This paper proposes Binomial models for the inclusion of historical data into flood frequency analysis, which recognize the uncertain nature of the perception threshold and the starting date of the historical period. The procedure is applied to a long systematic series 1816–2020 and 13 historical floods over 1500–1815 (Rhône River, Beaucaire, France). Inclusion of historical floods reduces the uncertainty flood quantiles, even when only the number of perception threshold exceedances is known.

Suspended-sand fluxes in rivers vary with time and space, complicating their measurement. The proposed method captures the vertical and lateral variations of suspended-sand concentration throughout a river cross section. It merges water samples taken at various positions throughout the cross section with high-resolution acoustic velocity and discharge measurements. The method also determines the sand flux uncertainty and can be easily applied to other sites using the available open-source code.

The dataset contains concentrations and fluxes of suspended particle matter (SPM) and several particle-bound contaminants along the Rhône River downstream of Lake Geneva. These data allow us to understand the dynamics and origins. They show the impact of flood events which mainly contribute to a decrease in the contaminant concentrations while fluxes are significant. On the contrary, concentrations are higher during low flow periods probably due to the increase of organic matter.