FoBench: A Python toolbox for Fibre Optic Sensing signal processing and data handling.

Over the past decade, studies involving fibre optic sensing (FOS) have increased substantially, and consequentialy, optical fibres have become a prominent option for seismo-acoustic data acquisition. FOS can leverage existing telecom infrastructure in locations where installing conventional seismic stations is difficult, while providing dense measurements at high spatial and temporal sampling rates. However, replacing conventional sensor workflows with fibre optic sensing introduces major practical challenges such as: very large data volumes, heterogeneous native formats, high computational demand for fast processing and visualisation, and the need for intuitive yet flexible programming interfaces.

Several open-source tools such as DASPy, DASCore and Xdas already address parts of these challenges. Here we present FoBench as a complementary architecture focused on practical, reproducible end-to-end workflows for FOS data handling and baseline signal processing. FoBench is designed to ease transition from conventional seismo-acoustic workflows by adopting usage patterns familiar to ObsPy and Pyrocko users. It supports native formats (I/O) from multiple interrogator manufacturers, provides high-speed interactive plotting for seamless data inspection, and organises campaign-scale archives through a structured Project-Unit-Dataset model.

FoBench also targets operational scalability. It includes discontinuity-aware dataset handling, configurable processing pipelines, and memory-efficient greater-than-memory workflows via chunked outputs, with four interchangeable parallel wrappers (multiprocessing, MPI, Dask, and Process Pools). In addition, FoBench’s architecture is aligned with the in-development Geo-INQUIRE proposed metadata scheme to improve consistency and interoperability across archives.

We present the design principles of FoBench, illustrate representative processing workflows on seismo-acoustic FOS datasets, and discuss benchmark comparisons against available toolboxes. Our goal is not to replace existing ecosystems, but to provide a lightweight, interoperable, and user-oriented framework that bridges the gap between raw FOS files and reproducible scientific products.