TL;DR
This study comprehensively evaluates the Hydrological Simulation Program-Fortran (HSPF) model's performance in simulating streamflow and nutrient transport in a mesoscale monsoon watershed in China. The findings demonstrate that HSPF is a robust tool for this purpose, achieving satisfactory to very good results across various time steps, which is crucial for effective watershed management.
What Problem Does This Paper Address?
Understanding and managing water resources and mitigating nonpoint source pollution are critical global challenges. These issues are driven by complex hydrological and biogeochemical processes within watersheds, which are influenced by numerous factors like climate, land use, and human activities. Accurately simulating these processes is essential for developing effective management strategies. While various hydrological models exist, a thorough evaluation of their performance, especially in diverse climatic regions like East Asian monsoon areas, is vital to ensure their reliable application. This paper addresses the need for a comprehensive performance assessment of the widely used HSPF model in such a context, specifically focusing on both hydrological and nutrient simulations.
Methods
The study employed the Hydrological Simulation Program-Fortran (HSPF) model, a well-established tool for simulating nonpoint source runoff and pollutant loadings. The model was set up for the Xitiaoxi River watershed, a mesoscale monsoon watershed in China.
Key methodological steps included:
- Model Construction: This involved acquiring and processing essential input data such as digital elevation models (DEM), land use data, soil data, and meteorological data.
- Calibration and Validation:
- Streamflow: The model was calibrated using daily observed streamflow data from January 2002 to December 2007 and validated using data from January 2008 to December 2010. The Model-Independent Parameter Estimation Tool (PEST) was used for automated runoff parameter calibration.
- Nutrients: Nutrient simulations (nitrate, ammonium, and orthophosphate) were calibrated and validated using monthly observed data from July 2009 to July 2010. This calibration was performed manually.
- Performance Evaluation: The model's performance was assessed using standard statistical metrics: the coefficient of determination (R²), Nash-Sutcliffe coefficient of efficiency (Ens), revised Ens (Ens'), and percent bias (PBIAS). These metrics were applied to annual, monthly, and daily time steps for streamflow, and to monthly time steps for nutrients.
Datasets
The study utilized the following datasets for the Xitiaoxi watershed:
- Digital Elevation Model (DEM): A 30 × 30 m resolution DEM derived from a 1:50,000 relief map.
- Land Use Data: 30 m grid size data interpreted from SPOT imagery acquired in 2004.
- Soil Type Map: A map generated in 2000 by the local soil survey department at a 1:600,000 scale.
- Meteorological Data: Data from seven meteorological stations within the watershed, including solar radiation, air temperature, dew-point temperature, precipitation, evapotranspiration, wind speed, cloud cover, and atmospheric pressure for the period 2002–2010.
- Hydrological Data: Daily observed streamflow data from the Gangkou hydrological station (watershed outlet) for the period January 2002 to December 2010.
- Water Quality Data: Monthly observed nutrient data (nitrate, ammonium, and orthophosphate) from the Hengtang and Ancheng Bridge water quality monitor stations for the period July 2009 to July 2010.
No new benchmarks were introduced; existing observed data were used for evaluation.
Key Insights and Ideas
- HSPF's Robustness in Monsoon Climates: The study confirms that HSPF, despite its general applicability, performs well even in the complex hydrological regimes of East Asian monsoon watersheds, characterized by distinct wet and dry seasons.
- Importance of Data Quality: The paper implicitly highlights that while HSPF is adaptable, its performance is significantly influenced by the quality and resolution of input data (DEM, land use, meteorological, and observed data).
- Nutrient Simulation Challenges: The limited availability of monthly nutrient data, compared to daily streamflow data, presented a challenge but still yielded satisfactory results, suggesting that even with less frequent observations, HSPF can provide valuable insights into nutrient transport.
- Parameter Sensitivity is Key: The study emphasizes the importance of identifying and calibrating key parameters, acknowledging that parametric uncertainty is a common issue in hydrological modeling.
Key Results
Hydrological Simulation:
- Annual Streamflow: Achieved very good performance with R² = 0.87 and Ens = 0.82 during calibration. Validation showed good agreement with R² = 0.94, though Ens was negative (-19.65), indicating some underestimation in wet years.
- Monthly Streamflow: Demonstrated good performance with R² = 0.77 and Ens = 0.76 during calibration. Validation also showed good results with R² = 0.94 and Ens = 0.87.
- Daily Streamflow: Showed satisfactory performance with R² = 0.65 and Ens = 0.65 during calibration. Validation results were also good with R² = 0.86 and Ens = 0.80.
- Overall, the model generally captured streamflow trends well, with some underestimation noted during rainy days and wet years.
Nutrient Simulation:
- Nitrate (NO₃⁻-N): Calibration showed satisfactory results with R² = 0.73 and Ens = 0.67. Validation yielded R² = 0.71 and Ens = 0.66.
- Ammonium (NH₄⁺-N): Performed well during calibration with R² = 0.82 and Ens = 0.74. Validation results were moderate with R² = 0.58 and Ens = 0.53.
- Orthophosphate (PO₄³⁻-P): Achieved very good results during calibration with R² = 0.92 and Ens = 0.86. Validation showed good performance with R² = 0.76 and Ens = 0.67.
- Despite data limitations, nutrient simulations were generally satisfactory, indicating HSPF's capability in this regard.
Limitations and Future Work
Limitations:
- Data Availability: The study acknowledges the limitation of having only monthly observed data for nutrient simulations, which restricts the ability to capture diurnal variations and potentially impacts the accuracy of nutrient modeling.
- Spatial Resolution of Rainfall: The paper suggests that the spatial resolution of rainfall gauges might affect storm runoff simulation, and localized storm events may not be fully captured by the current station network.
- Model Simplifications: HSPF, like any model, relies on empirical relationships and assumptions (e.g., well-mixed receiving water bodies), which may not perfectly represent all real-world processes.
Future Work:
- Parameter Library: The study suggests that the presented calibrated parameters could be useful for similar watersheds, and further work is needed to establish a comprehensive parameter library.
- Spatial Scale and Time Series: Further studies are required to assess HSPF's suitability across different spatial scales and time series.
- Integration with Other Models: The authors recommend integrating HSPF with other models to achieve more comprehensive simulations, especially for complex watershed systems.
- Improved Data: The paper highlights the need for more detailed, complete, and accurate data, particularly for water quality, to enhance model performance.
- Application in Developing Countries: With increasing accessibility of data like DEMs and land use, HSPF's application is expected to expand in developing countries, necessitating continued evaluation and adaptation.
Summary updated