Seasonal phenology of vegetation plays an important role in global carbon cycle and ecosystem productivity. In urban environments, vegetation phenology is also important because of its influence on public health (e.g., allergies), and energy demand (e.g. cooling effects). In this study, we studied the potential use of remotely sensed observations (i.e. Landsat data) to derive some phenology indicators for vegetation embedded within the urban core domains in four distinctly different U.S. regions (Washington, D.C., King County in Washington, Polk County in Iowa, and Baltimore City and County in Maryland) during the past three decades. We used all available Landsat observations (circa 3000 scenes) from 1982 to 2015 and a self-adjusting double logistic model to detect and quantify the annual change of vegetation phenophases, i.e. indicators of seasonal changes in vegetation. The proposed model can capture and quantify not only phenophases of dense vegetation in rural areas, but also those of mixed vegetation in urban core domains. The derived phenology indicators show a good agreement with similar indicators derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) and in situ observations, suggesting that the phenology dynamic depicted by the proposed model is reliable. The vegetation phenology and its seasonal and interannual dynamics demonstrate a distinct spatial pattern in urban domains with an earlier (9-14days) start-of-season (SOS) and a later (13-20days) end-of-season (EOS), resulting in an extended (5-30days) growing season length (GSL) when compared to the surrounding suburban and rural areas in the four study regions. There is a general long-term trend of decreasing SOS (-0.30day per year), and increasing EOS and GSL (0.50 and 0.90day per year, respectively) over past three decades for these study regions. The magnitude of these trends varies among the four urban systems due to their diverse local climate conditions, vegetation types, and different urban-rural settings. The Landsat derived phenology information for urban domains provides more details when compared to the coarse-resolution datasets such as MODIS, thus improves our understanding of human-natural systems interactions (or feedbacks) in urban domains. Such information is very valuable for urban planning in light of rapid urbanization and expansion of major metropolitans at the national and global levels.