Integration of Satellite Imagery and Climatic Time Series to Assess Multi‐Scale Drought–Vegetation Interactions in Polish Forests

ABSTRACT Drought is a major environmental stressor affecting forest health and productivity, particularly, in temperate regions under ongoing climate change. This study quantified spatial and temporal variability in drought–vegetation relationships across Poland from 2003 to 2023, using 1,379,295 valid 250‐m forest pixels (approximately 8.62 million ha) identified within a stable forest mask derived from CORINE Land Cover datasets. Vegetation dynamics were characterized using the Normalized Difference Vegetation Index (NDVI) from MODIS imagery, while drought conditions were represented by the Standardized Precipitation Evapotranspiration Index (SPEI) calculated over 1‐ to 12‐month accumulation periods. Monthly NDVI series were linearly detrended and pixel‐wise Pearson correlations were computed between detrended NDVI and SPEI across multiple accumulation periods and response months. For each forest pixel, the month of maximum NDVI–SPEI correlation was defined as the critical response month and the corresponding SPEI accumulation period as the optimal SPEI time scale. At the landscape scale, maximum NDVI–SPEI correlation coefficients ( r ) were ≥ 0.50 in 63.37% of valid forest pixels, ≥ 0.60 in 37.15%, and > 0.71 in 11.71%. Critical responses were strongly concentrated in summer, with June–August accounting for 82.23% of pixels; August was most frequent (33.76%), followed by July (31.33%) and June (17.14%). Short SPEI accumulation periods were most common, with SPEI‐2 the dominant optimal SPEI time scale (18.59%). SPEI‐1 to SPEI‐4 collectively accounted for 56.16% of the analysed forest area, while intermediate (SPEI‐5 to SPEI‐8) and longer periods (SPEI‐9 to SPEI‐12) accounted for 18.39% and 25.45%, respectively. Critical response months and optimal SPEI time scales varied across natural‐forest regions, elevation classes and dominant tree species, revealing substantial spatial and ecological heterogeneity. These findings show that Polish forest drought–vegetation relationships cannot be represented by a single drought accumulation period or uniform seasonal window, supporting a multiscale, spatially explicit framework for drought monitoring and regionally differentiated forest management.