Comparing the data-reduction pipelines of FRIPON, DFN, WMPL, and AMOS: Case study of the Geminids

Context . The number of meteor observation networks has expanded rapidly due to declining hardware costs, enabling professional and amateur groups to contribute substantial datasets. An accurate data reduction remains challenging, however, because variations in processing methods can significantly affect the trajectory reconstructions and orbital interpretations. Aims . Our goal is to thoroughly compare four professionally produced meteor data-reduction pipelines (FRIPON, DFN, WMPL, and AMOS) by reprocessing FRIPON Geminid observations. This analysis can be used for a comparison with other data-reduction methods. Methods . We processed a dataset of 5 84 Geminid fireballs observed by FRIPON between 2016 and 2023. The single-station astrometric data were converted into the global fireball exchange (GFE) standard format for uniform processing. We assessed variations in trajectory, velocity, and radiant and orbital element calculations in the pipelines and compared them to previously published Geminid measurements. Results . The radiant and velocity solutions provided by the four data-reduction pipelines are all within the range of previously published values, with some nuances. Particularly, the radiants estimated by WMPL, DFN, and AMOS are nearly identical, but FRIPON reports a systematic shift in right ascension (−0.3°) that is caused by an improper handling of the precession. Additionally, the FRIPON data-reduction pipeline also tends to overestimate the initial velocity (+0.3 km s −1 ), which is due to the deceleration model used as the velocity solver. The FRIPON velocity method relies on a well-constrained deceleration profile, but for the Geminids, many are low-deceleration events, which leads to an overestimation of the initial velocity. At the other end of the spectrum, the DFN tends to predict lower velocities, in particular, for poorly observed events. This velocity shift vanishes for the DFN when we considered Geminids alone with at least three observations or more, however. The primary difference identified in the analysis concerns the velocity uncertainties. Although all four pipelines achieved similar residuals between their trajectories and observations, their velocity uncertainties varied systematically. WMPL outputs the lowest values, followed by AMOS, FRIPON, and DFN. Conclusions . From this Geminid case study, we find that the default FRIPON data-reduction methods, while adequate for meteoritedropping events, are not optimal for all cases. Specifically, FRIPON tends to overestimate velocities for low-deceleration events because the fit is less strongly constrained, and the nominal radiants are not correctly output in J2000. On the other hand, the other data-reduction pipelines (DFN, WMPL, and AMOS) produce consistent results, provided that the observational data are sufficiently robust, that is, more than ∼50 data points from at least three observers. A key takeaway is that we need to reevaluate how the velocity uncertainties are estimated. Our results show that the uncertainty estimates vary systematically in different pipelines, even though the goodness-of-fit statistics is generally similar. The increasing availability of impact observations from varying sources (radar, video, photo, seismic, infrasound, satellite, telescopic, etc.) calls for greater collaboration and transparency in data-reduction practices.