Guyumus et al. (2026) HydroBlocks-MSSUBv0.1: a multiscale approach for simulating lateral subsurface flow dynamics in Land Surface Models

Identification

• Journal: Geoscientific model development
• Year: 2026
• Date: 2026-01-15
• Authors: D. Guyumus, Laura Torres‐Rojas, Luiz Bacelar, Chengcheng Xu, Nathaniel Chaney
• DOI: 10.5194/gmd-19-477-2026

Research Groups

• Department of Civil and Environmental Engineering, Duke University, Durham, NC, USA
• Program in Atmospheric and Oceanic Sciences, Princeton University, Princeton, NJ, USA

Short Summary

This study introduces HydroBlocks-MSSUBv0.1, a multiscale tiling scheme for Land Surface Models to efficiently simulate local, intermediate, and regional lateral subsurface flow dynamics. The scheme demonstrates consistency with a high-resolution benchmark simulation for key hydrological variables, offering computational efficiency for large-scale applications.

Objective

• To develop and evaluate HydroBlocks-MSSUBv0.1, a multiscale tiling scheme within the HydroBlocks Land Surface Model, to efficiently and accurately represent local, intermediate, and regional lateral subsurface flow dynamics, addressing the computational challenges of incorporating surface heterogeneity in large-scale hydrological simulations.

Study Configuration

• Spatial Scale: A 1° × 1° domain (approximately 8750 km²) in Southwest Colorado, USA. High-resolution input data at ~30 m. The multiscale experiment used 40,700 tiles, while the benchmark simulation used 1.4 million tiles. The soil column depth was set to 10 m, divided into six variable-depth layers (0.05 m, 0.1 m, 0.35 m, 1.0 m, 1.5 m, and 7 m).
• Temporal Scale: Simulations were run for 150 years with an hourly time step, using 10 years of hourly forcing data (2010-2019) repeated.

Methodology and Data

• Models used:
  • HydroBlocks v0.2 (Land Surface Modeling framework)
  • Noah-MP (vertical 1D-column model within HydroBlocks)
  • SIMTOP (simple groundwater model based on TOPMODEL runoff scheme within Noah-MP)
  • Darcy's law for lateral flow (extended to unsaturated zone with moisture-dependent hydraulic conductivity).
  • Richards' equation for vertical flow.
  • Hierarchical Multivariate Clustering scheme (HMC) for tiling.
• Data sources:
  • Elevation: National Elevation Dataset (NED, now 3DEP) at ~30 m resolution.
  • Soil Properties: POLARIS (30 m probabilistic soil property maps) for the contiguous United States, up to 2 m deep (linear extrapolation for deeper layers).
  • Meteorology: Princeton Climate Forcing (PCF) dataset at 1/32° (~3 km) spatial resolution, hourly, from 2010 to 2019.
  • Land Cover: National Land Cover Database (NLCD) at ~30 m resolution.

Main Results

• The multiscale scheme effectively simulates regional and intermediate water movement, primarily driven by elevation heterogeneity, while supporting parallel processing.
• Heterogeneity in land cover, soil properties, and meteorology significantly alters local subsurface flow patterns and magnitudes, with observed changes in mean divergence of ~36% (land cover), ~73% (soil properties), and ~11% (meteorology) compared to homogeneous conditions.
• Compared to a baseline (local lateral flow only), the multiscale scheme shows positive differences in Soil Moisture Content (SMC) in valleys and negative differences on ridges, increased Latent Heat Flux (LH) and decreased Sensible Heat Flux (SH) in valleys, increased Surface Runoff (R) in lower elevation streams, and a consistently lower Water Table Elevation (ZWT) across the domain, particularly in uplands, while preserving seasonal signals.
• When compared to a 1.4 million tile benchmark simulation, the multiscale scheme (40,700 tiles) shows strong agreement in the temporal variability of sensible and latent heat fluxes (high Pearson correlation coefficients).
• Minor discrepancies were observed in the total water budget for SMC, R, and ZWT (e.g., <1% difference in spatial mean for SMC, ~8% for R), with larger variability in ZWT in the multiscale scheme, potentially due to coarser representation of elevation heterogeneity.
• SMC at the root zone (0–0.05 m) shows strong agreement with the benchmark, but substantial differences emerge at the bottom of the soil column (7–10 m), highlighting the impact of elevation and aggregated subsurface heterogeneity at deeper layers.
• The multiscale implementation achieved an 8.5-fold speedup, consuming only 12% of the computational resources compared to the benchmark simulation for a 1-year run.

Contributions

• Introduction of HydroBlocks-MSSUBv0.1, a novel multiscale tiling scheme that explicitly incorporates local, intermediate, and regional lateral subsurface flow dynamics into Land Surface Models.
• Development of a computationally efficient approach to represent sub-grid heterogeneity in large-scale hydrological simulations, leveraging HydroBlocks' hierarchical clustering.
• Demonstration of the scheme's ability to reproduce hydrological patterns consistent with a high-resolution benchmark, providing a viable alternative to computationally intensive fully distributed models.
• Proof-of-concept for integrating multi-scale subsurface flow within a parallel computing framework, enabling future continental-scale simulations.
• Highlighting the critical role of elevation heterogeneity and the aggregation of subsurface properties in tiling approaches for accurate deep soil moisture and water table dynamics.

Funding

• National Oceanic and Atmospheric Administration (NOAA) grants:
  • NA23OAR40504311I ("Next-Generation of NOAA water modeling: Climate Risks & Interactive Sub-seasonal to Seasonal Predictability in the Earth System modeling framework – Bipartisan Infrastructure Law/Infrastructure Investment and Jobs Act")
  • NA24OARX431C0052-T1-01 ("Confronting the GFDL land model's sub-grid tiling scheme with observed space-time patterns of land surface temperature: Implications for hydrologic extremes.")

Citation

@article{Guyumus2026HydroBlocksMSSUBv01,
  author = {Guyumus, D. and Torres‐Rojas, Laura and Bacelar, Luiz and Xu, Chengcheng and Chaney, Nathaniel},
  title = {HydroBlocks-MSSUBv0.1: a multiscale approach for simulating lateral subsurface flow dynamics in Land Surface Models},
  journal = {Geoscientific model development},
  year = {2026},
  doi = {10.5194/gmd-19-477-2026},
  url = {https://doi.org/10.5194/gmd-19-477-2026}
}