Domínguez et al. (2013) Present-climate precipitation and temperature extremes over Spain from a set of high resolution RCMs

Identification

• Journal: Climate Research
• Year: 2013
• Date: 2013-09-11
• Authors: Marta Domínguez, Raquel Romera, Enrique Sánchez, Lluís Fita, Jesús Fernández, Pedro Jiménez‐Guerrero, Juan Pedro Montávez, WD Cabos, Giovanni Liguori, MÁ Gaertner
• DOI: 10.3354/cr01186

Research Groups

• Universidad Politécnica de Cartagena, Grupo de I+D+i Gestión de Recursos Hídricos, Unidad Predepartamental de Ingeniería Civil, Cartagena, Spain.
• Pontificia Universidad Javeriana, Departamento de Ingeniería Civil, Bogotá, Colombia.

Short Summary

This study develops and evaluates innovative multi-model ensemble methodologies for Regional Climate Models (RCMs) to reduce uncertainty in rainfall projections for Spain, identifying the most robust method and projecting a widespread decrease in mean annual rainfall for 2021-2050.

Objective

• To improve the reliability of precipitation time series projections in Spain by developing and evaluating innovative RCM ensemble methodologies.

Study Configuration

• Spatial Scale: Peninsular Spain, with observed data at 20 km x 20 km resolution and RCM data at 25 km x 25 km resolution.
• Temporal Scale: Reference period 1961-1990; projection period 2021-2050.

Methodology and Data

• Models used: Regional Climate Models (RCMs) from the ENSEMBLES project (17 selected models). A modified Reliability Ensemble Averaging (REA) method was used for ensemble construction.
• Data sources:
  • Observed precipitation data: Spain02 dataset (1950-2007).
  • RCM data: ENSEMBLES project (A1B emissions scenario, 1961-2050).
  • Metrics for goodness of fit: Smirnov-Kolmogorov (SK) test p-value and Perkins' Sscore.

Main Results

• The sensitivity analysis revealed that the choice of metric was the key parameter, with the precipitation ensemble being significantly more sensitive to it.
• The Smirnov-Kolmogorov (SK) test p-value metric yielded better results for ensemble performance compared to the Sscore.
• Using seasonal and annual cumulative distribution functions (Restacional) for the reliability factor provided better ensemble results than monthly CDFs (Rmensual).
• The combination of the SK test p-value metric and the seasonal/annual CDF scheme (referred to as R1) was identified as the most robust method for constructing probabilistic precipitation ensembles for the 1961-1990 reference period.
• Projections for the 2021-2050 period indicate a plausible and significant decrease in mean annual precipitation across almost all of Peninsular Spain, with the largest reductions anticipated in the southwestern region.

Contributions

• Development of innovative methodologies for multi-model RCM ensembles, including modifications to the original REA method to enhance model weighting and improve reliability.
• Probabilistic assessment of the influence of different statistical metrics and reliability factor formulations on rainfall projections for the Iberian Peninsula.
• Identification of the most robust ensemble construction method, providing a more confident approach to RCM-projections for spatio-temporal rainfall changes over extended regions.

Funding

• Project: CGL2012-39895-C02-01 Evaluación de variabilidad hidroclimática desde combinaciones multimodelo climáticas regionales (HYDROCLIM).
• Funded by: Secretaría de Estado de Investigación del Ministerio de Economía y Competitividad (Spain) and FEDER funds.

Citation

@article{Domínguez2013Presentclimate,
  author = {Domínguez, Marta and Romera, Raquel and Sánchez, Enrique and Fita, Lluís and Fernández, Jesús and Jiménez‐Guerrero, Pedro and Montávez, Juan Pedro and Cabos, WD and Liguori, Giovanni and Gaertner, MÁ},
  title = {Present-climate precipitation and temperature extremes over Spain from a set of high resolution RCMs},
  journal = {Climate Research},
  year = {2013},
  doi = {10.3354/cr01186},
  url = {https://doi.org/10.3354/cr01186}
}