Several researchers working on the axes of solid particle transport, river flow, as well as the vulnerability of structures built on rivers have focused their work on the phenomenon of local scour, where many scour formulas have been proposed. Most of these formulas have been developed using laboratory data, estimated from simplified scaled-down models. This led to uncertainties and considerable inaccuracies in the scour estimation, sometimes causing serious consequences for the stability of structures subject to scour. These formulas were rarely tested using field data, where, until recently, there was very little data to validate them. This study aims to propose a new local scour formula using multiple nonlinear regression (MNLR). The data used is in situ data from an extensive database. The new proposed formula includes different parameters of the three fields involved in the triggering and development of the local scour, namely; the bridge, the flow, and the suspended characteristic of the eroded particle. The results obtained in the training phase and the validation show relatively better performances of the new formula compared to the existing empirical formulas.