Evaluation of a non-point source pollution model, AnnAGNPS, in a tropical watershed
Introduction
Limited research exists that describes erosion, sedimentation, and water quality dynamics on watershed scale in Hawaii (Calhoun and Fletcher, 1999). Land use change causes a great impact on the sustainability of Hawaii's natural resources and environmental quality. An integrated approach is vital for successful management of watershed ecosystems (Verstraeten et al., 2003). One of the main components of this approach is numerical modeling, which is becoming an increasingly popular tool for decision making.
Selecting a suitable hydrologic model for a particular application is a challenging task, hence several alternative approaches were considered. A number of watershed-scale models with the three major components (water, sediment, and chemical) that vary in complexity and data input requirements have been developed (Borah and Bera, 2003, Borah and Bera, 2004). AGNPS, AnnAGNPS, ANSWERS-Continuous, DWSM, HSPF, MIKE SHE, and SWAT are among them. Event based models (AGNPS and DWSM) were considered not suitable for our application because a long-term simulation tool for assessment of management practices and land-use change was needed. MIKE SHE was cumbersome to use for long-term continuous simulations due to computationally intensive numerical schemes for runoff and subsurface flow. Of the continuous models only AnnAGNPS and SWAT use SCS TR-55 method for runoff calculation (SCS, 1986), which is a robust empirical approach suitable for wide variety of conditions including those where input data might be limited. Basic principles of AnnAGNPS are similar to that of SWAT; however, best management practices simulations appear to be the strength of AnnAGNPS (Srivastava et al., 2002).
AnnAGNPS and its single event predecessor AGNPS have been widely used for sediment and nutrient load predictions (Suttles et al., 2003) and in decision support systems (Mitchell et al., 1993). The model was tested over wide range of environments in Europe (Pekarova et al., 1999), North America (Perrone and Madramootoo, 1997, Yuan et al., 2001), Australia (Baginska et al., 2003), and Africa (Leon et al., 2003). The popularity of AnnAGNPS has recently increased due to integration with GIS capabilities, which significantly simplified its use (Tsou and Zhan, 2004).
Annualized and event based AGNPS were calibrated and validated under different conditions. Mostaghimi et al. (1997) used AGNPS to simulate water quantity and quality on a 1153-ha agricultural watershed in Virginia Piedmont region. Good agreement was found between observed and predicted runoff, sediment yield, N, and P loads (R2 = 0.97, 0.96, 0.4, and 0.55, respectively). Yuan et al. (2001) evaluated AnnAGNPS on an 82-ha watershed in the Mississippi delta. These authors found that the model accurately predicted overall daily runoff (R2 = 0.9) during average events, but underestimated it in extreme events. The model has been tested in humid subtropical climate in Georgia (Suttles et al., 2003). The amounts of predicted and observed runoff differed by 30% to 100%, depending on the storm and location on the watershed. The under prediction was attributed to the shortcoming of land cover representation. Baginska et al. (2003), using the model on a dry Australian watershed with infrequent heavy storms concluded that the model is well suited for gross estimations and comparative assessment of best management practices (BMPs). AnnAGNPS was successfully used for optimizing the placement and configurations of BMPs at the watershed scale. Zhen et al. (2004), using heuristic optimization techniques, were able to couple the model and BMP simulation module to find the least cost set of solutions that meet the pollutant load requirements.
Land-based pollutants are identified as the primary threat to coral reef ecosystem in Hawaii. In Hanalei river basin on Kauai Island, soil erosion yield is estimated at 1.4 Mg ha−1 y−1 (Calhoun and Fletcher, 1999). The sources of the bay pollutants are feral ungulates and alien plants that increase erosion in the upper watershed (increased suspended solids, nutrients and pathogens), cesspools and septic systems in urban areas, agricultural operations (taro ponds), water bird impoundments, and grazing land. Herbicides used to control invasive species and weeds in the area were found in trace amounts in runoff (Ozario et al., 2001).
The long-term goal of the Hanalei watershed management is the improvement of water quality and reduction of stress on estuarine and coral reef ecosystems. AnnAGNPS has been used and calibrated in variety of conditions (Baginska et al., 2003, Grunwald and Norton, 2000, Yuan et al., 2001); however, the novelty of this work lies in the type of environment where the model was applied. AnnAGNPS has not been previously tested in tropical volcanic island environment, where annual amount of precipitation may vary 5–10 fold, depending on the location on a 50-km2 watershed, and daily rainfall often exceeds 150 mm. The U.S. Natural Resources Conservation Service is considering the use of AnnAGNPS as a practical tool for watershed management in the Pacific region, hence the need to test its limitations in these extreme conditions. The main goal of this work was to evaluate the performance of AnnAGNPS in a tropical watershed. The specific objectives were: (i) calibrate and validate AnnAGNPS model for Hanalei watershed; (ii) perform sensitivity analysis to identify input parameters that are the most critical in tropical environment; and (iii) identify critical areas of the watershed and their contribution to sediment loading in the river.
