GITpy: An open-source, data-driven framework for robust ground-motion parameter estimation across tectonic settings

Quantifying earthquake ground motion requires the robust separation of source, propagation, and site effects, a long-standing challenge in seismology and seismic hazard studies. We present GITpy, an open-source Python framework that implements the non-parametric Generalized Inversion Technique (GIT), designed to flexibly and efficiently decompose S-wave Fourier amplitude spectra into their fundamental physical components. The software provides a unified, reproducible, and highly configurable environment for performing large-scale GIT-based inversions and for exploring the associated modeling choices.

GITpy adopts a one-step, data-driven inversion strategy that minimizes a priori assumptions and exploits the increasing availability of dense, high-quality seismic datasets. Its modular, object-oriented design allows users to customize all key aspects of the inversion workflow and to independently parameterize source and path terms, controlling source and attenuation complexity through flexible choices and solution constraints.

Post-inversion modules allow independent parameterization of attenuation functions and source spectra, as well as extraction of station-specific apparent source spectra. These capabilities make GITpy suited not only for deriving ground-motion parameters, but also for systematically diagnosing model performance and quantifying trade-offs between source, path, and site terms. The software architecture is designed to facilitate reproducible research, with standardized input/output formats, configuration files, and plotting routines that can be readily shared and extended.

GITpy is conceived as a general-purpose tool, and we demonstrate its performance through a validation test in Central Italy, where we compare inversion results with those obtained using the widely adopted GIT implementation of Oth et al. (2011). This benchmark confirms the robustness of the inversion core and illustrates how GITpy can reproduce established results while offering greater flexibility in model setup and analysis. Ongoing applications in northeastern Italy (Cataldi et al. 2026) further showcase its ability to manage complex tectonic settings and to explore spatial variations in source-related parameters, without requiring changes to the core inversion engine.

By integrating source, attenuation, and site analyses within a single open-source framework, GITpy promotes methodological consistency in ground-motion studies and facilitates comparisons across regions and tectonic environments. GITpy is under active development, with new modules and functionalities continuously added, including tools for systematic residual analysis and a dedicated module for the local calibration of the ML scaling law. It is intended as a community resource for seismological research, seismic hazard assessment, and ground-motion modeling.