FrenchPiezo: the French mainland groundwater level multivariate time series

FrenchPiezo is the French mainland multivariate time series dataset of groundwater levels (a.k.a piezometric levels) across the French mainland. The dataset contains 1026 multivariate time series, each made of 3 dimensions which are:

• the piezometric level (p),
• the precipitation (tp),
• the evapotranspiration (e).

Each time series is identified by a code (bss) which is the identifier of the piezometer used to measure the piezometric level. The measurements are sampled daily from January 2015 to January 2021, corresponding to 2221 days. The piezometric levels are collected from Hub'Eau, the French free API for accessing water data. The climate data tp and e are collected from the Copernicus ERA5 archive using their free API. These data are in the file dataset_2015_2021.csv in which time series with more than 50 missing values have been dropped. The file dataset_2015_2021_nomissing_linear.csv is the same dataset in which missing values have been imputed using linear interpolation.

In addition to the climate data, information about the nature of the soil where each time series is collected are downloaded from BDLISA and are given in the file dataset_stations.csv . BDLISA classifies soils as hydrologic entities (EH), characterized by a set of attributes described here: 

More information on how we created the dataset, the scripts used to collect the data, and some experiments on forecasting the future values of groundwater levels using global and local foresting methods can be found on our GitHub page.