Fan et al. (2026) When the Past Matters: How Model Initialization Can Lead to Surprises in Long‐Term Simulations in Glaciated Environments

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

Identification

• Journal: Hydrological Processes
• Year: 2026
• Date: 2026-04-01
• Authors: Xinyang Fan, Bettina Schaefli, Florentin Hofmeister, Gabriele Chiogna
• DOI: 10.1002/hyp.70507

Research Groups

Not specified in abstract.

Short Summary

This paper highlights that improper initialization of surface and subsurface water storage and the inability of hydrological models to account for landscape evolution lead to significant errors in long-term distributed streamflow simulations, recommending long exploratory runs to steady state.

Objective

• To investigate and emphasize the critical role of model initialization (e.g., surface and subsurface water storage, glacier area/volume, soil moisture) and the evolution of model topography on the accuracy of distributed long-term hydrological simulation results, particularly in non-stationary systems.

Study Configuration

• Spatial Scale: Catchment scale (specifically, a high Alpine catchment in Austria).
• Temporal Scale: Long-term (decadal to centennial, with recommended runs > 100 years).

Methodology and Data

• Models used: Hydrological models; Glacier models (implied for coupling). Specific model names are not provided.
• Data sources: Not specified in abstract.

Main Results

• Improper initialization of surface and subsurface water storage (e.g., glacier area and volume, soil moisture content) causes significant errors in streamflow simulation.
• The inability of hydrological models to account for landscape evolution (e.g., evolving topography) leads to significant errors in streamflow simulation.
• External coupling between distributed models (e.g., glacier and hydrological models) can be inadequate for non-stationary systems due to initial-condition dependence and difficulties in external state-updating.
• Long exploratory simulation runs (> 100 years) of various slice lengths are crucial for identifying potential modelling errors arising from model initialization.
• Hydrological models should be run to a steady state, similar to hydrogeological modelling practices, to mitigate initialization errors.

Contributions

• Identifies and quantifies a previously understudied challenge: the significant impact of model initialization and landscape evolution on long-term distributed hydrological simulations.
• Demonstrates the limitations of external coupling between distributed models for non-stationary hydrological systems.
• Provides practical recommendations for improving long-term hydrological simulations, including the necessity of long exploratory runs and steady-state initialization.

Funding

Not specified in abstract.

Citation

@article{Fan2026When,
  author = {Fan, Xinyang and Schaefli, Bettina and Hofmeister, Florentin and Chiogna, Gabriele},
  title = {When the Past Matters: How Model Initialization Can Lead to Surprises in Long‐Term Simulations in Glaciated Environments},
  journal = {Hydrological Processes},
  year = {2026},
  doi = {10.1002/hyp.70507},
  url = {https://doi.org/10.1002/hyp.70507}
}