In this study, a number of grassland plots with distinct differences in yield, canopy structure and plant species composition were investigated from the beginning of plant development until the first cut in 2004. Measurements by the ASD FieldSpec II spectrometer were performed and afterwards, above-ground plant material was harvested for deriving ground truth data (n=42). Overall, for all tested methods (vegetation indices, empirical statistical methods, Linear Spectral Mixture Analysis), a 3-endmember mixture model with optimized background component produced best results for estimating the structural variable of green leaf area index: The coefficients of determination were r²=0.72 and RMSE=0.52. Cross-validation proved robustness of the prediction model (r²=0.68, RMSE=0.56). Scaling up the model to HyMap image data (5 m ground resolution) provided estimation results both realistic and physically reasonable.