A national flood mark database spanning several centuries contains valuable information to characterize ancient flood events. However, the network of flood mark sites is distinct from the network of hydrometric stations, making this information difficult to use. This work describes a probabilistic model jointly describing flood marks at sites and flood peaks at stations. The model is based on the estimation of Hidden Climate Indices driving both flood marks and peaks: this allows transferring information between the two variables despite them being measured on distinct networks. The model is applied to about 300 flood mark sites (1705-2015) and 200 stations (1904-2015) in France. Results demonstrate that flood marks allow estimating the time-varying probability of exceeding some high discharge threshold at stations during the whole period 1705-2015, which largely predates the existence of stations. The resulting probability maps provide quantitative information on the extent and spatial structure of ancient floods.