Python-based event detection framework for space weather applications

Publicly available databases of space weather events are critically important for validating numerical models and improving the accuracy of space weather forecasts. Recently, the Heliophysics Event (HPE) List format has been introduced to address these problems through a standardized multi-platform event list architecture facilitating cross-mission event detection, tracking and investigation.

Combining, coordinating, matching and adapting multiple HPE lists may present certain challenges because different users could have different goals and approaches for defining the events, and may adhire to incompatible initial formats and standards. Creating intelligent data mining algorithms for an automatic run-time compilation of new HPE-formatted event lists from raw spacecraft data, based on flexible sets of observational signatures and constraints, can expand the capabilities of the modern forecasting methods, simplify data processing, and ultimately improve the accuracy of space weather forecasts supporting timely decision-making. In this talk, we present a new Python-based event detection package for creating HPE lists directly from observational databases serving the specific goals of a researcher. The package consists of three main units.

I. The event detection routines enable building new HPE lists from provided datasets using a customized initialization parameter file containing the required metadata and the list of unique processing parameters such as event detection thresholds, criteria for the amplitude and duration of event, etc. II. The quality check routines perform a post-filtration of the detected events, which can be done automatically or manually. At this stage, the users could compare the event detection results based on several different thresholds and other selection criteria. The final event list can be saved in a user-preferred HPE format, such as plain text HPE, VoTable HPE, or another commonly used format. III. The event list analysis and visualization routines providing the users with versatile tools for data processing and plotting customized for the needs of specific event list studies.

The presented event detection package has been successfully tested on a variety of observational datasets representing space weather activity in different parts of the Heliosphere.