Living Planet   


GOCESOFT: a software for GOCE data processing

Dr Pavel Ditmar(1), Prof. Roland Klees(1) , and Mr. Alexis Van Eck van der Sluijs(1)

(1) Delft University of Technology, Kluyverweg 1, 2629 HS Delft, Netherlands


The primary goal of the "GOCESOFT" software is the Earth's gravity field modeling on the basis of satellite-to-satellite and satellite gravity gradiometry data to be collected by the GOCE mission. The data can processed either separately or jointly. The necessary input information is: (i) observation points in the spherical geocentric coordinate system - they can be obtained directly from the reduced-dynamic orbit; (ii) the angles defining the orientation of the Local Orbital Reference Frame (LORF) at each observation point; (iii) gravity gradients in the LORF; (iv) average satellite accelerations in the LORF. The latter information is supposed to be obtained by the 3-point double differentiation of the kinematic orbit in an inertial frame. Importantly, both the gravity gradients and the satellite accelerations have to be pre-processed in order to remove: (i) signal from the reference static Earth's gravity field; (ii) signal from temporal gravity variations, e.g. related to tides; (iii) influence of non-gravitational effects, e.g. the atmospheric drag (if the drag-free control system fails to remove all such effects entirely). The result of the processing is a set of geopotential coefficients. These coefficients should be added to those describing the reference gravity field in order to obtain the final product: a model of the Earth's gravity field, which matches the GOCE data optimally. The GOCESOFT software can also be used to solve two other tasks: (i) simulation of gravity gradients or/and satellite accelerations on the basis of a static gravity field model; (ii) computation of the normal matrix and of the right-hand side vector in the system of normal equations in connection with gravity gradients or/and satellite accelerations. The results of the latter computation can be used: (i) to obtain the covariance matrix of the geopotential coefficients; (ii) to model the Earth's gravity field using both GOCE data and data from other sources jointly. At present, GOCESOFT has been adapted to and tested on: (i) Linux PC; (ii) multi-processor computer SGI Origin 3800; (iii) multi-processor computer SGI Altix 3700 (with Itanium-2 processors as the elementary base). We expect that GOCESOFT can also be used (as is or with minor modifications) on any other multi-processor platform with the Unix/Linux operating system; the only requirement is the presence of MPI and Lapack/Blas libraries.

We have put significant efforts in the optimization of GOCESOFT both in terms of CPU time and computer memory. In particular, the memory requirements are of the order of 200 bytes per observation point for either type of data (in the gravity field modeling mode). This means that one would need only about 12 Gb of memory to process all the GOCE data (about 3x10^7 observation points). By the time GOCE flies, this amount can probably be achieved even on a plain PC. As far as the CPU time is concerned, it is of the order of 1 hour on the SGI Origin 3800 platform with 64 CPUs for a 1-year data set of either type with 1-s sampling, provided that the maximum degree/order is set equal to 300, and the optimal regularization parameter is known. GOCESOFT is also capable of finding the optimal regularization parameter automatically (by means of the generalized cross-validation method). The CPU time in this case, however, increases by about the factor 30 (which is still well within reasonable limits).


Workshop poster

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry