Monitoring Vegetation Dynamics Inferred by Satellite Data Using the PhenoSat Tool

摘要

PhenoSat is an experimental software tool that produces phenological information from satellite vegetation index time series. The main characteristics and functionalities of the PhenoSat tool are presented, and its performance is compared against observed measures and other available software applications. A multiyear experiment was carried out for different vegetation types: vineyard, low shrublands, and seminatural meadows. Temporal satellite normalized difference vegetation index (NDVI) data provided by MODerate resolution Imaging Spectroradiometer and Satellite Pour l'Observation de la Terre VEGETATION were used to test the ability of the software in extracting vegetation dynamics information. Three important PhenoSat features were analyzed: extraction of the main growing season information, estimation of double growth season parameters, and the advantage of selecting a temporal region of interest. Seven noise reduction filters were applied: cubic smoothing splines, polynomial curve fitting, Fourier series, Gaussian models, piecewise logistic, Savitzky–Golay (SG), and a combination of the last two. The results showed that PhenoSat is a useful tool to extract NDVI metrics related to vegetation dynamics, obtaining high significant correlations between observed and estimated parameters for most of the phenological stages and vegetation types studied. Using the combination of SG and piecewise logistic to fit the NDVI time series, PhenoSat obtained correlations higher than 0.71, except for the seminatural meadow start of season. The selection of a temporal region of interest improved the fitting process, consequently providing more reliable phenological information.