Abstract:
Tree-ring width (TRW) is key for the identification of forest growth responses to climate, but field sampling inherently limits spatial and temporal coverage. Satellite-derived vegetation indices (VIs) can bridge this gap, yet most canopy-growth studies continue to rely on Normalized Difference Vegetation Index (NDVI), a greenness metric long known to saturate in high-biomass canopies and to decouple from stem growth under stress. Unlike previous studies relying on single vegetation indices and static seasonal metrics, we implemented an approach that integrates multiple vegetation indices across several accumulated temporal windows to capture intra-annual tree growth dynamics and stress responses. Rolling-sum correlations for the growth-year and previous-year linked canopy signals from greenness-, structural-, wetness- and pigment-related indices to annual TRW. The analysis used a 23-year (2000–2022) 8-day Landsat seamless data cube (SDC) at 30 m resolution covering 24 different forest sites in western and central Germany. Fifteen trees were investigated at each site, from mixed and pure stands consisting of the tree species Fagus sylvatica, Quercus petraea, Quercus robur, Pinus sylvestris, Pseudotsuga menziesii, and Larix decidua. Greenness indices, including the NDVI, consistently underperformed. Structure, including Near Infrared of Vegetation Index (NIRv), and wetness indices produced higher and more temporally targeted correlations, especially in pure stands and for beech during summer. Coniferous and mixed plots showed weaker, broader signals, while pigment-ratio indices proved comparatively most informative in mixed stands. High-resolution temporal analysis of the 8-day SDC Landsat time series revealed species-specific peak correlation windows, enabling the identification of critical periods driving radial tree growth. The results of this study by moving beyond greenness-only approaches and single-value, growing-season analyses reveals much better performance compared to studies averaging to growing-season means that blur these peaks and consequently report weaker or no correlations between tree growth and the satellite signal.
