Vegetation index retrieval by coupling optical and SAR images

Abstract

Monitoring spatial and temporal variability of Vegetation Indices (VIs) is important to manage land and water resources, with significant impact on the sustainability of modern agriculture Although algorithms based on optical data give accurate products, cloud cover dramatically reduces the temporal resolution of these outputs. The launch of new Synthetic Aperture Radar (SAR) constellations such as COSMO-Skymed opened new opportunities to develop agro-hydrological applications. Indeed, these satellites may represent a suitable source of data for operational applications due to their high spatial and temporal resolutions (10 m in StripMap PingPong acquisition mode, best revisit time with 4 satellites: 4 images per day at equator; every 7 hours on average at 40° latitude). Although X band is not optimal for agricultural and hydrological applications, reliable continuous assessment of the VIs can be achieved combing optical and SAR images. To this aim, an algorithm was implemented and validated coupling a VI derived from optical DEIMOS images (VIopt) and the crossed HV backscattering σ°HV (PingPong in HV polarization). A correlation analysis has been performed between σ° and VIs measurements taken simultaneously to Cosmo-SkyMed acquisitions in several plots. The correlation analysis was based on: incidence angle, spatial resolution and polarization mode. Results have shown that temporal changes of σ°HV (Δσ0HV), acquired with high angles (off nadir angle, θ > 40°) is characterized by the best correlation with variations of VI (ΔVI). The correlation between ΔVI and Δσ0HV is shown to be temporally robust. Based on this experimental evidence a model to infer VISAR at the time ti+1 once known the VIopt at a reference time, ti and Δσ0HV between times ti+1 and ti, was implemented and verified. The study is carried out over the SELE plain (Campania, Italy) mainly characterized by herbaceous crops. Five couples of COSMO-Skymed and optical DEIMOS images have been acquired between August and September 2011. Data have been collected within the COSMOLAND project (Use of COSMO-SkyMed SAR data for LAND cover classification and surface parameters retrieval over agricultural sites) funded by the Italian Space Agency (ASI). Results confirm that VISAR obtained using the combined model is a satisfactory surrogate of the VIopt.