A new python code to invert 36Cl cosmogenic nuclide dataset on normal fault bedrock scarps: comparison with previous published codes and results on the accuracy of the retrieved seismic history of two normal fault systems in Central Apennines

Measuring ³⁶Cl cosmogenic nuclides on exposed bedrock fault scarps has now been used in several places in the Mediterranean to retrieve ages of the fault seismic history (e.g. Mechernich et al. 2022, Iezzi et al. 2021 and Cowie et al. 2017).

In Central Apennines, around the Fucino basin, at least 15 ³⁶Cl sampling sites were analyzed in previous studies to interpret the ³⁶Cl data as seismic history or slip-rates. Several codes (e.g., Beck et al. 2018, Shlagenhauf et al. 2010) were used as a basis for solving ³⁶Cl production equations to calculate the ³⁶Cl concentration resulting from bedrock scarp exhumation history. Some codes included an MCMC routine to retrieve the seismic histories the closest to the dataset. The main differences between the various codes lie in: 1-the fault history prior to exhumation, 2-the parameters previous authors decided to inverse (as an example, mean density of the colluvium is inversed in Beck et al. 2018 but not in Tesson et al. 2019) and 3-the a priori distribution of those parameters (for instance, the time between two earthquakes follows an inverse gaussian distribution for Beck et al. 2018 but a uniform distribution for Tesson et al. 2019). I have compared the various codes and run them on the same dataset (one site at Campo Felice, one site at Roccapreturo and one site at Magnola) and found that retrieved seismic histories are similar, although the estimation of uncertainties differs.

Moreover, all previous cited codes run under Matlab or Fortran. Fortran codes have the advantage of fast computing time but could be cumbersomeI here propose a new code, adapted from Tesson et al. 2019, in the more accessible and widely used Python language. The inferred pre-exposure is also inversed and is a function of the height of the fault cumulative escarpment. The parameters considered are the number of events, ages of event, the associated slips, the long term slip rate, the quiescence and the pre-exposure and their optimal evaluation is done with a MCMC algorithm provided by Numpyro (Du Phan et al. 2019).

Using this new code, we have reanalyzed the 15 ³⁶Cl sites around the Fucino and, through a gaussian mixture algorithm, checked the hypothesis of common periods of activity throughout all the Fucino basin.