravitational forces. This is smart engineering, isn’t it?
Is the acceleration approach better than others?
Yes and no – the approach has its advantages, e.g. there is no accumulation of numerical integration errors or model errors since accelerations are directly linked to forces and no variational equations have to be solved. This also allows for a point-wise application which is especially suitable for regional and local applications. On the downside the acceleration approach applied to the case of GRACE needs not only the observation of range accelerations but also of a centrifugal component which requires the usage of GPS observations. Unfortunately this limits the precision of the solution to the precision of GPS and one cannot take full advantage of the precision of the K-Band observation if no proper measures are taken. Obviously this will not yield competitive solutions for GRACE but the problem can be overcome by reducing the observations to residual quantities.
Thus the acceleration approach is just another flavor of processing GRACE data. From a theoretical point of view all approaches should give identical results as they are all based on Newton’s equation of motion. Practically the different processing schemes come with their particular advantages and disadvantages. By combining the different solutions we will be able to benefit from the advantages and to mitigate the disadvantages. This is the strength of the project.
