Although vegetation indices (VIs) are a key parameter for describing vegetation conditions, their estimates are not routinely available at the required scale and the acceptable accuracy level. Satellite-based data are found to be acceptable for estimating seasonal patterns of the leaf area index (LAI) over large regions but inadequate for daily LAI variations for individual vegetation sites, when compared with ground-based direct and indirect measurements in different vegetation types. Scaling large region data to smaller scales and vice versa has long been an important research issue. In this study, we evaluated VI estimations with radiation reflected in the visible (400-700 nm) and near infrared (700-1000 nm) wavebands measured above the vegetation canopy. Measurements were conducted at flux tower sites located in grasslands, crops, and forests in the United States during the growing seasons for multiple years. Several algorithms of VIs were evaluated for the non-linear effects of atmospheric, vegetation, and soil surface factors, using daily measurements of hyperspectral reflectances, and these VIs were then compared with estimates from flux tower and satellite data. Overall, daily estimates of VIs not only captured the main effects of daily variations but were also robust in describing the seasonal green-ups and senescence for the different vegetation sites.