Lu et al. (2025) Blue and green water simulation in the river basin using remote sensing data fusion and dual-variable hydrological calibration

Identification

• Journal: Agricultural Water Management
• Year: 2025
• Date: 2025-12-13
• Authors: Yin Lu, Chang Feng, Zhonghui Guo, Yang Liu, Qing Liu
• DOI: 10.1016/j.agwat.2025.110046

Research Groups

• College of Geography and Tourism, Hengyang Normal University, Hengyang, China

Short Summary

This study developed and evaluated a dual-variable hydrological model calibration method, integrating remote sensing fusion evapotranspiration (ET) data with observed runoff, to improve the accuracy and reduce uncertainty in blue water and green water simulations in the Xiangjiang River Basin. The method significantly enhanced the simultaneous simulation accuracy of both blue and green water compared to traditional single-variable calibration.

Objective

• To propose a blue and green water simulation framework that integrates remote sensing data fusion with the SWAT hydrological model to improve watershed simulation accuracy.
• To evaluate the role and advantages of fused ET products in dual-variable calibration within the Xiangjiang River Basin.

Study Configuration

• Spatial Scale: Xiangjiang River Basin (XRB), China, covering approximately 82,900 km², divided into 30 sub-basins.
• Temporal Scale: Study period from 2000 to 2013, with a warm-up period (2000–2001), calibration period (2002–2007), and validation period (2008–2013). Data fusion and model calibration were performed at a monthly scale.

Methodology and Data

• Models used:
  • Hydrological Model: SWAT (Soil and Water Assessment Tool)
  • Remote Sensing Spatiotemporal Fusion Algorithm: ESTARFM (Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model)
• Data sources:
  • Remote Sensing ET products: MODIS16A2 ET (500 m spatial resolution, 8-day temporal resolution), GLDAS ET (0.25° × 0.25° spatial resolution, 3-hour temporal resolution). These were fused to create high spatiotemporal resolution ET data.
  • Soil Moisture (for validation): ERA5-Land and GLEAM v3.7 data.
  • Observed Data: Monthly runoff monitoring data from Xiangtan Station (2000–2013); daily meteorological data (precipitation, maximum/minimum temperature, average wind speed, average relative humidity, sunshine duration, small-scale evapotranspiration monitoring data) from 43 meteorological stations; site-measured ET reference values from 21 meteorological stations (interpolated using Thiessen polygons).
  • Geospatial Data: Digital Elevation Model (DEM) (30 m ASTER GDEMV3), Land Use (30 m CNLUCC), Soil data (1 km HWSD-V1.2).

Main Results

• The remote sensing fusion evapotranspiration (ET) data achieved the highest accuracy (correlation coefficient R = 0.87, relative error Re = 37.5 %, root mean square error RMSE = 56.56 mm/month), outperforming individual MODIS ET and GLDAS ET products by balancing authenticity and accuracy.
• The dual-variable calibration scheme (Runoff + Fusion ET, S7) significantly improved the simultaneous simulation accuracy of both blue water (runoff) and green water (ET) compared to single-variable calibration schemes.
• The dual-variable scheme (S7) substantially enhanced green water simulation accuracy (mean R improved by 19.64%, KGE by 37.50% compared to traditional runoff-only calibration S1), while maintaining high blue water simulation performance (mean R² = 0.92, NSE = 0.80, KGE = 0.71).
• SWAT model parameter sensitivity was found to be dynamic, influenced by the chosen calibration variables and remote sensing ET products, with dual-variable calibration optimizing a broader range of parameters affecting both surface and groundwater exchange.
• The S7 scheme significantly improved green water storage (SW) simulation accuracy, showing better correlation (average CC = 0.58) and lower RMSE (60.03 mm) when validated against ERA5-Land data, and consistent temporal trends with both ERA5-Land and GLEAM.

Contributions

• Proposed and validated a novel blue and green water simulation framework that integrates remote sensing data fusion with the SWAT hydrological model, enhancing watershed simulation accuracy.
• Demonstrated the effectiveness of spatiotemporal fusion algorithms in generating high-quality ET data, which addresses the trade-off between spatial and temporal resolution and reduces input data uncertainty.
• Provided an effective approach for accurate simulation of blue-green water at the watershed scale, offering scientific support for water resources optimal allocation, vegetation water conservation assessment, and ecosystem service function quantification.
• Addressed the "point-to-area" scale inconsistency problem in green water calibration by effectively integrating in-situ observations with remote sensing-fused ET data.
• Offered new insights into addressing parameter uncertainty and equifinality in hydrological modeling through dual-variable calibration, improving the physical realism and identifiability of model parameters.

Funding

• National Natural Science Foundation of China (grant Nos. 42001024 and 41901026)
• Hunan Provincial Natural Science Foundation of China (grant No. 2025JJ50199)
• Graduate Research Innovation Foundation of Hunan Province (grant No. CX20231238)

Citation

@article{Lu2025Blue,
  author = {Lu, Yin and Feng, Chang and Guo, Zhonghui and Liu, Yang and Liu, Qing},
  title = {Blue and green water simulation in the river basin using remote sensing data fusion and dual-variable hydrological calibration},
  journal = {Agricultural Water Management},
  year = {2025},
  doi = {10.1016/j.agwat.2025.110046},
  url = {https://doi.org/10.1016/j.agwat.2025.110046}
}