Improving the temporal and spatial vegetation variability in land surface models based on satellite observations

Land-atmosphere interactions are largely controlled by vegetation, which is dynamic across spatial and temporal scales. Most state-of-the-art land surface models do not adequately represent the temporal and spatial variability of vegetation, which results in weaknesses in the associated variability of modelled surface water and energy states and fluxes. The objective of this work is to evaluate the effects of integrating spatially and temporally varying vegetation characteristics derived from satellite observations on modelled evaporation and soil moisture in the land surface model HTESSEL. Specifically, model fixed land cover was replaced by annually varying land cover, and model seasonally varying Leaf Area Index (LAI) was replaced by seasonally and inter-annually varying LAI. Additionally, satellite data of Fraction of green vegetation Cover (FCover) was used to formulate and integrate a spatially and temporally varying model effective vegetation cover parameterization. The effects of these three implementations on model evaporation and soil moisture were analysed using historical offline (land-only) model experiments at a global scale, and compared to reference datasets. The enhanced vegetation variability lead to considerable improvements in correlation of anomaly evaporation and surface soil moisture in semiarid regions during the dry season. These improvements are related to the adequate representation of vegetation-evaporation-soil moisture feedback mechanisms during water-stress periods in the model, when integrating spatially and temporally varying vegetation. These findings emphasize the importance of vegetation variability for modelling land surface-atmosphere interactions, and specifically droughts. This research contributes to the understanding and development of land surface models, and shows that satellite observational products are a powerful tool to represent vegetation variability.