Implementation of ROSA radio occultation data handling into EUMETSAT and GRAS SAF processing

Topic of the GRAS-SAF Visiting Scientist activity 16 has been the implementation of ROSA radio occultation data handling into EUMETSAT and GRAS SAF processing tools. ROSA data referred to observations taken on-board OCEANSAT-2 mission. Even if we are talking of standard Radio Occultation data, the format of raw binary data is peculiar, and processing chains already implemented for other Radio Occultation missions, should be adapted to this further format. Moreover, ROSA antennas, ROSA coverage, ROSA sampling rate, etc. are different from GRAS or COSMIC data. Main goals of this VS activity have therefore been the implementation of ROSA processing into the EUMETSAT YAROS prototype (in order to produce excess-phases, amplitudes and bending angles over impact parameter) and into the GRAS SAF ROPP processor (for the processing into bending angle, refractivity and higher level atmospheric profiles), dealing mainly with data interfacing issues. The first part of the activity was carried out in EUMETSAT, where the YAROS prototype has been adapted to handle ROSA data. Before doing that, an in-depth analysis has been done in order to identify all the ROSA data characteristics and issues. In particular, the Thales Alenia Space “decoding” software has been completely review, in order to understand how these data were formatted inside the native binary ROSA data stream coming from the OCEANSAT-2 telemetry, and inside the standard Level1 Engineered data file available for users. This preliminary analysis allows the definition of several ROSA data structures that could be easily implemented directly inside the YAROS prototype. At the end of this first part, two different releases of the EUMETSAT prototype were available: the first one able to manage only the scientific ROSA engineered data, containing the basic observables for Radio Occultation purposes (the Level 1a data, the one normally made available for users); the second one which implements ad-hoc data structures where all the data generated by the ROSA receiver (and contained in the raw binary format, the Level 0 data) are stored. About thirty hours of ROSA observations, for which orbit and clocks were already available, were used as input to the adapted YAROS/ROSA prototype and completely analyzed. Only the YAROS prototype adapted to ROSA Engineered Level 1 format was used. A quality check of L1 and L2 basic observables (mainly L1 and L2 Excess-Phase and Signal-to-Noise ratio profiles) contained in the Level 1a YAROS output file has been performed. Even if a complete ROSA dataset is available for one entire month of observations, the production of all the necessary POD products for the entire month was not undertaken. Its in-depth analysis will be carried out in the future, when the Memorandum of Understanding between Italian Space Agency and EUMETSAT for the ROSA data processing will be signed. During this activity we preferred to concentrate the work on the YAROS prototype adaptation to ROSA data, making the prototype ready to handle all the ROSA data available, and not only the one contained in the Level1 Engineered data normally available to the user. In order to let these NetCDF Level1a YAROS/ROSA output files to be further processed into higher Radio Occultation products, a converter from NetCDF 4.0 to NetCDF 3.0 file (with the required data structures correctly set for the GRAS-SAF ROPP processing). This was performed during the second part of the Visiting Scientist activity at DMI, where YAROS/ROSA Level 1a files were both processed to bending angles, to refractivity and higher level products. Analysis of the bending angle and refractivity data against ECMWF and collocated radio occultation data were also performed, in order to have a preliminary idea on the quality of the final atmospheric products that can be achieved analyzing ROSA data. Moreover, YAROS/ROSA Level 1a files were also processed into Level 1b products (bending angle profiles) using the L1a2L1b YAROS code and compared with the correspondent forward modelled ECMWF collocated data. As a general comment on the ROSA data quality it has to be noted that, even thought L1 data are in line with that observed by the other Radio Occultation instruments, L2 data shows some problems. Problems which are mainly related to OCEANSAT-2 issues. ROSA on-board OCEANSAT-2 was equipped only with the Velocity Radio Occultation antennas, therefore only rising events can be recorded. Moreover local multipath on-board OCEANSAT-2 is very strong because of solar panels and because of the scatterometer antenna which both are moving. In particular this local multipath was not modelled or measured on ground, since an in-orbit platform manoeuvre was made necessary and created a permanent and unexpected yaw bias on the platform. And these problems mainly impact L2 tracking, which basically starts too high in atmosphere and which is affected by long data gaps. Only for a small number of events L2 is available in troposphere. In these cases, the quality of atmospheric profiles retrieved using the GRAS-SAF ROPP processing chain is in line with that obtainable from other Radio Occultation missions. But, for all the other cases for which L2 is available higher in the atmosphere, modifications to the L2 extrapolation/interpolations and different Statistical Optimization setting are probably necessary to adapt the GRAS-SAF ROPP_PP Processor to such bad data. From the other point of view, YAROS software correctly handle such ROSA data, even if a high number of rejected profiles are seen with respect what normally happens in processing data from other Radio Occultation observations