Spring phenology, taken as the timing of leaf appearance, is both an indicator of climatic changes and a factor influencing the vegetation annual carbon uptake. We developed a remote sensing methodology to measure the timing of the leaf appearance in boreal deciduous forests without being affected by snow, as shown with the 8 day RMS difference with ground observations of the date of leaf appearance. These remote sensing estimates of the date of leaf appearance were used to calibrate a phenology model. All together, in situ measurements, remote sensing estimates and modeled dates were used to study the variations in spring phenology since 1920. Moreover, the model was integrated in a dynamic vegetation model to take into account phenological variations in the forest carbon budget. Finally, the phenological modeling was extended to the tundra area.