Overcoming Optical Gaps: Evaluating SAR–Optical Consistency for Cotton Phenology

Abstract. Reliable monitoring of cotton phenology is challenged by cloud-induced gaps in optical satellite observations, especially during key growth stages. This study evaluates the temporal consistency between C-band Synthetic Aperture Radar (SAR) features from Sentinel-1 and vegetation indices from Sentinel-2 across eleven adjacent cotton fields in Didim, Türkiye, during the 2024 season. Sentinel-1 dual-polarization (VV, VH) backscatter, polarimetric decomposition parameters (Entropy, Alpha, Anisotropy), and Stokes-derived metrics (e.g., g₀) were extracted from Single Look Complex (SLC) data processed in SNAP. Sentinel-2 Level-2A imagery was used to compute seven vegetation indices (NDVI, EVI, ARVI, NDRE, NDWI, MSAVI, GCI) after cloud masking via the Scene Classification Layer. A ±2-day temporal matching strategy aligned SAR and optical acquisitions, enabling inter-field correlation analysis. Results show strong, consistent relationships between VH backscatter and chlorophyll–biomass–oriented indices, with NDVI–VH (mean r = 0.930), ARVI–VH (0.928), and EVI–VH (0.924) exhibiting synchronized seasonal trajectories. Stokes g₀ correlated highly with biomass- and moisture-sensitive indices, including MSAVI (0.863), EVI (0.860), and NDWI (0.858), highlighting its utility as a cloud-resilient surrogate for canopy status. In contrast, H/A/α parameters demonstrated weak coupling to optical indices (e.g., GCI–Entropy mean r = 0.115), reflecting their sensitivity to structural scattering mechanisms not directly linked to pigment dynamics. These findings indicate that combining VH backscatter and g₀ with optical indices provides a robust, cloud- tolerant monitoring framework, while H/A/α offer complementary structural diagnostics. The proposed approach is scalable using free, global datasets and is transferable to other phenology-driven agricultural monitoring applications.