Perin et al. (2025) Assessing the cumulative impact of on-farm reservoirs on modeled surface hydrology

Identification

• Journal: Hydrology and earth system sciences
• Year: 2025
• Date: 2025-11-18
• Authors: Vinicius Perin, Mirela G. Tulbure, Shiqi Fang, A. Sankarasubramanian, Michele L. Reba, Mary Yaeger
• DOI: 10.5194/hess-29-6353-2025

Research Groups

• Planet Labs Inc.
• North Carolina State University (Center for Geospatial Analytics; Department of Civil, Construction and Environmental Engineering)
• USDA-ARS Delta Water Management Research Unit
• The University of Memphis (Center for Applied Earth Science and Engineering Research)

Short Summary

This study developed a novel framework integrating remote sensing data with a hydrological model (SWAT+) to assess the cumulative spatial and temporal impacts of on-farm reservoirs (OFRs) on surface hydrology in eastern Arkansas, finding significant reductions in annual flow (14%–24%) and peak flow (43%–60%).

Objective

• To assess the spatial and temporal variability of the cumulative impact of OFRs at the watershed and subwatershed levels.
• To quantify the intra- and interannual impacts of OFRs on flow and peak flow at the channel scale.

Study Configuration

• Spatial Scale: A watershed in eastern Arkansas, USA, covering 7107 km², divided into 68 subwatersheds and 642 Hydrological Response Units (HRUs), with analysis performed at the channel scale for 69 aggregated OFRs.
• Temporal Scale: Daily simulations for 30 years (1990–2020), including a five-year warm-up period.

Methodology and Data

• Models used:
  • Soil Water Assessment Tool+ (SWAT+, v.2.1.9) with QSWAT+ (v.2.1.9) interface and SWAT+ Editor (v.2.1.0).
  • A top-down, data-driven, remote sensing-based algorithm (Kalman filter) for daily OFR surface area dynamics.
• Data sources:
  • Digital Elevation Model (DEM): Shuttle Radar Topography Mission DEM (30 m resolution).
  • Land Use and Land Cover: National Land Cover Database (30 m resolution).
  • Soil Data: Gridded Soil Survey Geographic Database (gSSURGO) (100 m resolution).
  • Climate Data: Gridded Surface Meteorological Datasets (daily precipitation, minimum/maximum temperatures, surface downward shortwave radiation, wind velocity, relative humidity).
  • OFR Inventory: Digitally mapped 330 OFRs (Yaeger et al., 2017).
  • OFR Surface Area Time Series: Derived from multi-sensor satellite imagery (PlanetScope, RapidEye, Sentinel-2) using a Kalman filter (Perin et al., 2022).
  • Flow Observations: Monthly measured flow from three United States Geological Survey (USGS) stations for calibration and validation (14 to 25 years of data).

Main Results

• Model Performance: Calibration and validation showed good agreement for monthly flow, with r² ranging from 0.71 to 0.93, Nash–Sutcliffe model efficiency coefficient (NSE) from 0.68 to 0.90, and Percent Bias (PBIAS) generally within ±12% (with one instance at 18.12%).
• Impact on Annual Flow: The presence of OFRs was associated with a mean decrease in annual flow ranging from 14.6% to 24.2% across different flow classes.
• Temporal Variability of Flow Impact:
  • Largest flow reductions occurred between January and May (mean decrease of 30.0% to 37.6%).
  • Milder impacts were observed from June to December (mean decrease of 1.4% to 12.5%).
  • Positive impacts (increase in flow, up to 8.7%) were noted for larger flow classes during August and October.
• Impact on Peak Flow: OFRs led to a mean reduction in peak flow ranging from 43.9% to 60.7% across flow classes, primarily occurring between January and May.
• OFR Surface Area Scenarios: The lower, mean, and upper OFR surface area scenarios showed similar impacts on flow and peak flow, with differences generally smaller than 5%, despite a significant range in total OFR surface area (2176 ha to 3370 ha).
• Spatial Variability: The highest impacts on annual flow (both positive and negative, exceeding 100%) occurred in subwatersheds that contributed least (<10%) to the main model outlet, corresponding to channels with smaller flow magnitudes. Subwatersheds with the highest reservoir capacities did not consistently exhibit the highest impacts on annual flow.

Contributions

• Proposed a novel framework that systematically integrates dynamic OFR conditions derived from multi-year satellite imagery time series with a process-based hydrological model (SWAT+).
• Provided the first study to combine spatial and temporal variability of OFRs from satellite imagery with a hydrological model to assess cumulative impacts.
• Quantified OFR impacts on flow and peak flow at the channel scale, offering a higher level of detail than previous subwatershed-scale analyses.
• Demonstrated the intra- and interannual variability of OFR impacts, highlighting that effects are not uniformly distributed across a watershed and depend on OFR spatial distribution and storage capacity.
• Developed an approach transferable to other watersheds globally to support water resource management and planning for new OFR construction.

Funding

• NASA through the Future Investigators in NASA Earth and Space Science and Technology fellowship.

Citation

@article{Perin2025Assessing,
  author = {Perin, Vinicius and Tulbure, Mirela G. and Fang, Shiqi and Sankarasubramanian, A. and Reba, Michele L. and Yaeger, Mary},
  title = {Assessing the cumulative impact of on-farm reservoirs on modeled surface hydrology},
  journal = {Hydrology and earth system sciences},
  year = {2025},
  doi = {10.5194/hess-29-6353-2025},
  url = {https://doi.org/10.5194/hess-29-6353-2025}
}