Stochastic and time series drought forecast using rainfall oscillations in arid and Semi-arid environments
By: Abbasali Vali, Mostafa Dastorani, Adel sepehr, Chooghi Bairam Komaki
Key Words: Stochastic models, Rainfall forecasting, Arid-Semi Arid, SPI index.
J. Bio. Env. Sci. 9(1), 245-256, July 2016.
The importance of water supplies in the world, underscores the need for estimating and forecasting the trend of meteorological phenomena, understanding atmospheric phenomena and its trend in economic management. This includes optimization of profitability and productivity impact, especially in arid and semi-arid schedules. Conversely, climate and rainfall are highly non-linear and complicated phenomena, which require non-linear mathematical modeling and simulation for trusted accurate prediction. In this study, monthly rainfall data were obtained from 10 synoptic stations from 1985 to 2014. Thereafter, R software was employed in predicting the height of rainfall in 10 synoptic stations (2003 to 2014) using monthly height of rainfall data (1985 to 2014). In this research, five models (AR, MA, ARMA, ARIMA, and SARIMA) with 12 different structures were tested. After deciding on the optimal model to be used for each station, rainfall was forecast for 120 months (2014 to 2024) and then for the years 2014 and 2024 iso-rainfall maps were outlined. From the findings of this research, it was observed that in 80% of data, ARMA (2,1) had better results than the other models and according to the simulated and predicted rainfall by time series models, the drought situation was evaluated using standardized precipitation index (SPI). The result thus revealed that in comparison to 2014, severe drought will have decreased by the end of 2024.