Prediction of Net Primary Production Changes in Different Phytogeographical Regions of Iran from 2000 to 2016, Using Time Series Models

Vegetation cover is an important component of terrestrial ecosystems that changes seasonally. Accurate parameterization of vegetation cover dynamics through developing indicators of periodic patterns can assist our understanding of vegetation-climate interactions. The current study was conducted to investigate and model vegetation changes in some phytogeographical regions of Iran including, Khazari, Baluchi, semi-desert, temperate steppe, warm semi-steppe and arid forest and to compare their stochastic behavior. To study the vegetation changes the net primary production (NPP) was used, based on the products of Moderate Resolution Imaging Spectroradiometer (MODIS) sensor (MOD17A2 series). Seasonal Auto Regressive Integrated Moving Average (SARIMA) time series model was used for modeling NPP. The Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) of time series showed that these areas were static with seasonality in 12-month periods. It also showed that the vegetation in Khazari region was more stable, which indicates a stable environmental condition with the least deviation in water, light and nutrients. We also found that most of the vegetative regions of Iran can be modeled with SARIMA and its changes can be reliably predicted. Estimated models for Khazari (Root-Mean-Square Error, (RMSE) = 0.12, R2 = 0.87, Mean Relative Absolute Error (MARE) = 0.083) and semi-desert (RMSE = 0.12, R2 = 0.95, MARE = 0.048) were more suitable models than other regions.