Soto-Escobar et al. (2026) Developing Intensity-Duration-Frequency (IDF) curves using sub-daily gridded and in situ datasets: characterising precipitation extremes in a drying climate

Identification

• Journal: Hydrology and earth system sciences
• Year: 2026
• Date: 2026-01-12
• Authors: Cristóbal Soto-Escobar, Mauricio Zambrano‐Bigiarini, Violeta Tolorza, René Garreaud
• DOI: 10.5194/hess-30-91-2026

Research Groups

• ICASS, Santiago, Chile
• Department of Civil Engineering, Universidad de la Frontera, Temuco, Chile
• Center for Climate and Resilience Research, Universidad de Chile, Santiago, Chile
• Department of Geophysics, Universidad de Chile, Santiago, Chile

Short Summary

This study investigates the influence of stationary assumptions and data length on annual maximum precipitation intensities (Imax) in continental Chile using sub-daily gridded and in situ datasets. It reveals significant decreasing trends in Imax across Central Chile, with non-stationary models generally yielding slightly lower intensities, though the impact of data length on Imax values is minor.

Objective

• To understand how the stationary assumption and data length of hourly precipitation data influence the annual maximum intensities of precipitation events in continental Chile.
• To determine the spatial distribution of stationary annual maximum precipitation intensities for different durations.
• To assess if significant trends in stationary annual maximum precipitation intensities justify using non-stationary Intensity-Duration-Frequency (IDF) curves.
• To evaluate the significance of differences between stationary and non-stationary IDF curves.
• To quantify the impact of typical data length of precipitation products on estimating stationary and non-stationary IDF curves.

Study Configuration

• Spatial Scale: Continental Chile (17.5–56.5° S, approximately 76° W to 66° W), covering approximately 756,626 km². The area is divided into five macro-climatic zones. Gridded data resolutions range from 0.1° to 0.25° (approximately 8 km to 28 km).
• Temporal Scale:
  • Main analysis period: 2001–2021 (20-21 years) for all precipitation products.
  • Extended analysis period for ERA5 and ERA5-Land: 1981–2021 (40 years).
  • Precipitation durations: 1, 2, 4, 6, 8, 10, 12, 18, 24, 48, and 72 hours.
  • Return periods: 2, 5, 10, 25, 50, and 100 years.
  • Data frequency: Hourly (some aggregated from half-hourly products).

Methodology and Data

• Models used:
  • Stationary Gumbel distribution (a two-parameter Generalized Extreme Value (GEV) distribution with shape parameter ξ=0) for Intensity-Duration-Frequency (IDF) curve development.
  • Non-stationary Gumbel distribution, where the location parameter (μ) is modeled as a linear function of time, while the scale parameter (σ) remains constant.
  • Modified Mann–Kendall test for identifying monotonic trends in annual maximum intensities, adjusted for autocorrelation.
  • Thin plate spline interpolation for spatially continuous bias-correction factors.
• Data sources:
  • In situ: 161 quality-checked hourly rain gauges from the General Directorate of Water (DGA), Meteorological Directorate of Chile (DMC), Agrometeorological Network of the Agricultural Research Institute (Agromet), and Center for Advanced Studies in Arid Zones (CEAZA). Data from 2000 to 2021, with a selection criterion of >80% hourly data availability between 1 January 2013 and 31 December 2017.
  • Gridded precipitation datasets:
    • IMERGv06B Final Run (Integrated Multi-satellitE Retrievals for GPM): 30-minute, 0.1° resolution, 2001-2020.
    • IMERGv07B Final Run: 30-minute, 0.1° resolution, 2001-2021.
    • ERA5 (ECMWF Reanalysis v5): Hourly, 0.25° resolution, 1981-2021.
    • ERA5-Land: Hourly, 0.1° resolution, 1981-2021.
    • CMORPH-CDR (Climate Prediction Center morphing technique - Climate Data Record): Half-hourly, 8 km resolution, 2001-2021.
  • Ancillary data: Digital elevation model, mean annual precipitation, and Köppen–Geiger climate classification maps.

