Abstract
This paper discusses the coupling between ecological process models and satellite data. Three major strategies are reviewed, with chosen examples in natural and cultivated vegetation modelling. The first strategy uses satellite data to derive input variables required by models. The spatial and/or temporal variations of these variables allow the modelling of some vegetation processes over large area and/or during the time when observations are available. The second strategy uses satellite data (or secondary variables derived from satellite data) to validate model outputs, and the third one uses them to control the modelling by the calibration of some key processes or the retrieval of initial conditions. Temporal series allows the assimilation of satellite observations in a complete modelling of the temporal signal at the satellite level. Such an approach improves the description of some processes not well known, and a better understanding of their spatial variations. An important point is the capability of describing these processes when observations are not available or unsuitable for a direct inversion of canopy parameters. This strategy can be considered as a model‐based interpolation of satellite data, and can help to build models usable in a prognostic mode at the global scale with more confidence.
Notes
Present address: Potsdam Institut für Klima Polgen Porschung, PO Box 60 12 03, D‐14 412 Potsdam, Germany.