The Validation of VEMAP Phase 2 Model Experiments Using Hydrologic Data.

Gordon, W.S., J.S. Famiglietti, N.L. Fowler, K.A. Hibbard, T.G.F. Kittel, and VEMAP Members.  2001.  The Ecological Society of America 86th Annual Meeting, August 2001.  The Ecological Society of America, Washington, DC.

Abstract

The Vegetation/Ecosystem Modeling and Analysis Project (VEMAP) Phase 2 model experiments investigate the response of biogeochemical and biogeographical models to variability in climate over the conterminous United States using historical and projected transient scenarios of climate and atmospheric CO2. Here we evaluate the behavior of six models by comparing simulated runoff to gauged streamflow from the USGS Hydro-Climatic Data Network (HCDN).   These records cover much of the 20th century providing a lengthy record of monthly flows against which to validate the models.  GIS software is used to manipulate the two distinct types of spatial data: VEMAP grid cells and USGS watersheds. Using time-series analysis, we present the average, monthly, under- or over-estimate of runoff by the models. While the errors are large, the time-series analysis demonstrates the ability of the models to simulate seasonal patterns that are qualitatively similar to those seen in the observed data. We also present the correlation between each of the models and the observed data set for 13 selected watersheds.  We find that generally the performance of the models is quite similar.  The greatest variability in model performance occurs across watersheds.  The correlations between simulated and observed runoff range between 0.12 and 0.94.  Within a single watershed, those correlation coefficients differ by as much as 0.43 and as little as 0.05. This validation exercise increases our understanding of the strengths and shortcomings of these models.



HOME PAGE   | CV MAIN PAGE  | CV  PUBLICATION INDEX  |ECOLOGICAL RESEARCH OVERVIEW  |  CLIMATE RESEARCH OVERVIEW


rev. 22 Mar 2001