Rivoire et al. (2026) The future is in the past? A flexible resampling approach to generate multivariate time series

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Environmental Research Letters
• Year: 2026
• Date: 2026-03-13
• Authors: Pauline Rivoire, Tatjana Milojevic, Michael Lehning
• DOI: 10.1088/1748-9326/ae51af

Research Groups

Not explicitly specified in the abstract. The study utilizes data from the Swiss Alpine measurement station Adelboden and climate model projections from MeteoSwiss (CH2018).

Short Summary

This paper introduces a straightforward method for generating synthetic climate time series by constrained sampling of observations, demonstrating its ability to preserve physical consistency and multivariate dependencies while simulating multi-day extremes under future climate scenarios.

Objective

• To propose and demonstrate a straightforward method for generating synthetic meteorological time series based on constrained sampling of observations, ensuring physical consistency, preserving short temporal structure, and enabling exploration of future climate scenarios and potential extremes.

Study Configuration

• Spatial Scale: Point scale, specifically the Swiss Alpine measurement station Adelboden.
• Temporal Scale: Daily to multi-day (for extremes like heatwaves), based on sufficiently long observational time series.

Methodology and Data

• Models used: A straightforward method for time series generation based on constrained sampling of observations.
• Data sources: In-situ meteorological observations (temperature, precipitation, incoming solar radiation, wind speed) from Adelboden, Switzerland; climate model projections (MeteoSwiss: CH2018).

Main Results

• The proposed constrained sampling method effectively preserves the physical consistency and short-term temporal structure of observed meteorological variables.
• It maintains the multivariate dependence structure found in both historical data and climate projections.
• While not generating daily values outside the observed range, the method successfully simulates multi-day extremes, such as heatwaves, that can exceed historical records.

Contributions

• This study introduces a novel, straightforward, and flexible method for synthetic time series generation that explicitly preserves multivariate physical consistency and short-term temporal structures from observations.
• It uniquely demonstrates the ability to simulate multi-day extremes exceeding historical records, offering a valuable tool for climate impact studies, extreme event analysis, and downscaling tasks without relying on complex physical or statistical assumptions.

Funding

Not specified in the abstract.

Citation

@article{Rivoire2026future,
  author = {Rivoire, Pauline and Milojevic, Tatjana and Lehning, Michael},
  title = {The future is in the past? A flexible resampling approach to generate multivariate time series},
  journal = {Environmental Research Letters},
  year = {2026},
  doi = {10.1088/1748-9326/ae51af},
  url = {https://doi.org/10.1088/1748-9326/ae51af}
}