Application of BEL1D for sNMR data interpretation

The interpretation of sNMR data is still mainly performed
using deterministic or stochastic inversion schemes.
sNMR signal to noise ratio is often low regarding
electromagnetic noise pollution which coupled to nonuniqueness makes uncertainty quantification challenging.
Here, we propose a new Bayesian scheme relying on a
learning step and a prediction step to perform the
interpretation of sNMR data including uncertainty
quantification: BEL1D. With it, it is possible to estimate
the uncertainty of models parameters from a given dataset
in a rapid manner compared to stochastic inversion and
reach an equivalent posterior estimation after iterative
prior resampling. The learning step can even be used to
multiple datasets to improve performances with only the
prediction required. Additionally, BEL1D could be used
with any geophysical methods.