Kalura et al. (2026) Enhancing Hydrologic Model Performance in a Data-Scarce Basin Using Satellite-Based Soil Moisture Data

Identification

• Journal: Water Resources Management
• Year: 2026
• Date: 2026-01-01
• Authors: Praveen Kalura, Anant Pandey, Deen Dayal, V. M. Chowdary
• DOI: 10.1007/s11269-025-04440-y

Research Groups

• Department of Water Resources Development and Management, Indian Institute of Technology Roorkee, Roorkee, India
• National Remote Sensing Centre (NRSC), Indian Space Research Organisation, Hyderabad, India

Short Summary

This study evaluates the effectiveness of incorporating satellite-derived soil moisture (SM) into the calibration of the Variable Infiltration Capacity (VIC) model in the data-scarce Wardha River Basin, demonstrating that multivariate calibration with SM significantly improves streamflow predictions, especially low-flow conditions, and enhances internal hydrological state representation.

Objective

• To evaluate the effectiveness of incorporating satellite-derived soil moisture (SM) data, using multivariate calibration strategies, to enhance the performance of the Variable Infiltration Capacity (VIC) hydrological model in a data-scarce basin.

Study Configuration

• Spatial Scale: Wardha River Basin (WRB), approximately 46,250 square kilometers, with the VIC model configured at a 0.05 degree (5 kilometer) spatial resolution.
• Temporal Scale: Daily time step simulation from 2001 to 2017, including a 2-year warm-up (2001–2002), a 9-year calibration period (2003–2011), and a 5-year independent evaluation period (2012–2016).

Methodology and Data

• Models used:
  • Variable Infiltration Capacity (VIC) hydrological model.
  • Lohmann routing model for streamflow generation.
  • Dynamically Dimensioned Search (DDS) global optimisation algorithm for calibration.
• Data sources:
  • Meteorological forcing: India Meteorological Department (IMD) precipitation (0.25°, 2003–2017) and temperature (1°, 2003–2017), ERA5 wind speed (0.1°).
  • Topography: SRTM (90 meters).
  • Soil properties: Harmonized World Soil Database (HWSD) (1 kilometer).
  • Land surface parameters: Decadal Land Use/Land Cover (LULC) classifications (100 meters), MODIS-derived Leaf Area Index (LAI) and albedo (1 kilometer).
  • Observed streamflow: Central Water Commission (CWC) gauges.
  • Satellite-derived soil moisture (SM) and evapotranspiration (ET): Global Land Evaporation Amsterdam Model (GLEAM) version 4.2 (0.1° resolution, 1980–2023).

Main Results

• Incorporating satellite-derived soil moisture (SM) data into VIC model calibration significantly improved streamflow simulation, particularly under low-flow conditions, and enhanced predictive skill for upstream and tributary sub-basins.
• Multivariate calibration strategies (M2, M3, M4) generally achieved a more balanced performance across various hydrological components compared to the streamflow-only approach (M1), which, while performing best at the calibrated outlet, poorly represented internal soil moisture dynamics.
• The dry-season mean absolute error in streamflow was substantially reduced from 76 cubic meters per second (M1) to 10 cubic meters per second under the spatial SM calibration (M2).
• The spatial correlation coefficient (R) between simulated and observed soil moisture increased from 0.15 (M1) to 0.35 when SM metrics were included, and the temporal SM correlation rose from 0.38 to 0.57.
• SM-informed calibration strategies (M2–M4) led to 15–80% reductions in streamflow prediction error at validation sites lacking calibration data.
• Integrating SM objectives did not negatively impact evapotranspiration simulations, maintaining comparable accuracy against GLEAM data across all calibration scenarios.
• The optimal calibration strategy was found to be context-dependent, varying with basin characteristics such as land cover and size (e.g., M4 was most beneficial for small agricultural basins, while M2 performed best in forested sub-basins).

Contributions

• Joint calibration using both SM and streamflow (multivariate strategies M2–M4) substantially improved model predictive skill, increasing the mean Nash–Sutcliffe Efficiency (NSE) by approximately 13–14% and streamflow correlation by about 7% in ungauged sub-basins compared to single-variable calibration.
• Achieved marked improvements in low-flow (dry-season) performance through the inclusion of spatial and temporal SM metrics.
• Demonstrated significant gains in representing internal soil moisture dynamics without degrading other components of the water balance, leading to a more realistic model.
• Helped mitigate the limitations of sparse gauge networks by effectively utilizing satellite SM data, resulting in reduced streamflow prediction error at validation sites.
• Encourages the adoption of multi-variate calibration techniques for improved prediction of both high-flow and low-flow conditions and a better depiction of water balance components, which is crucial for operational water management in data-limited regions.

Funding

• National Remote Sensing Centre (NRSC), Indian Space Research Organisation, Hyderabad (Grant number: ISR-1312-WRC).

Citation

@article{Kalura2026Enhancing,
  author = {Kalura, Praveen and Pandey, Anant and Dayal, Deen and Chowdary, V. M.},
  title = {Enhancing Hydrologic Model Performance in a Data-Scarce Basin Using Satellite-Based Soil Moisture Data},
  journal = {Water Resources Management},
  year = {2026},
  doi = {10.1007/s11269-025-04440-y},
  url = {https://doi.org/10.1007/s11269-025-04440-y}
}