OverviewOpenMarineStream v1.0 is a raw multi-sensor marine time series dataset created for research on online anomaly detection, sensor fusion, and concept drift in environmental monitoring systems. The archive contains nine CSV files, each representing an individual sensor stream exported over the period 10-28 to 10-30: C3 Temperature C3 Turbidity Flow Flow Flow Temperature Optode Concentration Optode Saturation Optode Temperature SEB45 Conductivity SEB45 Salinity A Labelled Preprocessed Dataset Each file includes a common timestamp column and numerical readings from the corresponding instrument. Together, these streams capture water temperature, turbidity, flow rate, conductivity, salinity, dissolved oxygen concentration, oxygen saturation, and related probe temperatures in a FerryBox-style flow-through system. Preprocessing (reference pipeline)For RoLA V2.1 experiments, these raw CSV files were merged and preprocessed as follows: Timestamp harmonisation & merging Convert all timestamp columns to a standard datetime format. Merge the nine datasets on the timestamp using an inner join, so each row represents a time step where all selected sensors have readings. Standardise column names to a consistent schema (e.g. C3_Temperature, C3_Turbidity, Flow_Flow, Flow_Temperature, Optode_Concentration, Optode_Saturation, Optode_Temperature, SEB45_Conductivity, SEB45_Salinity). Handling missing data Apply forward-fill (ffill) and backward-fill (bfill) to propagate recent valid values through short gaps. Use linear interpolation to estimate remaining missing values in continuous numeric columns. Feature selection & normalisation Retain all numerical sensor readings as candidate features for anomaly detection. Optionally apply Min–Max normalization to scale each series into 0,10, 10,1 for certain models; RoLA V2.1 can also operate on raw values. Intended useThe dataset is designed to support: Development and benchmarking of real-time anomaly detection algorithms on multivariate environmental data. Experiments with sensor fusion (optical, electrochemical, and flow-through sensors). Studies of concept drift and robustness in marine monitoring streams. It is directly compatible with the RoLA V2.1 streaming anomaly detection demo, which supports synthetic data, USGS/UK Hydrology sources, and user-uploaded CSVs based on this schema.