Fine spatial resolution urban vegetation datasets are crucial for monitoring change in green space and guiding planning and policy initiatives to promote liveable and sustainable cities. Previous studies have demonstrated that finer spatial resolution products are more suited to monitoring urban vegetation than coarse spatial resolution datasets. However, there are differences in the generation of fine spatial resolution datasets that could affect how urban vegetation is represented. To explore how sensitive monitoring and analysis tasks are to the choice of vegetation dataset, a series of comparative analyses were undertaken using three fine spatial resolution datasets in Perth, Western Australia. A technique for quantitative comparison of spatial pattens was used to compare vegetation datasets. There were large areas of Perth where spatial patterns of vegetation were substantially different across datasets. The level of differences in spatial patterns between datasets varied with geographic context such as land use. Elements of a spatial pattern related to grass and tree composition were similar across datasets but there were differences in how shrubs and the configuration of vegetation features were captured. There were differences in the temporal change detection of vegetation patterns across datasets. Spatial patterns of vegetation and land cover generated using the three datasets were used as predictor variables of surface temperatures in machine learning workflows. In some cases, the models learned dataset specific relationships between elements of a vegetation pattern and surface temperature outcomes. In a final modelling analysis, spatial patterns of vegetation were considered as an outcome responding to a disturbance event, a change in dwelling density. The size of the effect of dwelling density change on vegetation patterns varied across vegetation datasets.