Exploration of Spatial Scaling Effects on Remote Sensing Phenology in the Qinghai–Tibetan Plateau Grassland

Land surface phenology (LSP), a sensitive indicator of ecosystem reactions to climatic changes, plays a crucial role in exchanging materials between terrestrial ecosystems and the atmosphere. Previous studies have utilized various satellite datasets to investigate LSP. However, the phenological scaling effects across different spatial resolutions have not been comprehensively addressed and thoroughly researched. Therefore, we employed two methods (the average and percentile approaches) to aggregate the LSP of the Qinghai-Tibetan Plateau (QTP) grassland at 250 m resolution to 8 km and ecoregion division scales. Based on this, we analyzed the scaling effects of LSP in the QTP grassland. Additionally, this study systematically analyzed factors that would influence the scaling effects. The findings showed that (1) the average absolute difference (AAD) for LSP (250 m) and LSP (8 km, division scale) obtained by the two methods was close. In most areas in the QTP grassland, the percentile approach generated AAD values smaller than the average approach. For pixel and division scales, the optimal aggregation percentile usually ranged from the 40th to 50th percentile instead of a single value; (2) whether in pixel or division scales, the AAD values of SOS were larger than those of EOS; (3) we also found that the terrain complexity (CV) exhibited the most significant influence on scaling effects; followed by altitude, vegetation growth/senescence speed (MP) and vegetation coverage (FVC); terrain slope and aspect had relatively little impact. The findings of this research helped enhance the understanding of the phenology scaling effects of grassland on the QTP.