Abstract
Accurate mapping of complete landscape gradients is critical to understanding the evolution patterns and landscape structure of socio-natural ecosystems. However, existing datasets focus primarily on the delineation of urban and rural settlements, neglecting the functional attributes of the landscape and neglecting the two gradients of suburban and natural. Therefore, we fused impervious surface and cultivated land data and used morphological spatial pattern analysis (MSPA) to produce a set of year-by-year urban-suburban-rural-natural (USRN) landscape data, and verified the accuracy with the published data. The spatiotemporal accuracy assessment showed that the USRN data had an average R(2) higher than 0.8 with other city-wide products and was superior in spatial integrity to a reference dataset integrating three different data sources. The R(2) of suburban versus socio-economic data is lower than that of urban, confirming spatial heterogeneity of suburban landscapes; natural landscapes have higher consistency with nature reserves. USRN expands on the traditional urban-rural binary structure, helps us assess sustainability goals more carefully and provides a sustained and systematic understanding of landscape dynamics.