Chauveau et al. (2026) A Köppen–Geiger classification derivative tailored for numerical modeling of ecosystems within watershed dynamics

Identification

• Journal: The Science of The Total Environment
• Year: 2026
• Date: 2026-01-22
• Authors: Benoît Chauveau, Arnaud Pujol, Ahmed Mama
• DOI: 10.1016/j.scitotenv.2026.181445

Research Groups

• IFP Energies Nouvelles, Rueil-Malmaison, France

Short Summary

This study develops a novel framework to generate climate scenarios for hydro-environmental modeling by subdividing Köppen–Geiger classes into 108 subclasses based on distinct climatic curve patterns, further organizing them into 8 macro-environmental groups, and demonstrates its applicability in simulating lake dynamics.

Objective

• To develop a reproducible framework for generating realistic climatic scenarios suitable for hydro-environmental and ecosystem modeling, particularly in contexts where direct climate observations are sparse or unavailable (e.g., past or future periods).
• To extend the applicability of the Köppen–Geiger climate classification by associating each class with a set of representative multivariate climatic patterns and scenario-generation tools.

Study Configuration

• Spatial Scale: Global coverage, with 48,910 locations sampled. Data resolutions varied: 0.1° for precipitation, 0.5° x 0.625° for most meteorological variables, and 1° for solar radiation and MODIS cloud cover. A case study was performed on Lake Mendota, USA.
• Temporal Scale: Climate data collected for the 1991–2020 period (30 years for most variables, 2001–2020 for MODIS cloud cover). Data were sampled at a daily time step, aggregated into twelve-month representations for pattern identification. The Lake Mendota simulation spanned a 37-year period (1979–2015).

Methodology and Data

• Models used:
  • General Lake Model (GLM, version 3.3.0)
  • Aquatic EcoDynamics (AED2) library (coupled with GLM)
  • Statistical methods: Principal Component Analysis (PCA), k-means clustering, hierarchical clustering, elbow method, Silhouette Index, Gap Statistic.
• Data sources:
  • NASA POWER Project (2025) database (daily/hourly air temperature, precipitation, solar radiation, relative humidity, wind speed).
  • MODIS satellite products (Level-3 MODIS Atmosphere Eight-Day Global Product for cloud cover).
  • ERA5 reanalysis dataset (Hersbach et al., 2023) (for cloud cover comparison).
  • Beck et al. (2018, 2023) Köppen–Geiger climate classification maps (reference for present-day and future climate zones).
  • North American Land Data Assimilation System Phase 2 (NLDAS-2) (hourly meteorological forcing for reference Lake Mendota study).
  • United States Geological Survey (USGS) (in-situ streamflow measurements for calibration in reference study).

Main Results

• Clustering methods identified 108 distinct climatic subclasses within the Köppen–Geiger classification, each characterized by unique annualized temporal patterns of six key climatic variables (air temperature, precipitation, solar radiation, wind speed, relative humidity, and cloud cover).
• A complementary hierarchical clustering analysis reorganized these 108 subclasses into 8 macro-environmental groups (e.g., high-altitude continental, monsoon-like tropical, oceanic, Mediterranean), providing a broader context for scenario selection.
• The developed framework enables the generation of synthetic multivariate annual climate scenarios that satisfy user-defined annual mean constraints while preserving seasonal structure and cross-variable dependencies.
• A case study on Lake Mendota demonstrated that the geographically relevant subclass (Dfa.3) combined with the inclusion of a diurnal cycle in shortwave radiation (Dfa.3, HF scenario) accurately reproduced observed lake thermal stratification, dissolved oxygen profiles, and seasonal gross primary productivity (GPP).
• Simulations without high-frequency diurnal shortwave radiation (Dfa.3, LF scenario) underestimated surface temperatures and produced deviations in summer temperature and oxygen profiles, highlighting the importance of sub-daily variability.
• The NASA POWER dataset, when augmented with a diurnal shortwave cycle, proved to be a robust alternative for lake modeling, reproducing reference GPP dynamics with a low annual error (+2.97%).

Contributions

• Proposes a novel, reproducible framework that extends the Köppen–Geiger climate classification by linking each class to representative, multivariate climatic patterns, making it directly compatible with hydro-environmental modeling requirements.
• Identifies and quantifies significant intra-class climatic variability within Köppen–Geiger classes, leading to the definition of 108 data-driven subclasses.
• Develops a flexible statistical framework for generating consistent multivariate annual climate scenarios from these subclasses, allowing for user-defined constraints and preserving empirical inter-variable relationships.
• Introduces a transversal classification of subclasses into 8 macro-environmental groups, offering a pragmatic approach to guide scenario selection in data-sparse or paleoenvironmental contexts.
• Demonstrates the critical importance of both appropriate climatic subclass selection and the representation of high-frequency (e.g., diurnal) radiative forcing for accurate simulations of lake thermal structure and primary productivity.
• Makes the derived subclasses and the Python scripts for climate scenario generation openly available via a Zenodo repository (DOI: 10.5281/zenodo.16419447).

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Citation

@article{Chauveau2026KöppenGeiger,
  author = {Chauveau, Benoît and Pujol, Arnaud and Mama, Ahmed},
  title = {A Köppen–Geiger classification derivative tailored for numerical modeling of ecosystems within watershed dynamics},
  journal = {The Science of The Total Environment},
  year = {2026},
  doi = {10.1016/j.scitotenv.2026.181445},
  url = {https://doi.org/10.1016/j.scitotenv.2026.181445}
}