Impact of rainfall anomalies on Fourier parameters of NDVI time series of northwestern Argentina

Abstract

This paper describes a method to detect the impact of rainfall anomalies on vegetation phenology, in terms of timing (phase) and greenness, by using Fourier series to fit a time series of Normalized Difference Vegetation Index (NDVI) observations. The study was conducted in the northern semiarid region of Argentina, where rainfall is the driving factor of vegetation phenology. A 9-year time series of monthly NDVI Global Area Coverage (GAC) images, obtained with the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR), was split into nine series of 12-monthly images, each corresponding to a yearly growth cycle. A Fast Fourier Transform (FFT) algorithm was applied to each cycle, and derived parameters were analysed according to rainfall anomalies for irrigated and rainfed crops, grasslands and native forest. Derived Fourier parameters were: mean NDVI, amplitude and phase. Both negative and positive rainfall anomalies had a significant impact on the Fourier parameters. Amplitude and phase were the most sensitive parameters. Droughts modified the monomodal structure of the yearly cycle by decreasing the contribution of the 12-month periodic component and increasing the contribution of the 6-month component. The impact of drought on the Fourier parameters was highest for rainfed crops. Yearly values of Fourier parameters for grasslands and native forest were affected by prevailing hydrological conditions over the previous year.