Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: A Retrospective Analysis of Model Uncertainty for Forecasting Hydrologic Change 1884

Authors
item Semmens, D. - US-EPA
item Hernandez, M. - UNIVERSITY OF ARIZONA
item Goodrich, David
item Kepner, W. - US-EPA

Submitted to: Environmental Modelling & Software
Publication Type: Proceedings
Publication Acceptance Date: July 1, 2006
Publication Date: August 1, 2006
Citation: Semmens, D., Hernandez, M., Goodrich, D.C., Kepner, W.G. 2006. A retrospective analysis of model uncertainty for forecasting hydrologic change. In: Voinov, A., Jakeman, A., Rizzoli, A. (eds). Proceedings of the iEMSs Third Biennial Meeting: "Summit on Environmental Modelling and Software". International Environmental Modelling and Software Society, July 2006, Burlington, VT. 2006. CD ROM.

Interpretive Summary: A relatively new planning process called “Alternative Futures” is being more widely applied for local and regional planning and decision making. In this process a future landscape is developed e.g. 20 years in the future) for the area of interest typically under three scenarios (e.g. restricted development, development as currently conducted, and unrestricted development). These future landscapes can then be used as input into watershed computer models to evaluate potential environmental and watershed impacts. Care must be exercised in this analysis as errors can, and typically do exist in watershed computer models. In this study, historical landscape change over a 24-year period was used to measure the watershed model error with known (i.e. observed alternative landscapes). Results demonstrate that if future land-use/cover and climate conditions are known precisely, the model does a satisfactory job of predicting observed conditions almost 25 years into the future. Unfortunately, future land-use/cover and climate conditions can never be known with certitude. The goal of regional planning efforts is to explore desired outcomes, and it is assumed that policy can be used to shape future change and guide it towards a particular outcome. As a result, climate conditions are the primary unknown in projecting future hydrologic response. Results of the present study indicate that by holding climate constant, it is possible to evaluate qualitatively the broad spatial patterns of hydrologic response to landscape change within a basin. Given the sensitivity of hydrologic response to climatic conditions, future research will focus more attention on the use of climate scenarios to characterize hydrologic response for a range of climatic conditions.

Technical Abstract: GIS-based hydrologic modeling offers a convenient means of assessing the impacts associated with land-cover/use change for environmental planning efforts. Alternative future scenarios can be used as input to hydrologic models and compared with existing conditions to evaluate potential environmental impacts as part of this process. Model error, however, can be significant and potentially compounded when projecting future land-cover/use change and management conditions. To address this problem we have utilized repeat observations of land cover/use as a proxy for projected future conditions. A systematic analysis of model efficiency during simulations based on observed land-cover/use change is used to quantify error associated with simulations for a series of known “future” landscape conditions over a 24-year period. Calibrated and uncalibrated assessments of relative change over different lengths of time are also presented to determine the types of information that can reliably be used in planning efforts for which calibration is not possible.

Last Modified: 8/30/2014
Footer Content Back to Top of Page