An et al. (2025) Future projections of wet and dry spells in southern Sweden: The impact of climate model resolution

Identification

• Journal: Atmospheric Research
• Year: 2025
• Date: 2025-10-08
• Authors: Dong An, Jonas Olsson, Yiheng Du, Johanna Sörensen, Cíntia Bertacchi Uvo, Peter Berg
• DOI: 10.1016/j.atmosres.2025.108547

Research Groups

• Division of Water Resources Engineering, Faculty of Engineering (LTH), Lund University, Lund, Sweden
• Hydrology Research Unit, Swedish Meteorological and Hydrological Institute, Norrk¨oping, Sweden
• United Nations University Hub on Water in a Changing Environment (WICE) at Lund University, United Nations University Institute for Water, Environment and Health (UNU-INWEH), Lund, Sweden
• Finnish Environment Institute, Helsinki, Finland
• NorCP (Nordic Convection Permitting Climate Projections) project group (Danish Meteorological Institute (DMI), Finnish Meteorological Institute (FMI), Norwegian Meteorological Institute (MET Norway), Swedish Meteorological and Hydrological Institute (SMHI))
• European Centre for Medium-Range Weather Forecasts (ECMWF)
• Swedish National Supercomputing Center (NSC), Link¨oping University

Short Summary

This study evaluates the performance of five Regional Climate Models (RCMs) with resolutions from 44 km to 3 km in reproducing historical wet and dry spells in southern Sweden and projects future changes under RCP8.5. It finds that the convection-permitting model (CPM) AROME (3 km) outperforms coarser RCMs, projecting increases in wet spells and extreme precipitation intensities, while dry spells show more modest, non-monotonic changes in duration.

Objective

• How does RCM resolution, particularly convection-permitting models (CPMs) and non-convection-permitting models (non-CPMs), influence the accuracy of wet and dry spell representations during the historical period?
• How consistent are the projected future changes in wet and dry spells, including extremes, across CPMs and non-CPMs at different resolutions?

Study Configuration

• Spatial Scale: Six observational stations in Scania, southern Sweden. Model grid resolutions ranging from approximately 3 km to 50 km.
• Temporal Scale: Historical period: 1996–2005 (observations), 1986–2005 (model baseline P1). Future periods: 2041–2060 (P2) and 2081–2100 (P3). Observational data at 15-minute time-steps; model outputs at 15-minute or 1-hour time-steps. Wet spells analyzed at hourly scale, dry spells at daily scale.

Methodology and Data

• Models used:
  • Global Climate Model (GCM): EC-Earth member r12i1p1
  • Regional Climate Models (RCMs) for downscaling:
    • HCLIM38-AROME (AROME, ~3 km resolution) - Convection-permitting model
    • HCLIM38-AROME (AROME4x4, ~12 km resolution) - Aggregated AROME
    • HCLIM38-ALADIN (ALADIN, ~12 km resolution) - Parameterized convection
    • RCA4–011 degree (RCA011, ~12.5 km resolution)
    • RCA4–022 degree (RCA022, ~25 km resolution)
    • RCA4–044 degree (RCA044, ~50 km resolution)
• Data sources:
  • Local precipitation observations: 15-minute time-step data from six GEONOR automated precipitation gauges (SMHI) in southern Sweden (1996–2005).
  • Regional Climate Model projections: GCM-driven RCM simulations under Representative Concentration Pathway (RCP) RCP8.5.
  • Extreme value analysis: Generalized Extreme Value (GEV) distribution fitted to annual maxima of precipitation and annual maximum consecutive dry days.
  • Climate factors: Wet Climate Factor (WCF) and Dry Climate Factor (DCF) calculated as ratios of future to historical return levels.

Main Results

• Historical Performance (1996–2005):
  • The convection-permitting model (CPM) AROME (3 km) consistently provides the most accurate representation of both wet and dry spell characteristics, particularly for sub-daily extremes (≤3 hours).
  • Higher resolution generally improves the representation of dry spell characteristics, while wet spell results vary more strongly with resolution.
  • Biases generally increase with decreasing model resolution. Coarser models tend to overestimate the duration of wet events and the number of dry events, while underestimating their average and maximum lengths.
  • Models generally perform better in reproducing precipitation maxima at accumulation durations of 6 hours or longer; biases enlarge with decreasing spatial resolution.
• Future Projections (RCP8.5, P2: 2041–2060, P3: 2081–2100):
  • Wet Spells: All models project a gradual increase in average daily precipitation (8–12 % by P3). The number of wet events, their average depth, and duration are also projected to increase to varying degrees.
  • Extreme Precipitation (WCF): AROME and ALADIN project higher Wet Climate Factors (WCFs) for extreme precipitation, especially at shorter durations (1–3 hours), with AROME showing median WCFs between 1.2 and 1.3 in P2, and exceeding 1.3 in P3. Coarser RCMs project smaller increases.
  • Dry Spells: The annual number of dry days shows a slight increase (0–3 % by P3). The number of dry events is projected to decrease, while both the average duration and the annual maximum consecutive dry days are projected to increase.
  • Extreme Dry Spells (DCF): The Dry Climate Factor (DCF) indicates a decrease in the length of extreme dry spells in the near future (P2/P1 DCF 0.8–1.0), followed by a slight increase by the end of the century (P3/P1 DCF 0.9–1.3). DCF shows less dependence on resolution, but AROME exhibits the most consistent behavior across return periods.

Contributions

• Provides a comprehensive multi-resolution analysis of both wet and dry spells, including extremes, using five RCMs (from 44 km to 3 km) driven by the same GCM under RCP8.5 for southern Sweden.
• Confirms the superior performance and improved reliability of convection-permitting models (CPMs), specifically AROME, in reproducing historical wet and dry spell characteristics and projecting future variability, particularly for short-duration precipitation extremes.
• Introduces and applies the Dry Climate Factor (DCF) to quantify projected changes in extreme dry spells, complementing the established Wet Climate Factor (WCF).
• Offers valuable insights for regional climate services and adaptation planning by clarifying the added value and limitations of moving towards convection-permitting scales compared to existing coarser resolution standards.

Funding

• GlobalHydroPressure (EU Water JPI through the Swedish Research Council Formas, Grant no. 2018-02379)
• EDUCAS (Swedish Research Council Formas, Grant no. 2019-00829)
• Digital Waters Flagship project (DIWA Flagship) (Research Council of Finland)

Citation

@article{An2025Future,
  author = {An, Dong and Olsson, Jonas and Du, Yiheng and Sörensen, Johanna and Uvo, Cíntia Bertacchi and Berg, Peter},
  title = {Future projections of wet and dry spells in southern Sweden: The impact of climate model resolution},
  journal = {Atmospheric Research},
  year = {2025},
  doi = {10.1016/j.atmosres.2025.108547},
  url = {https://doi.org/10.1016/j.atmosres.2025.108547}
}