Land surface phenology (LSP) is useful to understand patterns of terrestrial ecosystems. Detecting LSP in drylands is more challenging when compared to agricultural and mesic environments due to vegetation heterogeneity, the presence of evergreen and seasonal species, and the dominant role of water (which is often received episodically with variable timing) in determining vegetation growth. In this study, LiDAR-derived vegetation classes are defined to guide and improve the interpretation of LSP metrics extracted using temporally decomposed Landsat fPAR time series. This method was applied to waterholes within the Cooper Creek floodplain, in dryland Australia, which are important for ecological conservation. Results showed that phenology is mostly associated with the recurrent vegetation (approximately 80% of all identified phenological events) in all waterholes. However, during high streamflow periods, the number of phenological events associated with the persistent vegetation greatly increased (up to 40% of the identified events). Non-annual phenology was also identified, with more than one phenological event occurring across a water year during high streamflow periods. The duration of the phenological events of the persistent vegetation exceeded one water year during periods of high streamflow. Phenological differences of the LiDAR-derived vegetation classes occupying the riparian zone of the waterholes were also identified. Streamflow movement across the floodplain exerts an important influence on the vegetation phenology, as suggested by a lag in the phenology when comparing southern and northern waterholes. The method developed herein can be applied to other highly spatially heterogeneous ecosystems where vegetation species simultaneously present permanent and seasonal patterns. Knowing the vegetation life cycle is useful to understand terrestrial ecosystems. However, understanding these cycles in drylands is more complex when compared to agricultural and mesic environments due to heterogeneity and presence of evergreen and seasonal vegetation. In this study, vegetation structure is characterized to guide and improve the interpretation of the vegetation cycles identified using the temporally decomposed time series of a vegetation index. This method was applied at the Cooper Creek floodplain, in dryland Australia. Results showed vegetation cycles are mostly associated with the seasonal vegetation (approximately 80% of all identified cycles). However, during high streamflow periods, the number of cycles associated with the evergreen vegetation greatly increased (up to 40% of the identified events). Multiple cycles were also identified within one water year during high streamflow periods. The duration of the cycles of the evergreen vegetation exceeded one water year during periods of high streamflow. Differences on the life cycle of the different vegetation types were also identified. Streamflow exerts an important influence on the vegetation life cycles by creating a north-to-south gradient. The method developed herein can be applied to other highly spatially heterogeneous ecosystems where vegetation species simultaneously present permanent and seasonal patterns. Vegetation phenology in riparian zones of waterholes in central Australia is studied by integrating LiDAR data and Landsat time series LiDAR-derived vegetation classes assisted in interpreting temporally decomposed Landsat fPAR time series providing ecohydrological insights Reflective satellite mixed pixel problem commonly found in drylands was addressed by extracting the phenology of each vegetation component