Calibration of a Simple Rainfall-Runoff Model for Long-Term Hydrological Impact Evaluation

INTRODUCTION Continued land development and land-use changes within cities and at the urban fringe present considerable challenges for environmental management. Hydrologic changes including increased impervious area, soil compaction, and increased drainage efficiency generally lead to increased direct runoff, decreased groundwater recharge, and increased flooding, among other problems (Booth 1991). Hydrologic models, especially simple rainfall-runoff models, are widely used in understanding and quantifying the impacts of land-use changes, and to provide information that can be used in land-use decision making. Many hydrologic models are available, varying in nature, complexity, and purpose (Shoemaker et al. 1997). One such model, Long-Term Hydrological Impact Assessment (L-THIA), a simple rainfall-runoff model based on the U.S. Department of Agriculture's Curve Number (CN) method (USDA 1986), was developed to help land-use planners and watershed managers obtain initial insight into the hydrologic impacts of different land-use scenarios, including historic, current, and future alternatives (Harbor 1994). Like other models, L-THIA is based on empirical relationships that capture the main processes and controls on runoff, but do not account for all the conditions and controls specific to particular study sites, and do not predict the baseflow component of stream flow. Where close correspondence between predicted and observed runoff values is required, rather than simply a relative measure of change, it is necessary to produce a modified (calibrated) model. Calibration of rainfall-runoff models with respect to local observational data is used to improve model predictability. When model results match observed values from stream-flow measurement, users have greater confidence in the reliability of the model. In the present study, a simple method based on univariate linear regression has been used to calibrate L-THIA, using land-use change data, model predicted direct runoff, and direct runoff derived from stream-flow data using hydrograph separation. This calibration approach is field-verified and can be used with any simple rainfall-runoff model, if there are observational data available. Interestingly, calibration and verification test results for the Little Eagle Creek watershed in Indiana show the usefulness of this approach in general and at the same time raise new questions about the sensitivity of L-THIA model predictions to land-use changes, precipitation, and selection of CN values. L-THIA--A SIMPLE RAINFALL-RUNOFF MODEL Modeling rainfall-runoff relationships can be complicated and time-consuming because of the numerous variables that are involved (Bhaduri et al. 2001). Models that capture many of the factors controlling runoff typically require extensive input data and user expertise. Some types of users, such as watershed managers or urban planners, need various levels of models to support decision making, including initial assessment tools that can produce results with minimal data and user expertise. Initial assessments can be a cost-effective way to identify areas of importance that can be targeted for further analysis using a more detailed model or field-based study. Providing users with a simple assessment model can help them reach decisions more quickly and efficiently than immediately performing analysis with highly detailed hydrologic models. The L-THIA model, developed to fill the need for a simple assessment tool, has the capability to provide relative estimates of direct runoff and nonpoint source (NPS) pollution from different land uses (Bhaduri 1998). The L-THIA model details, utility, and applicability have been demonstrated in several studies (e.g., Leitch and Harbor 1999, Harbor et al. 2000, Bhaduri et al. 2000, Grove et al. 2001), and L-THIA is now widely accessible through a Web-based version of the model (http://www.ecn. purdue.edu/runoff, Pandey et al. …

[1]  David J. Goodman,et al.  Personal Communications , 1994, Mobile Communications.

[2]  Peter M. Allen,et al.  Automated Base Flow Separation and Recession Analysis Techniques , 1995 .

[3]  Pao-Shan Yu,et al.  Fuzzy multi-objective function for rainfall-runoff model calibration , 2000 .

[4]  B. Bhaduri,et al.  Assessing Watershed-Scale, Long-Term Hydrologic Impacts of Land-Use Change Using a GIS-NPS Model , 2000, Environmental management.

[5]  Kyoung Jae Lim,et al.  Developing a Web-enabled Tool to Assess Long-term Hydrologic Impacts of Land-use Change : Information Technology Issues and a Case Study , 2000 .

[6]  J. Ndiritu,et al.  An improved genetic algorithm for rainfall-runoff model calibration and function optimization , 2001 .

[7]  Ronald A. Sloto,et al.  HYSEP: A Computer Program for Streamflow Hydrograph Separation and Analysis , 1996 .

[8]  Henrik Madsen,et al.  Automatic calibration of a conceptual rainfall-runoff model using multiple objectives. , 2000 .

[9]  Bernard A. Engel,et al.  COMPOSITE VS. DISTRIBUTED CURVE NUMBERS: EFFECTS ON ESTIMATES OF STORM RUNOFF DEPTHS 1 , 1998 .

[10]  B. Bhaduri,et al.  A geographic information system-based model of the long-term impact of land use change on nonpoint-source pollution at a watershed scale , 1998 .

[11]  Derek B. Booth,et al.  Urbanization and the Natural Drainage System -- Impacts, Solutions, and Prognoses , 1991 .

[12]  J. Harbor A Practical Method for Estimating the Impact of Land-Use Change on Surface Runoff, Groundwater Recharge and Wetland Hydrology , 1994 .

[13]  R. T. Roberts,et al.  Urban Hydrology for Small Watersheds (TR-55 Rev.) , 1985 .

[14]  S. Sorooshian,et al.  Automatic calibration of conceptual rainfall-runoff models: sensitivity to calibration data , 1996 .

[15]  A. Elshorbagy,et al.  Performance Evaluation of Artificial Neural Networks for Runoff Prediction , 2000 .

[16]  V. Gupta,et al.  A physically based filter for separating base flow from streamflow time series , 2001 .