Postprint version. Published in Journal of the American Water Resources Association, Volume 35, Issue 2, April 1, 1999, pages 457-468.
The definitive version is available at http://dx.doi.org/10.1111/j.1752-1688.1999.tb03603.x.
An evaluation was conducted on three forested upland watersheds in the northeastern U.S. to test the suitability of TOPMODEL for predicting water yield over a wide range of climatic scenarios. The analysis provides insight of the usefulness of TOPMODEL as a predictive tool for future assessments of potential long-term changes in water yield as a result of changes in global climate. The evaluation was conducted by developing a calibration procedure to simulate a range of climatic extremes using historical temperature, precipitation, and streamfiow records for years having wet, average, and dry precipitation amounts from the Leading Ridge (Pennsylvania), Fernow (West Virginia), and Hubbard Brook (New Hampshire) Experimental Watersheds. This strategy was chosen to determine whether the model could be successfully calibrated over a broad range of soil moisture conditions with the assumption that this would be representative of the sensitivity necessary to predict changes in streamfiow under a variety of climate change scenarios. The model calibration was limited to a daily time step, yet performed reasonably well for each watershed. Model efficiency, a least squares measure of how well a model performs, averaged between 0.64 and 0.78. A simple test of the model whereby daily temperatures were increased by 1.7°C, resulted in annual water yield decreases of 4 to 15 percent on the three watersheds. Although these results makes the assumption that the model components adequately describe the system, this version of TOPMODEL is capable to predict water yield impacts given subtle changes in the temperature regime. This suggests that adequate representations of the effects of climate change on water yield for regional assessment purposes can be expected using the TOPMODEL concept.