A multi-year, high-frequency (15-minute) archive of public urban-mobility feeds for the city of València, Spain, self-collected and curated because the official portals expose only the live state and retain no history.

1. Data Records

All material is sourced from the Ajuntament de València open-data portal and is redistributed here under CC-BY 4.0.

1a. PRIMARY streams

These four streams come from the Ajuntament de València open-data portal (OpenDataSoft), published under Creative Commons Attribution 4.0 (CC-BY 4.0), which permits redistribution with attribution. They are included in full:

Folder	Phenomenon	Geometry	Entities	Sampling
`estat_trafic_VLC/`	Road-segment operational state	LineString	376	96/day (15 min)
`espires_VLC2/`	Car loop-detector intensity	Point	1277	96/day (15 min)
`espires_bici_VLC/`	Bicycle loop-detector intensity	Point	118–120	96/day (15 min)
`valenbici/`	Valenbisi station availability	Point	276	96/day (15 min)

1b. SUPPLEMENTARY 2020 car loop-detector bundle

A separate bundle, recovered from the city's legacy real-time open-data (RTOD) JSON API, extends the car loop-detector record back to 2020 and thus through the first COVID-19 lockdown:

espires_VLC2/2020_json/ — 2020-02-28 → 2020-11-25 (112 days; gap 2020-04-16 → 2020-09-21), one JSON bundle per snapshot.

It is Ajuntament-sourced (CC-BY 4.0) and included in full. Note the autumn 2020 intensity values appear on a rescaled footing relative to spring (documented in the caveats in section 4 below); restrict cross-period comparisons accordingly.

1c. AIR-QUALITY & METEOROLOGY COMPANION

A companion environmental layer is bundled to make the mobility archive directly usable for cross-domain (traffic ↔ air quality ↔ weather) studies:

gva_air_quality/ — hourly air-pollutant and meteorological observations for the nine Generalitat Valenciana (RVVCCA) monitoring stations in the city of València, 2020-01-01 → 2025-12-31, one zipped CSV per year.

Unlike the mobility feeds, these data are already public and historically retained by the GVA open-data portal (dadesobertes.gva.es); they are not a novel recovery. They are redistributed here, filtered to València city and harmonised/time-aligned with the mobility record, purely for convenience. Source Generalitat Valenciana, CC-BY; the citable contribution of the deposit remains the mobility archive.

2. Directory layout

<root>/
├── estat_trafic_VLC/    analysis-ready Parquet layer    
├── espires_VLC2/       analysis-ready Parquet layer 
├── espires_bici_VLC/        analysis-ready Parquet layer 
├── valenbici/               analysis-ready Parquet layer 
├── data_quality/         analysis-ready Parquet layer 
├── data_manifest.csv        machine-readable inventory of uploaded files
├── gva_air_quality/         analysis-ready Parquet layer
├── raw/                 raw ZIPs
├── CITATION.cff
└── README.md

Inside each daily ZIP, snapshot files embed the timestamp, e.g. valenbici_01-01-2023_00-15-02.csv. CSVs are semicolon-separated with a header row; coordinates appear both as GeoJSON geo_shape and as geo_point_2d lat/lng.