Section snippets
Description of the watershed
The study was conducted in Hanalei River basin that extends from the Mount Waialeale (1570 m) to Hanalei Bay, and located in the northern part of Kauai Island, Hawaii. The upper portion of the basin located above USGS stream gauge (−159.469° W; 22.184° N) with area of 4800 ha was used in the modeling (Fig. 1). The watershed is represented by steep mountain slopes and deep fluvial valleys through which flows the upper 20 km of Hanalei River with its tributaries. The lower portion of the river flows
Calibration of AnnAGNPS
Runoff and soil loss in AnnAGNPS are simulated using separate functions; however, nutrient load calculation is flow and sediment dependent, which indicates that in order to use the model in management decision making, particular attention should be given to accurate flow and erosion calibration (Baginska et al., 2003). The rainstorms that occurred on the watershed during the study period were characterized by high frequency and intensity (Table 3). The average annual precipitation for the
Summary and conclusion
This study examined the applicability of AnnAGNPS model to a tropical watershed with volcanic soils and high spatial and temporal variability of rainfall. Measured and simulated annual surface runoffs were similar (2230 mm and 2405 mm, respectively). Monthly totals were predicted with R2 = 0.90 (P < 0.05); however, up to 50% difference between the actual and simulated data was observed during dry season month (May and July). Prediction of daily runoff yielded low coefficient of correlation (R2 = 0.55, P
Acknowledgments
The authors would like to thank USDA-NRCS for providing funding to complete this project. The authors wish to express their appreciation to the staff of the US Geological Survey, Honolulu District Office, Honolulu, HI, for providing the data used in this study. The authors also thank Dr. R. Bingner and Dr. F. Theurer for their technical support and valuable suggestions.
References (43)
- et al.
Modelling nutrient transport in Currency Creek, NSW with AnnAGNPS and PEST
Environmental Modelling and Software
(2003) - et al.
Measured and predicted sediment yield from a subtropical, heavy rainfall, steep-sided river basin: Hanalei, Kauai, Hawaiian Islands
Geomorphology
(1999) - et al.
Water balance and nutrient inputs in bulk precipitation in tropical montane cloud forest in Panama
Journal of Hydrology
(1997) - et al.
Calibration and validation of a non-point source pollution model
Agricultural Water Management
(2000) - et al.
Testing and evaluation of the agricultural non-point source pollution model (AGNPS) on Augucho catchment, western Hararghe, Ethiopia
Agriculture Ecosystems and Environment
(2003) - et al.
Surface runoff, soil erosion and water quality modelling in the Alpone watershed using AGNPS integrated with a Geographic Information System
European Journal of Agronomy
(1997) - et al.
Watershed management modelling in Malawi: application and technology transfer
Environmental Modelling and Software
(2003) - et al.
Modelling pesticide fate with GLEAMS
European Journal of Agronomy
(1995) - et al.
Assessment of management alternatives on a small agricultural watershed
Water Research
(1997) - et al.
River flow forecasting through conceptual models. Part 1. A discussion of principles
Journal of Hydrology
(1970)
Why soil erosion models over-predict small soil losses and under-predict large soil losses
Catena
Testing of AGNPS model application in Slovak microbasins
Physics and Chemistry of the Earth Part B-Hydrology Oceans and Atmosphere
Watershed-scale hydrologic and nonpoint-source pollution models: Review of mathematical bases
Transactions of the ASAE
Watershed-scale hydrologic and nonpoint-source pollution models: Review of applications
Transactions of the ASAE
Using geostatistics to select grid-cell layouts for the AGNPS model
Transactions of the ASAE
Partitioning of rainfall into throughfall, stemflow and interception: effect of forest type, ground cover and climate
Hydrological Processes
A statistical topographic model for mapping climatological precipitation over mountainous terrain
Journal of Applied Meteorology
Second Approximation to the Solution of the Suspended Load Theory
Cited by (87)
A framework for evaluating county-level non-point source pollution: Joint use of monitoring and model assessment
2020, Science of the Total EnvironmentEnhanced field-scale characterization for watershed erosion assessments
2019, Environmental Modelling and SoftwareAn R package for estimating river compound load using different methods
2019, Environmental Modelling and SoftwareSimulating the hydrological response of a small tropical forest watershed (Mata Atlantica, Brazil) by the ANNAGNPS model
2018, Science of the Total Environment