Monjo et al. (2023) Estimation of future extreme rainfall in Barcelona (Spain) under monofractal hypothesis

Identification

• Journal: International Journal of Climatology
• Year: 2023
• Date: 2023-03-29
• Authors: Robert Monjo, Luca Locatelli, John Milligan, Luis Torres, Marc Velasco, Emma Gaitán, Javier Pórtoles, Darío Redolat, Beniamino Russo, Jaime Ribalaygua
• DOI: 10.1002/joc.8072

Research Groups

• Climate Research Foundation – Fundación para la Investigación del Clima (FIC), Madrid, Spain
• Department of Algebra, Geometry and Topology, Complutense University of Madrid (UCM), Madrid, Spain
• AQUATEC (AGBAR Group), Barcelona, Spain
• Flumen Research Institute, Universitat Politècnica de Catalunya - BarcelonaTech (UPC), Barcelona, Spain

Short Summary

This study projects future extreme subdaily rainfall in Barcelona, Spain, by comparing two monofractal-based approaches using downscaled CMIP5 climate models, finding a consensual increase in both rainfall concentration (n-index) and absolute intensities for short-duration events by the end of the 21st century.

Objective

• To obtain change projections of intensity–duration–frequency (IDF) curves for subdaily precipitation events (from 5 minutes up to 2 hours) in Barcelona, comparing two approaches: (a) direct scaling of expected probable precipitation and (b) a new semi-stochastic approach combining daily climate model outputs with stochastic simulation of a concentration index (n-index).
• To analyze projected changes of the rainfall concentration n-index throughout the 21st century, used as an approximation of the simple-scaling exponent or monofractal parameter.

Study Configuration

• Spatial Scale: Barcelona metropolitan area and the Ter-Llobregat hydrological system (Catalonia, Northern Spain). The study used 27 subdaily rainfall stations around Barcelona and 300 daily rain gauges across the Ter-Llobregat system. The Jardí gauge of Fabra Observatory in Barcelona served as a long reference series.
• Temporal Scale:
  • Data Resolution: 5-minute, subdaily (up to 2 hours), daily, and supradaily (up to 5 days).
  • Historical Period: Observed data from 1927 (Fabra Observatory) or 1980–2021 (other stations); CMIP5 historical experiments for 1951–2005.
  • Future Projections: 2006–2100, analyzed in three 30-year periods: 2011–2040, 2041–2070, and 2071–2100.

Methodology and Data

• Models used:
  • Climate Models: 10 Coupled Model Intercomparison Project Phase 5 (CMIP5) models (ACCESS1-0, BCC-CSM1-1, CanESM2, CNRM-CM5, GFDL-ESM2M, HADGEM2-CC, MIROC-ESM-CHEM, MPI-ESM-MR, MRI-CGCM3, NorESM1-M) under two Representative Concentration Pathways (RCP4.5 and RCP8.5).
  • Downscaling:
    • Spatial Downscaling: Two-step analogue/transference approach linking surface point observations to large-scale predictor fields (wind speed/direction at 1000 hPa and 500 hPa, specific humidity at 700 hPa). Includes analogue stratification and model output statistics with quantile mapping bias correction.
    • Time Scaling (Subdaily Disaggregation):
      1. Direct Scaling: Applies the monofractal hypothesis to daily time series to derive Climate Change Factors (CCFs) for subdaily durations, assuming relative changes from supradaily scales apply to subdaily scales.
      2. Semi-stochastic Scaling: Generates subdaily time series by combining daily physical forcing from downscaled climate models with stochastic simulation based on the probability distribution of the concentration n-index and other observed rainfall features (wet spell duration, reference intensity).
  • Statistical Distributions: Gamma, Weibull, Gumbel, reversed Gumbel, log-logistic, and Generalized Extreme Value (GEV) distributions were fitted to empirical cumulative distribution functions (ECDFs) for IDF curve estimation.
  • Validation: Kolmogorov–Smirnov (KS) test for goodness-of-fit, cross-validation for direct scaling.
• Data sources:
  • Observed Rainfall:
    • 5-minute resolution records from Jardí gauge of Fabra Observatory (Barcelona) since 1927 (with gap filling for 1992-2008).
    • Subdaily precipitation records from 27 stations in the Barcelona area (BCASA, 1980–2021).
    • Daily precipitation records from 300 stations in the Ter-Llobregat hydrological system (Spanish State Meteorology Agency - AEMET, 1980–2021).
  • Reanalysis Data: ERA-Interim reanalysis for predictor variables in downscaling.

