Characterizing Forest Species Seasonality Using Time Series of C-Band Sentinel-1 Backscattering Coefficient: A Case Study in the Spanish Central Range

Time series of remote sensing data are essential for understanding the seasonal patterns and long-term forest dynamics. The Sentinel-1 mission offers a unique opportunity to monitor forests with high spatiotemporal resolution using radar technology. This study explored the capability of Sentinel-1 backscatter time series to portray seasonal variations in different Mediterranean forest types of Spain. We used Sentinel-1 GRD data collected over the 2017–2021 period to characterize four dominant species with diverse leaf habits. To reduce noise in time series, we employed the STL decomposition approach. The seasonal plus trend components were retained and Savitzky-Golay filtered to accurately extract the seasonal signatures from a dataset of 50 samples per species. Our analysis revealed marked seasonality in VH polarization backscatter for all species, with specific seasonal patterns varying among forest types. Deciduous forests (European beech and Pyrenean oak) exhibited a decrease in VH backscatter during spring and an increase in late autumn. Conversely, evergreen forests (Holly oak and Scots pine) displayed a distinct seasonal pattern with increasing VH backscatter in spring, peaking in summer, and decreasing in autumn. At VV polarization, no distinct seasonal pattern was observed for deciduous species, whereas for evergreen forests the seasonal pattern resembled that of VH polarization. This study highlights the potential of Sentinel-1 data in characterizing the seasonality of different species. By overcoming the limitations of optical remote sensing in adverse weather conditions, radar data offers new insights into species dynamics, enhancing forest monitoring, and capturing complex seasonal variations of diverse forest types.