Poster Abstract

P10.26 Kinga Albert (Max Planck Institute for Solar System Research)

Theme: Data processing pipelines

Metadata and its importance in SO/PHI's on-board data processing

The Polarimetric and Helioseismic Imager (PHI) is the first solar spectropolarimeter on a deep space probe: the Solar Orbiter (SO). Due to telemetry limitations, long command-response turnaround times and high accuracy requirements in the data products, SO/PHI performs extensive on-board data processing.

The on-board processing system runs a full science data pipeline on the acquired raw data containing several irreversible steps, performs on-board instrument characterisation, as well as calculates operational parameters. Due to the limited telemetry the download of raw or partial data products is not feasible, therefore in-operation testing of the pipeline will be a challenge. At the same time, unexpected results are foreseen: on one hand due to the novelty of the mission, on the other hand since on-ground pipeline tests are restricted to synthetic data, with only limited possibility for extension with real SO/PHI images during commissioning. This will force us during operations to perform blind debugging of the pipeline. As support in this, besides tools as the instrument simulator, we have implemented a consistent and extensive metadata logging system.

Metadata logging starts at the time of image read-out, when the instrument parameters used to acquire the dataset are recorded. This is extended during each step of the data processing. We record metadata on several software layers to ensure security as well as provide the necessary information for its interpretation on ground. We log each operation performed, together with its parameters and its return value, detected errors and generated warnings. The metadata serves several purposes: it is interpreted by the pipeline and used to determine necessary operations, it serves as information for the scientists interpreting the data, and it supports in-flight debugging of the pipeline.

This contribution presents the metadata logging system together with examples on its support for blind debugging.','Data processing pipelines'