Dataset used in the paper : Interpretable deep learning for characterizing sinkhole to supply well transfer dynamics in karst aquifers

The dataset consists of an aggregated set of time series. The first column contains the dates, and the second column corresponds to the turbidity measured at the supply well.The columns Maximum rainfall intensity (mm/h), Rainfall depth over the two previous days (mm), Effective duration (min), and Rainfall depth (mm) represent the hydroclimatic variables used as inputs in both the WaterSed model and the deep learning models.Columns with the prefix SED_ correspond to sediment fluxes, while those with the prefix RUISS_ correspond to runoff values. The number following each prefix identifies the associated sinkhole. Consequently, for each sinkhole, the dataset includes one sediment flux variable and one runoff variable.