Abstract: Harmonic analysis, also termed Fourier analysis, is used to analyze time series data. It can decompose a time-dependent periodic phenomenon into a series of sinusoidal functions, each defined by unique amplitude and phase values. The harmonic analysis of NDVI time series can characterize the seasonal change of land cover. In this paper, a harmonic analysis was made on the 1-year MODIS NDVI time series data of Shandong Province, with the harmonic characteristics (additive term, amplitude, and phase image) of several typical land covers in the province extracted. The results showed that the original NDVI time series could be reconstructed by using several harmonic terms with low frequency, and the reconstructed time series profile was smoother, suggesting that harmonic analysis had removed noise when processing time series. Different land covers had different characteristic parameters of harmonic term, being correlated with the seasonal change of land cover. Additive term indicated the arithmetic mean of NDVI; amplitude indicated the variation of temporal NDVI; while phase image indicated the time of peak value. Each land cover had its different harmonic parameters. It was suggested that the results of harmonic analysis could be used for studying vegetation classification and vegetation dynamics.

Key words: Harmful algal bloom area, Terrestrial input, Al