Esbrí et al. (2025) Intense Rainfall in Urban Areas: Characterization of High-Intensity Storms in the Metropolitan Area of Barcelona (2014–2022)

Identification

• Journal: Atmosphere
• Year: 2025
• Date: 2025-12-28
• Authors: Laura Esbrí, Tomeu Rigo, María Carmen Llasat
• DOI: 10.3390/atmos17010041

Research Groups

• GAMA Team, Department of Applied Physics, University of Barcelona, 08028 Barcelona, Spain
• IdRA, Water Research Institute, 08017 Barcelona, Spain
• Meteorological Service of Catalonia (Servei Meteorològic de Catalunya—SMC), 08017 Barcelona, Spain

Short Summary

This study characterizes intense rainfall in the Metropolitan Area of Barcelona (2014–2022) by comparing radar-derived VIL density (DVIL) with rain gauge observations and urban impact data, establishing a linear relationship between DVIL and short-duration rainfall intensity and evaluating its nowcasting potential.

Objective

• Characterize the spatiotemporal structure of intense convective events in the Metropolitan Area of Barcelona (AMB).
• Examine the relationship between DVIL and extreme rainfall.
• Assess the nowcasting potential of radar-derived storm tracking.
• Analyze directional patterns, propagation speeds, and seasonal variability of convective storms.

Study Configuration

• Spatial Scale: Metropolitan Area of Barcelona (AMB), approximately 636 km². Two regions of interest (ROI) were defined: ROI1 (Barcelona municipality plus a 5 km buffer zone, 101.3 km²) and ROI2 (a square area of approximately 20 km side around Barcelona, encompassing the entire AMB). Radar data had a spatial resolution of 1 km.
• Temporal Scale: The study period covered 2014–2022. Radar data had a temporal resolution of 6 min. Rain gauge data provided 1 min and 5 min rainfall intensities. Impact-related datasets were available daily. Convective cells were typically short-lived, often lasting less than 30 min.

Methodology and Data

• Models used: RaNDeVIL (Radar Nowcasting with Density of VIL) algorithm for storm cell detection and tracking.
• Data sources:
  • Volumetric radar products (Vertically Integrated Liquid (VIL), Echo Top 12 dBZ (TOP12), Constant Altitude Plan Position Indicator (CAPPI), and VIL density (DVIL)) from the Servei Meteorològic de Catalunya’s (SMC) XRAD C-band Doppler weather radar network (1 km spatial, 6 min temporal resolution).
  • High-resolution rainfall data (1 min and 5 min intensities) from the BCASA rain gauge network (24 tipping bucket sensors, 0.1 mm resolution).
  • Drainage network incident reports from Barcelona Municipality and BCASA (2011–2022).
  • Arrival ATFM Delays at Josep Tarradellas Barcelona-El Prat Airport attributed to meteorological causes from EUROCONTROL (2014 onwards).
  • INUNGAMA flood database for Catalonia (1901–2022).

Main Results

• 45 intense rainfall episodes were analyzed, with a strong seasonal pattern showing most events concentrated between June and November (51% in September and October).
• A positive Pearson correlation was found between maximum DVIL (DVILmax) and maximum 1 min rainfall intensity (R² = 0.423, p < 0.001), and a slightly lower but significant correlation with maximum 5 min rainfall intensity (R² = 0.332).
• Segmented regression identified a break point in DVILmax at 1.14 g/m³ for 1 min intensity and 2.27 g/m³ for 5 min intensity, indicating a change in the relationship slope at higher values.
• 322 life-cycle storm cells were tracked using RaNDeVIL across 30 of the 45 selected days. Most cells were short-lived, with 32% decaying in under 12 min and 33% before 30 min. More persistent cells (>61 min) generally exceeded 3 g/m³ DVILmax.
• Seasonal variability showed most storm centroids detected during autumn and summer. July and August exhibited the highest median DVILmax (1.55 g/m³ and 1.45 g/m³) and Echo Top 12 dBZ (10.4 km and 10.8 km), and largest storm areas (July median ~40 km²).
• Diurnal patterns varied seasonally: summer convective centroids peaked between 03:00 and 13:00 UTC (absolute maximum at 10:00 UTC), while autumn showed two peaks, early morning (~05:00 UTC) and late afternoon (16:00–19:00 UTC).
• The RaNDeVIL nowcasting system for convective precipitation (5 min intensity > 35 mm/h) achieved a Probability of Detection (POD) of 0.728, a False Alarm Ratio (FAR) of 0.333, a Critical Success Index (CSI) of 0.534, and a Bias of 1.092.
• The median lead time for anticipating convective precipitation was 30 min. 39% of cases anticipated convective precipitation, 27% were late detections (no lead time), 19% were false alarms, and 15% were misses.

Contributions

• This is the first study to compare radar-derived VIL density (DVIL) fields with short-duration rainfall intensity from rain gauges, establishing a linear relationship and identifying a breakpoint at higher DVIL values.
• Provides an integrated characterization of intense convective storms in the Metropolitan Area of Barcelona using a multi-sensor approach (radar, rain gauges, and urban impact data).
• Evaluates the nowcasting potential of the RaNDeVIL algorithm for intense urban rainfall, offering specific performance metrics (POD, FAR, CSI, Bias) for the region.
• Details the spatiotemporal structure, seasonal and diurnal variability, persistence, and propagation characteristics of convective cells in a densely urbanized coastal environment.

Funding

• This research was carried out within the framework of the I-CHANGE project (Individual Change of HAbits Needed for Green European transition), from the European Union’s Horizon 2020 program.
• Institutional support was received from the Water Research Institute of the University of Barcelona (IdRA).
• The work was developed in collaboration with Barcelona Cicle de l’Aigua, S.A. (BCASA) and the Meteorological Service of Catalonia (SMC).

Citation

@article{Esbrí2025Intense,
  author = {Esbrí, Laura and Rigo, Tomeu and Llasat, María Carmen},
  title = {Intense Rainfall in Urban Areas: Characterization of High-Intensity Storms in the Metropolitan Area of Barcelona (2014–2022)},
  journal = {Atmosphere},
  year = {2025},
  doi = {10.3390/atmos17010041},
  url = {https://doi.org/10.3390/atmos17010041}
}