Result of a visual detection of non-natural records in streamflow time series for the Explore2 project

Résumé 0

The dataset collects anomalies reported during a visual inspection of for 570 (611 minus 41 stations belonging to the EDF network) streamflow time series available on the French database Hydroportail (https://hydro.eaufrance.fr/, date of extraction: 2021-03-01). Visual inspections were conducted in the first semester of 2022 as part of the Explore2 project. Streamfow time series were visually analysed by 43 evaluators (mostly hydrologists) in order to flag periods of anomalies that may reflect non-natural streamflow or measurement disturbance. Five types of anomalies were available: linear interpolation, drops, noise, point anomaly, and other. Description of the fields: code_station_h3: Code of the hydrometric station in relation to the SANDRE reference used in Hydroportail [10 characters] code_site_h3: Code of the hydrometric site in relation to the SANDRE reference used in Hydroportail [10 characters] code_h2: Former code of the hydrometric station in relation to the SANDRE reference used in Banque Hydro [8 characters] name_station_h3: Name of the station (Name of the river and nearest town) in Hydroportail [characters] name_site_h3: Name of the site (Name of the river and nearest town) in Hydroportail [characters] name_h2: Former name of the station (Name of the river and nearest town) used in Banque Hydro [characters] start: Date of the first day of the anomaly [YYYY-MM-DD] end: Date of the last day of the anomaly [YYYY-MM-DD] type: Type of anomaly reported [“Linear interpolation”, “Drops”, “Noise”, “Point anomaly”, and “Other”] evaluator: Anonymized evaluator code [2 characters]