Software for analysis of vegetation indices dynamics

Abstract

The article describes a software package designed for processing and analyzing of vegetation indices based on satellite data. A program is developed in the free-distributed environment Octave. The input data of the program are vegetation indices (MODIS NDVI and EVI). The data are pre-processed by removing the peak values using the Savitsky-Golay filter and a seasonal decomposition by a moving average. To exclude the influence of snow cover and the period without vegetation, data of the growing season are used in the calculation. The vegetation period is determined by phenological parameters of the beginning and end of the growth period. The program determines the phenological parameters by a double logistic function, an asymmetric Gaussian function, and a fourth-degree polynomial to approximate the original time series. After the preliminary processing, a linear trend of the series, its statistical significance, the minimum and maximum values, coefficients of variation, etc., are calculated. The correlation function between the vegetation indices and the time series set by the user is also implemented. The output data are exported in raster and text formats. In conclusion, an example of using the software for calculating the NDVI trends in the Uda River basin in Transbaikalia is given.