Khedhaouiria et al. (2026) The Canadian Surface Reanalysis (CaSR) v3.2 precipitation dataset: a 45-year high-resolution analysis for North America (1980–2024)

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-09-24
• Authors: Dikra Khedhaouiria, Nicolas Gasset, Vincent Fortin, Milena Dimitrijevic, Maxim Bulat, Xihong Wang
• DOI: 10.5194/hess-30-5971-2026

Research Groups

• Numerical Modelling and Prediction Research Division, Environment and Climate Change Canada, Dorval, QC, Canada
• Meteorological Service of Canada, Environment and Climate Change Canada, Dorval, QC, Canada

Short Summary

This study introduces and evaluates the Canadian Surface Reanalysis (CaSR) v3.2 precipitation dataset, a 45-year high-resolution product for North America that integrates surface observations with a numerical weather prediction background. The new version demonstrates substantial improvements in precipitation estimates, particularly in data-sparse regions, offering a reliable resource for hydrological and climatological studies.

Objective

• To document the configuration and evaluate the performance of the Canadian Surface Reanalysis (CaSR) v3.2 offline precipitation component (CaPA-24h) for North America.
• To compare CaPA-24h v3.2 against its predecessor (v2.1) and independent datasets (ERA5-Land, PRISM) using various skill scores, high-impact precipitation metrics, and diurnal cycle representation.

Study Configuration

• Spatial Scale: North America, with a horizontal resolution of approximately 10 kilometers (km).
• Temporal Scale: 45-year period (1980–2024), providing daily 24-hour accumulated precipitation and derived hourly precipitation reanalysis.

Methodology and Data

• Models used:
  • Canadian Surface Reanalysis (CaSR) system (online and offline components)
  • Canadian Precipitation Analysis (CaPA) system (CaPA-24h, CaPA-6h)
  • Regional Deterministic Reforecast System (RDRS)
  • Canadian Land Data Assimilation System (CaLDAS)
  • Global Environmental Multiscale (GEM) model
  • Optimal Interpolation (OI) algorithm for precipitation assimilation
• Data sources:
  • Assimilation: Surface station observations from Canadian and United States networks, Integrated Surface Database (ISD), Environment and Climate Change Canada (ECCC) operational archives, Adjusted Daily Rainfall and Snowfall (AdjDlyRS), Adjusted Hourly Rain and Snow (AdjHlyRS).
  • Evaluation: Leave-one-out validation using surface station observations, ERA5-Land (ECMWF's fifth-generation reanalysis land component), PRISM (Parameter-elevation Regressions on Independent Slopes Model).
  • Near-real-time extension (operational CaPA-RDPA): Radar Quantitative Precipitation Estimation (QPE), IMERG precipitation estimates.

Main Results

• CaPA-24h v3.2 shows substantial improvements over v2.1, especially in data-sparse regions, with reduced frequency bias and enhanced representation of precipitation events across different intensities.
• CaPA-24h v3.2 provides more accurate seasonal and regional precipitation patterns compared to ERA5-Land, and closer agreement with PRISM, particularly in the eastern contiguous United States (CONUS).
• Biases persist in southern and western mountainous areas, especially for orographic precipitation, for both CaSR and ERA5-Land.
• Decomposition of seasonal precipitation bias reveals that differences in total accumulation are primarily driven by errors in wet-day frequency rather than systematic intensity errors.
• For extreme precipitation indices (e.g., annual daily maximum precipitation, Rx1day), CaPA-24h v3.2 generally achieves higher Kling–Gupta Efficiency (KGE) values (0.7–0.9) than ERA5-Land in the eastern CONUS.
• The hourly disaggregated product captures the expected seasonal contrast in diurnal variability, reproducing late-afternoon peaks for warm-season precipitation. However, non-physical peaks at synoptic hours (06:00, 12:00, 18:00 UTC) are present due to the current lead-time stitching and rescaling strategy.
• The operational CaPA-RDPA is broadly consistent with CaSR v3.2 at climatological scales, making it suitable for extending time series for climate-oriented applications, but caution is advised for weather-scale or event-based analyses.

Contributions

• Development and comprehensive evaluation of a new 45-year (1980–2024) high-resolution (approximately 10 km) gridded precipitation reanalysis (CaSR v3.2) for North America.
• Demonstrates significant improvements in precipitation estimates over previous versions and other widely used reanalysis products, particularly in regions with limited observational data.
• Provides a first-time assessment of the hourly disaggregated precipitation product, identifying both strengths in capturing diurnal cycles and limitations related to disaggregation methodology.
• Offers a valuable, reliable, and well-established dataset for hydrological, climatological, and impact studies across North America.
• Introduces updated quality control and assimilation procedures, along with a denser and better-curated gauge archive.

Funding

• International Joint Commission, through its International Watersheds Initiative.

Citation

@article{Khedhaouiria2026Canadian,
  author = {Khedhaouiria, Dikra and Gasset, Nicolas and Fortin, Vincent and Dimitrijevic, Milena and Bulat, Maxim and Wang, Xihong},
  title = {The Canadian Surface Reanalysis (CaSR) v3.2 precipitation dataset: a 45-year high-resolution analysis for North America (1980–2024)},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-5971-2026},
  url = {https://doi.org/10.5194/hess-30-5971-2026}
}