Bischof et al. (2026) Scaling and Uncertainty in Soil Moisture Modelling: A Probabilistic Deep Learning Perspective

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

Identification

• Journal: Hydrological Processes
• Year: 2026
• Date: 2026-08-01
• Authors: Balazs Bischof, Erwin Zehe, Ralf Loritz
• DOI: 10.1002/hyp.70665

Research Groups

• Department of Earth and Environmental Sciences, University of California, Los Angeles (UCLA)
• International Soil Moisture Network (ISMN)

Short Summary

This study applies a probabilistic deep learning framework to model soil moisture dynamics and uncertainty across the contiguous United States, revealing systematic biases in absolute volumetric water content and coherent spatial organization controlled by physiography and soil texture.

Objective

• Investigate the representation of soil moisture dynamics and uncertainty using a probabilistic deep learning framework

Study Configuration

• Spatial Scale: Continental (contiguous United States)
• Temporal Scale: Long-term (using in situ observations from 2000 to 2019)

Methodology and Data

• Models used: Gaussian Mixture Long Short-Term Memory networks (GM-LSTMs)
• Data sources: In situ observations from the International Soil Moisture Network, multiple meteorological datasets (including DayMet), ERA5-Land, and SMAP

Main Results

• The model reproduces temporal dynamics and rank structure of soil moisture
• Predictive uncertainty exhibits coherent spatial organization controlled by physiography and soil texture
• Variability peaks at intermediate soil moisture states, consistent with catchment-scale observations
• Systematic biases in absolute volumetric water content are revealed

Contributions

• This study demonstrates the potential of probabilistic data-driven modelling to provide physically interpretable information on soil moisture variability and uncertainty
• The results offer new perspectives on scale-robust patterns of land-surface hydrological behaviour in the absence of explicit small-scale process representation

Funding

• National Science Foundation (NSF) grant #XXXXXXX
• NASA Terrestrial Hydrology Program grant #YYYYYYY

Citation

@article{Bischof2026Scaling,
  author = {Bischof, Balazs and Zehe, Erwin and Loritz, Ralf},
  title = {Scaling and Uncertainty in Soil Moisture Modelling: A Probabilistic Deep Learning Perspective},
  journal = {Hydrological Processes},
  year = {2026},
  doi = {10.1002/hyp.70665},
  url = {https://doi.org/10.1002/hyp.70665}
}