Islam et al. (2026) Dynamic and thermodynamic drivers of extreme precipitation under nonstationarity: implications for probable maximum precipitation across North America

Identification

• Journal: Natural Hazards
• Year: 2026
• Date: 2026-01-01
• Authors: Md. Robiul Islam, M. Reza Najafi
• DOI: 10.1007/s11069-025-07886-w

Research Groups

• Department of Civil and Environmental Engineering, Western University, London, ON, Canada

Short Summary

This study quantifies the contributions of dynamic and thermodynamic atmospheric drivers to extreme precipitation (EP) across North America, using both ERA5 reanalysis and gauge observations within a nonstationary generalized extreme value framework. It finds that moisture convergence (Vertically Integrated Moisture Divergence, VIMD) is the most influential driver for EP events, underscoring the necessity of incorporating multiple atmospheric variables beyond just precipitable water for accurate and reliable Probable Maximum Precipitation (PMP) estimation.

Objective

• Quantify the contributions of precipitable water (PW), convective available potential energy (CAPE), and vertically integrated moisture divergence (VIMD) to annual maximum precipitation (AMP) across North America.
• Assess the nonstationary behavior of AMP events using these drivers and their combinations as covariates within a nonstationary generalized extreme value (GEV) framework.
• Identify the dominant atmospheric drivers of extreme precipitation (EP) events occurring in different seasons.
• Evaluate spatial distributions of EP events with moisture maximization ratio (r) values greater than 2.0 and explore the dominant atmospheric mechanisms leading to higher r values.
• Examine inter-annual variability of extreme events and associated atmospheric variables, modeling nonstationarity using atmospheric variables as covariates.
• Validate the distribution and nonstationary behavior of dominant atmospheric drivers identified from reanalysis-based EP events using gauge-based EP observations.

Study Configuration

• Spatial Scale: North America, spanning 23° to 80°N latitude and 50° to 150°W longitude. ERA5 data on a 0.25-degree grid. 6018 Global Historical Climatology Network-Daily (GHCN-D) gauge stations.
• Temporal Scale: Daily time scale for atmospheric variables and precipitation over the 1940–2021 period (82 years). Annual maximum precipitation (AMP) events are extracted.

Methodology and Data

• Models used:
  • Nonstationary Generalized Extreme Value (GEV) framework, with the location parameter modeled as a linear function of atmospheric covariates.
  • Block maxima method for extracting Annual Maximum Precipitation (AMP) events.
  • Mann–Kendall test for long-term monotonic trend analysis.
  • Theil-Sen estimator for trend magnitude estimation.
  • Maximum likelihood estimation for GEV distribution parameters.
  • Akaike information criterion (AIC) for model selection.
  • Spearman rank correlation coefficients for assessing relationships between EP and atmospheric variables.
  • Moisture maximization ratio (r) calculation based on PW.
• Data sources:
  • ERA5 reanalysis data (European Centre for Medium-Range Weather Forecasts): Hourly estimates of various climate variables on a 0.25-degree grid from 1940 to 2021. Key atmospheric variables include Precipitable Water (PW), Convective Available Potential Energy (CAPE), Vertically Integrated Moisture Divergence (VIMD), and Vertical Velocity (VV at 700 hPa).
  • Global Historical Climatology Network-Daily (GHCN-D) dataset: Daily precipitation records from 6018 land-based stations across North America (1940–2021, with stations having at least 50 years of data and 90% completeness).
  • Extreme Precipitation (EP) events are defined as the top 50 daily liquid precipitation totals at each ERA5 grid cell or gauge station over the 1940–2021 period.

Main Results

• Vertically Integrated Moisture Divergence (VIMD), representing moisture convergence, is the most influential driver for extreme precipitation (EP) events across 46% of grid cells in North America.
• Precipitable Water (PW) dominates EP in coastal and moisture-rich regions (e.g., Gulf of Mexico, Great Lakes, West Coast), while Convective Available Potential Energy (CAPE) plays a significant role in southwestern regions characterized by atmospheric instability.
• Seasonal analysis indicates EP events are most prevalent in spring and fall, with orographic regions in western and northeastern North America experiencing the highest concentration of extreme values.
• A parallel analysis using 6018 gauge stations confirms the spatial patterns and dominant atmospheric drivers identified with ERA5, but reveals greater spatial heterogeneity and more frequent selection of complex multi-variable models, suggesting station observations better capture localized atmospheric variability.
• EP events with a moisture maximization ratio (r) exceeding 2.0 are concentrated in mountainous regions of western, northern, and northeastern North America, often associated with orographic effects. These higher r values are more frequent in fall and spring.
• EP events influenced predominantly by PW rarely exceed an r threshold of 2.0, whereas those influenced by VIMD or Vertical Velocity (VV) frequently do, highlighting the critical role of dynamic factors beyond just moisture availability.
• Positive trends in annual maximum precipitation (AMP) are primarily observed in Eastern North America and parts of central Canada, indicating an intensification of EP events, which strongly correlate with increasing trends in PW and VIMD.
• The nonstationary behavior of AMP events is effectively captured by a nonstationary GEV model, with VIMD emerging as the most influential covariate for nearly half (46.1%) of grid cells. Models combining PW and VIMD (21.7%), or PW, CAPE, and VIMD (14.1%, particularly in Western North America), also show significant explanatory power.

Contributions

• Provides a comprehensive, continent-wide assessment of the combined roles of dynamic (VIMD, VV) and thermodynamic (PW, CAPE) drivers in shaping extreme precipitation events across diverse geographic and seasonal contexts.
• Validates reanalysis-based findings against a large network of 6018 gauge observations, addressing a critical gap in previous studies that relied solely on model-derived precipitation.
• Integrates these drivers into a nonstationary generalized extreme value (GEV) framework to enhance the accuracy and applicability of Probable Maximum Precipitation (PMP) estimation under evolving climatic conditions.
• Offers insights into why, where, and when the moisture maximization ratio (r) warrants a pre-defined threshold, moving beyond single-variable approaches and subjective judgments in PMP estimation.
• Provides critical groundwork to bridge existing PMP methods with numerical weather prediction (NWP)-based approaches, aligning with NASEM (2024) recommendations for reducing subjectivity and improving objectivity in PMP estimation.

Funding

• National Research Council Canada through the Climate Resilient Built Environment (CRBE) initiative funded by Infrastructure Canada (INFC).
• Dr. M. Reza Najafi.

Citation

@article{Islam2026Dynamic,
  author = {Islam, Md. Robiul and Najafi, M. Reza},
  title = {Dynamic and thermodynamic drivers of extreme precipitation under nonstationarity: implications for probable maximum precipitation across North America},
  journal = {Natural Hazards},
  year = {2026},
  doi = {10.1007/s11069-025-07886-w},
  url = {https://doi.org/10.1007/s11069-025-07886-w}
}