DEEPSEEK AI Server —2024 Hourly Dataset for Water & Environmental Modeling.

This package provides harmonized hourly data for 2024 to support training, validation, and benchmarking of predictive models in urban water and environmental contexts. It includes time series for network flow rate, pressure, reservoir level, rainfall, air temperature, and a model-based demand forecast across three reference entities (site:scada-NET01, site:res-ALFA01, site:met-BO01), a derived feature store for ML, a station catalogue (~1,200 entries), a 1,000-cell geospatial grid (GeoJSON, EPSG:4326), and an operational events log for the same period. Data are produced via reproducible pipelines with explicit provenance and quality flags, enabling reliable reuse in research and innovation workflows.

• FAIR note: Rich (DataCite-style) metadata, open formats (CSV/GeoJSON; Parquet-ready), stable identifiers, provenance (provenance_id) and quality flags (qf), clear licensing, and documented processing ensure findability, accessibility, interoperability, and reusability.

• Units: All variables use UCUM encodings (e.g., m3/h, kPa, m, mm, Cel).

• Time zone: Timestamps follow ISO-8601 in UTC (Z) at hourly resolution.