| Abstract: | This paper compares various ways of quantifying the importance of land–atmosphere feedback. A widely used land surface hydrology model is used in coupled (to a planetary boundary layer model) and uncoupled modes to compare the adequacy of different feedback indices. It is found that existing feedback indices are primarily based on ‘one factor at a time’ sensitivity analysis and cannot adequately capture the interaction between land and atmosphere. A new index is used which combines factorial design concepts and traditional sensitivity analysis. This index is shown to capture and quantify the strength of interaction between land surface parameters and atmosphere. To assess the effects of forcing characteristics on the stand alone model sensitivity, several ways to specify near-surface atmospheric conditions are evaluated. It is found that commonly used forcing conditions (e.g. model generated or observed time-series of near-surface atmospheric variables) may not be adequate to mimic the coupled model environment for evaluating the land surface representations. The partially coupled model sensitivity is shown to capture a major feedback loop related to water holding capacity, surface fluxes and near-surface atmospheric processes. These results suggest that sensitivity from the stand alone model should be interpreted with caution and future evaluations should strive to incorporate land–atmosphere feedback, at least within a partially coupled model. © 1997 John Wiley & Sons, Ltd. |