Main Results

• Validation:
  • Spatial downscaling of daily precipitation showed good performance, with systematic errors generally smaller than 10% for maximum daily precipitation and passing the KS test for most models after bias correction.
  • Direct time scaling introduced systematic errors generally smaller than 5% for rainfall durations less than 2 hours, increasing up to 10%–20% for longer durations (e.g., 360 minutes).
  • Semi-stochastic time scaling adequately simulated subdaily rainfall features (wet spell precipitation, n-index, reference intensities, wet/dry spells), passing the KS test for most cases.
• Projected Changes (2071–2100):
  • Rainfall Concentration (n-index): Projected to increase by up to 10% (from a median of 0.38 to 0.41), especially under the RCP8.5 scenario, indicating more concentrated rainfall events.
  • Extreme Rainfall Intensities:
    • Maximum intensities for 2-hour events with 2-year to 500-year return periods are projected to increase by approximately 20% (ranging from 15% to 30%).
    • 12-hour extreme rainfall is projected to increase by up to 30%–40% by 2071–2100, with statistical significance, particularly for 2-year and 10-year return periods.
  • Climate Change Factors (CCFs):
    • For the 2041–2070 period, a CCF of approximately 1.15 ± 0.15 is expected for all return periods (2 to 500 years) for 2-hour events.
    • By 2071–2100, CCFs for 1-hour extreme events can reach up to 1.5 for higher return periods.
• Uncertainty: Systematic errors of the methods (5%–10%) are smaller than the projected climate change signal (15%–30% variation range for CCF). Both emission scenarios (RCP4.5 and RCP8.5) show similar increases in rainfall concentration and maximum reference intensity by the end of the century.

Contributions

• Developed and compared two monofractality-based methods (direct scaling and a novel semi-stochastic approach) for projecting subdaily Intensity-Duration-Frequency (IDF) curves from daily climate model outputs, demonstrating their robustness and consistency for Mediterranean climates.
• Provided the first analysis of projected changes in the rainfall concentration n-index throughout the 21st century, linking it to the monofractal dimension and offering insights into the evolving temporal structure of subdaily rainfall events.
• Demonstrated that the systematic errors of the proposed methods are smaller than the projected climate change signal, providing robust projections for decision-making in stormwater management infrastructure design.
• The semi-stochastic technique offers the added value of simulating entire subdaily time series and statistics of general wet spells, which is beneficial for hydrological models requiring continuous subdaily inputs beyond just synthetic design storms.

Funding

• European Commission, Horizon 2020 Research and Innovation Program, Grant/Award Number: 700174 (RESilience to cope with Climate Change in Urban arEas - RESCCUE project)
• European Commission, Horizon Europe, Cluster 5—Climate, Energy and Mobility, EC Project Code: 101093806 (Improving ClimAte Resilience of crItical Assets - ICARIA project)
• Ministry for the Ecological Transition and the Demographic Challenge (MITECO) of the Spanish Government (IMpacts of climate change on wetlands Affected by GroUndwAter - IMAGUA project, two phases)

Citation

@article{Monjo2023Estimation,
  author = {Monjo, Robert and Locatelli, Luca and Milligan, John and Torres, Luis and Velasco, Marc and Gaitán, Emma and Pórtoles, Javier and Redolat, Darío and Russo, Beniamino and Ribalaygua, Jaime},
  title = {Estimation of future extreme rainfall in Barcelona (Spain) under monofractal hypothesis},
  journal = {International Journal of Climatology},
  year = {2023},
  doi = {10.1002/joc.8072},
  url = {https://doi.org/10.1002/joc.8072}
}