POTENTIAL OF THE SENTINEL-2 RED EDGE SPECTRAL BANDS FOR ESTIMATION OF ECO-PHYSIOLOGICAL PLANT PARAMETERS

摘要

In this study we investigated importance of the spacebornerninstrument Sentinel-2 red edge spectral bands andrnreconstructed red edge position (REP) for retrieval ofrnthe three eco-physiological plant parameters, leaf andrncanopy chlorophyll content and leaf area index (LAI), inrncase of maize agricultural fields and beech and sprucernforest stands. Sentinel-2 spectral bands and REP of therninvestigated vegetation canopies were simulated in thernDiscrete Anisotropic Radiative Transfer (DART) model.rnTheir potential for estimation of the plant parametersrnwas assessed through training support vector regressionsrn(SVR) and examining their P-vector matrices indicatingrnsignificance of each input. The trained SVR were thenrnapplied on Sentinel-2 simulated images and the acquiredrnestimates were cross-compared with results from highrnspatial resolution airborne retrievals. Results showedrnthat contribution of REP was significant for canopyrnchlorophyll content, but less significant for leafrnchlorophyll content and insignificant for leaf area indexrnestimations. However, the red edge spectral bandsrncontributed strongly to the retrievals of all parameters,rnespecially canopy and leaf chlorophyll content.rnApplication of SVR on Sentinel-2 simulated imagesrndemonstrated, in general, an overestimation of leafrnchlorophyll content and an underestimation of LAIrnwhen compared to the reciprocal airborne estimates. Inrnthe follow-up investigation, we will apply the trainedrnSVR algorithms on real Sentinel-2 multispectral imagesrnacquired during vegetation seasons 2015 and 2016.