Phenological variables derived from remote sensing are useful in determining the seasonal cycles of ecosystems in a changing climate. Satellite remote sensing imagery is useful for the spatial continuous monitoring of vegetation phenology across broad regions; however, its applications are substantially constrained by atmospheric disturbances such as clouds, dusts, and aerosols. By way of contrast, a tower-based ground remote sensing approach at the canopy level can provide continuous information on canopy phenology at finer spatial and temporal scales, regardless of atmospheric conditions. In this study, a tower-based ground remote sensing system, called the "Phenological Eyes Network (PEN)", which was installed at the Gwangneung Deciduous KoFlux (GDK) flux tower site in Korea was introduced, and daily phenological progressions at the canopy level were assessed using ratios of red, green, and blue (RGB) spectral reflectances obtained by the PEN system. The PEN system at the GDK site consists of an automatic-capturing digital fisheye camera and a hemi-spherical spectroradiometer, and monitors stand canopy phenology on an hourly basis. RGB data analyses conducted between late March and early December in 2009 revealed that the 2G_RB (i.e., 2G - R - B) index was lower than the G/R (i.e., G divided by R) index during the off-growing season, owing to the effects of surface reflectance, including soil and snow effects. The results of comparisons between the daily PEN-obtained RGB ratios and daily moderate-resolution imaging spectroradiometer (MODIS)-driven vegetation indices demonstrate that ground remote sensing data, including the PEN data, can help to improve cloud-contaminated satellite remote sensing imagery.