Exploring spatial and temporal variability of CR index and its uncertainty in Europe using Sentinel-1 data and Google Earth Engine

• Sentinel-1 radar data was used to compute Cross-ratio index (CR) multitemporal maps. • CR spatio-temporal variability over Europe was explored using Google Earth Engine. • CR uncertainty was modelled and mapped using variance propagation law. • Significant CR values differences in time and space domains were modelled and mapped. In this work authors, with reference to Sentinel-1 data, computed and mapped cross-ratio index (CR) over the whole Europe along the year 2023 looking for peculiar behaviors possibly relatable to the main land cover classes. CR presented a flat behavior along the year over urban areas (about −7.5 dB) and over bare soils (about −9 dB). Conversely, CR proved to be selectively changing in vegetated areas supporting the idea of its suitability for vegetation phenology monitoring. Nevertheless, when two acquisitions are compared by differencing it is difficult to separate CR significant differences from noise-related ones. In this context, an uncertainty analysis of CR has not been currently explored in literature. For this reason, authors coupled to the CR mapping an approach aimed at estimating the theoretical uncertainty of CR as obtainable while applying the variance propagation law (VPL). This was implemented directly in GEE to map CR uncertainty over Europe along time. Resulting CR difference uncertainty ( σ Δ C R ) over vegetation macro-class ranges between 1.96 and 2.14 dB. σ Δ C R over urban areas show a constant value around 1.7 dB with no significant variation over the year. σ Δ C R over bare soils show values ranging between 2.14 and 2.57 dB. Resulting uncertainty values can be considered a reasonable threshold over that CR differences (in time) can be assumed as significant. These uncertainties are expected to improve change detection deductions using the a-priori approach based VPL.