Main Results

• Bias in Annual Maximum Intensities (Imax): IMERG products generally overestimated Imax for short durations (1-6 hours) with median bias correction factors between 0.65 and 0.82 mm h−1, improving for longer durations (0.92-0.98 mm h−1 for 24-72 hours). ERA5 and ERA5-Land underestimated Imax for short durations (1-6 hours, median factors 1.16-1.49 mm h−1) and slightly overestimated for longer durations (0.83-0.95 mm h−1 for 24-72 hours). CMORPH-CDR showed mixed biases depending on the macroclimatic zone and duration. Bias variability was highest in the Far North and Far South of Chile.
• Trends in Imax: Significant decreasing trends in Imax were observed across Chile for CMORPH-CDR. ERA5 and ERA5-Land showed decreasing trends in Central-Southern Chile (32–43° S). IMERGv06B and IMERGv07B exhibited isolated, divergent trends (some increasing in the Near North and Far South, decreasing in Central Chile). These decreasing trends are consistent with observed drying trends and fewer winter storms in Central Chile.
• Spatial Distribution of Imax: The spatial distribution of Imax differs significantly from the mean annual precipitation pattern in continental Chile. While mean annual precipitation increases steadily southward, Imax reaches its maximum values in Central-Southern Chile (32.2–43.7° S) for all durations. For durations of 24 hours or more, the highest intensities are primarily found in the Andes, particularly between the Maule and Araucanía regions (35–40° S). A strong longitudinal gradient in Imax was observed, with lower values in the intermediate depression and higher values in the Andes, increasing with longer durations.
• Comparison of Stationary vs. Non-stationary Imax: Non-stationary Imax values were generally slightly lower than their stationary equivalents (differences in the range of 0 to 5 mm h−1). These differences became smaller for longer durations (greater than 8 hours). Median differences between the two approaches were close to zero (0 to 2 mm h−1) for most datasets and durations, with decreasing dispersion for increasing durations. Locations with statistically significant trends in Imax did not necessarily exhibit significant differences between stationary and non-stationary Imax.
• Impact of Data Length: Comparing 20-year (2001–2021) and 40-year (1981–2021) records for ERA5 and ERA5-Land revealed only minor differences in the resulting Imax values (median differences below 1 mm h−1) for both stationary and non-stationary models across all durations and macroclimatic zones. While Gumbel parameters showed larger percentage differences (e.g., scale parameter up to ±40%), these did not translate into substantial differences in Imax.
• Dataset Agreement: IMERGv07B, ERA5, and ERA5-Land showed strong agreement in spatial distribution and Imax values. IMERGv06B exhibited higher Imax values, and CMORPH-CDR the highest, especially in the Andes. CMORPH-CDR and IMERGv06B also presented the largest number and magnitude of outliers.

Contributions

• First comprehensive comparison of annual maximum precipitation intensities derived from stationary and non-stationary statistical models using two distinct families of state-of-the-art gridded precipitation datasets (IMERGv06B/IMERGv07B and ERA5/ERA5-Land).
• Pioneering study to provide high spatial and temporal resolution Intensity-Duration-Frequency (IDF) curves for continental Chile using state-of-the-art gridded precipitation datasets, significantly advancing knowledge of extreme precipitation in mountainous regions.
• Demonstrated that the spatial distribution of annual maximum intensities (Imax) in continental Chile fundamentally differs from the mean annual precipitation pattern, highlighting critical implications for civil infrastructure design and flood risk assessment.
• Provided evidence that, despite observed decreasing trends in Imax in Central Chile, the differences between stationary and non-stationary IDF curves are generally small, suggesting that for the Chilean context, the choice of model may not significantly alter Imax estimations.
• Developed and made publicly available a user-friendly web platform (https://curvasIDF.cl/) to disseminate the generated IDF curves, offering practical guidance for hydrological and engineering applications in Chile.

Funding

• Chilean Agencia Nacional de Investigación y Desarrollo (ANID)
  • ANID-Fondecyt Regular 1212071
  • ANID-PCI NSFC190018
  • ANID-Fondecyt Iniciación 11190864
  • UFRO Postdoctoral grant VRIP20P001
  • ANID-FONDAP 1523A0002

Citation

@article{SotoEscobar2026Developing,
  author = {Soto-Escobar, Cristóbal and Zambrano‐Bigiarini, Mauricio and Tolorza, Violeta and Garreaud, René},
  title = {Developing Intensity-Duration-Frequency (IDF) curves using sub-daily gridded and in situ datasets: characterising precipitation extremes in a drying climate},
  journal = {Hydrology and earth system sciences},
  year = {2026},
  doi = {10.5194/hess-30-91-2026},
  url = {https://doi.org/10.5194/hess-30-91-2026}
}