Data Processing
Defining product contents, implementing processing algorithms and generating the end products. From plan-driven approaches to data-driven approaches.
Payload Data processing activities first require a deep analysis into the Mission and Instrument objectives. The goals range from the definition of the interfaces and Product contents, to the analysis and implementation of processing algorithms suitable to the Instrument(s), and their integration into the Ground Segment infrastructure.
During the design of Data Processing facilities, a careful study must be performed on the scientific community needs for data coming from the Instrument(s), and also on the available processing algorithms. Prototyping activities may be initiated in case sufficient knowledge is not available, or new ideas are being worked out. Outputs from this initial stage are the Auxiliary data needs, Mission Products formatting and contents and Algorithms Baselines.
Regarding processing strategies, there are two main currents nowadays: plan-driven and data-driven. Plan-driven processing (e.g. ENVISAT PDS) relies on a-priori information on the Instrument(s) configuration and data downlinks to establish an optimised processing plan. The processing algorithms are ed according to this plan and the output is expected to follow the plan as well. Data-driven processing does not rely on any knowledge in turn, it analyses all the input data it finds, in order to try to determine the best processing algorithm.
As it can be seen, plan-driven processing allows for better optimisation of the Ground Segment resources, but in turn is less flexible to exceptions that may appear when the data received is different from the expected plan. On the other hand, data-driven approaches are completely flexible, they depend on the level of autonomy built in their knowledge database, but may have difficulties in the optimisation of GS resources under certain circumstances.
DEIMOS is currently working in several projects involving data processing, covering all the different steps into the design and final implementation of Data Processors. It has implemented an early prototype of the instruments on-board Darwin, designed to detect and characterize planets, to be used by the scientific community. Following the prototyping activities a number of projects are currently focused on improving the processing chain for optical instruments, having in mind future missions. In the field of operational processors, DEIMOS is implementing the SMOS Near Real Time Processor (NRT), and, in our subsidiary at Portugal, the SMOS L1 Processor Prototype, detailing the required L1 Product Formats and L1 Algorithms, as well as implementing a data-driven L1 Processor quite close to the operational one. DEIMOS has also been responsible for the migration of the operational ENVISAT AATSR Processors from AIX environment to linux, adapting also latest interface conventions at ESA/ESRIN, as well as analysing the performance differences between the two OS.
|
