Terrestrial vegetation and water balance - hydrological evaluation of a dynamic global vegetation model

Earth's vegetation plays a pivotal role in the global water balance. Hence, there is a need to model dynamic interactions and feedbacks between the terrestrial biosphere and the water cycle. Here, the hydrological performance of the Lund-Potsdam-Jena model (LPJ), a prominent dynamic global vegetation model, is evaluated. Models of this type simulate the coupled terrestrial carbon and water cycle, thus they are well suited for investigating biosphere-hydrosphere interactions over large domains. We demonstrate that runoff and evapotranspiration computed by LPJ agree well with respective results from state-of-the-art global hydrological models, while in some regions, runoff is significantly over- or underestimated compared to observations. The direction and magnitude of these biases is largely similar to those from other macro-scale models, rather than specific to LPJ. They are attributable primarily to uncertainties in the climate input data, and to human interventions not considered by the model (e.g. water withdrawal, land cover conversions). Additional model development is required to perform integrated assessments of water exchanges among the biosphere, the hydrosphere, and the anthroposphere. Yet, the LPJ model can now be used to study inter-relations between the world's major vegetation types and the terrestrial water balance. As an example, it is shown that a doubling of atmospheric CO2 content alone would result in pronounced changes in evapotranspiration and runoff for many parts of the world. Although significant, these changes would remain unseen by stand-alone hydrological models, thereby emphasizing the importance of simulating the coupled carbon and water cycle. (C) 2003 Elsevier B.V. All rights reserved.