The Accuracy of Extrapolated Design Estimates with Hydrologic and Environmental Models

by Dawn Dlinesmith,

Document Type: Proceeding Paper

Part of: World Water & Environmental Resources Congress 2003


Extrapolation error in water resources and environmental modeling has not been the subject of adequate study in the past. In research and design, estimates obtained from regression models are generally accepted as accurate within the range of the measured data. Results presented herein prove otherwise. The accuracy of regression model prediction and estimation is addressed. The effects of several parameters including the extrapolation length, sample size, population correlation, and intercorrelation are evaluated using Monte Carlo simulation. Analysis of one-predictor power models and two-predictor power models reveals a high probability for inaccurate predictions when hydrologic and environmental models are extrapolated beyond the range of measured data. For data with moderate correlation, predictions within the data region may even be inaccurate.

Subject Headings: Model accuracy | Hydrologic models | Errors (statistics) | Data analysis | Water resources | Regression analysis | Hydro power

Services: Buy this book/Buy this article


Return to search