DECOMPOSING TREE AND GRASS PHENOLOGY IN DRYLANDS: ANALYZING THE MIXED PIXEL EFFECT WITH NEAR-SURFACE AND SATELLITE OBSERVATIONS

This study investigates the phenological dynamics of a heterogeneous dryland ecosystem in Alice Springs, Australia, through the integration of high-resolution phenocam observations and multi-sensor satellite data. By decomposing the phenological responses of grasses and tree canopies, we found that grass phenology, despite covering less than 30% of the satellite-pixel footprint, exerts a dominant influence on the overall land surface phenology (LSP) at the landscape scale, with stronger correlations evident in coarser-resolution satellite data. The analysis further shows that while lower spatial resolution sensors capture broad-scale phenological patterns effectively, they often mask finer-scale variations critical for accurate ecosystem assessments in drylands. These findings highlight the need for techniques to mitigate the mixed-pixel effect in LSP studies, enhancing the precision of remote sensing applications.