3. Schemas (header fields)

valenbici (English header, until ~2023-06): gid;name;number_;address;open;available;free;total;ticket;updated_at;…;geo_shape;geo_point_2d
valenbici (Spanish header, from ~2023-06): Direccion;Numero;Activo;Bicis_disponibles;Espacios_libres;Espacios_totales;ticket;fecha_actualizacion;geo_shape;geo_point_2d (same semantics; parser maps both)
espires_VLC2 (car): gid;Ángulo;idpm;ih;created_date;…;geo_shape(Point);geo_point_2d
espires_bici_VLC: Geo Point;Geo Shape;angulo;fecha_actualizacion;hora_actualizacion;idpm;ih
estat_trafic_VLC: gid;Denominación;Estado;Id.Tram;…;geo_shape(LineString);geo_point_2d
2020 RTOD JSON (supplementary car bundle): {summary, resources:[{modified, intensidad, angulo, ycoord, xcoord, sensor, uri}]}
air companion (gva_air_quality/air_valencia_YYYY.csv): unmodified GVA schema, semicolon-separated, UTF-8, decimal comma ,, missing marker -, 42 columns: COD_ESTACION;NOM_ESTACION;FECHA;HORA;SO2;CO;NO;NO2;NOx;O3;PM10;PM2.5;PM1;NH3;C6H6;C7H8;C8H10;Direc.;H.Rel.;Precip.;Pres.;R.Sol.;Ruido;Temp.;UV-B;Veloc.;Veloc.max.;As;BaA;BaP;BbFA;BjFA;BkFA;Cd;DahA;FA;HMN;H2S;IcdP;Ni;Pb;PST. One row per station × hour (FECHA=YYYY-MM-DD, HORA=00..23). Coordinates are not in the CSV → join COD_ESTACION to stations_valencia.csv.

4. Known caveats

Day-level completeness varies by stream: valenbici 7.5% missing, espires_bici 12.9%, espires_VLC2 30.2%, estat_trafic 34.6%. Largest gaps documented in the quality flags.
Refresh ("frozen export") regimes, from a monthly dynamic test over the whole archive (see data_quality/regime_probes.csv):
- valenbici: dynamic throughout, except frozen 2024-04..05 and 2024-08.
- espires_bici_VLC: static until 2023-09; dynamic 2023-11..2024-07; intermittent 2024-08..12; dynamic from 2025-01.
- espires_VLC2 (car): dynamic until 2024-10; intermittent frozen 2024-11..2025-05; dynamic from 2025-06.
- estat_trafic_VLC: dynamic until 2023-04; static 2023-05..2024-02; dynamic from 2024-03.
Car intensity -1 is a sentinel for missing/invalid → treat as NA.
Road-state denomination field has UTF-8 mojibake; numeric/geometric fields are clean.
No gap-filling is applied; mask gaps rather than interpolate.
A machine-readable quality mask is provided: data_quality/quality_flags.csv (one row per stream × calendar day: present, expected_snapshots, observed_snapshots, mean_entities, regime ∈ {missing, static, dynamic}, notes); regime is propagated from the monthly probes, so every present day is classified (no "unverified"). The directly-measured evidence is in data_quality/regime_probes.csv. Regenerate at exact per-day resolution with code/validation/quality_flags.py --archive <dir>.
air companion (gva_air_quality/): hourly (not 15-min) — aggregate the mobility snapshots to hourly to join. Pollutants and meteorology are at different stations (NO2/NOx 7/9, O3 6/9, PM10/PM2.5 5/9; temperature/wind/pressure/radiation only at the 2 meteorological stations); heavy metals/PAHs/UV-B largely empty for VLC. See air_coverage_summary.csv. Already-public, historically retained at source; no gap-filling applied. Regenerate with code/acquisition/gva_air_quality.py --link.

5. License & attribution

All streams (Ajuntament-sourced, including the supplementary 2020 car bundle): CC-BY 4.0 (top-level LICENSE). Attribution: "Source: Ajuntament de València open-data portal (valencia.opendatasoft.com), CC-BY 4.0; compiled by C. Ferri et al." plus the acquisition date range.
Air-quality & meteorology companion (gva_air_quality/): Generalitat Valenciana, CC-BY. Attribution: "Source: Generalitat Valenciana, RVVCCA air-quality open data (dadesobertes.gva.es), CC-BY." Redistributed for convenience; the authoritative, full-history copy remains at the GVA portal.

6. Citation

Ferri, C., Martí-Pérez, F., Martínez-Plumed, F., & Moros-Daval, Y. (2026). A high-frequency open dataset of urban mobility in València, Spain, spanning 2020–2025 (1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.20817050

A machine-readable CITATION.cff is included; fill in authors, DOI and version tag.

Downloads last month: 51

Total file size:

14.4 GB