Analysis and prediction of drought using time series in a number of rain gauge stations in Kheyrabad Basin, Kohgiluyeh and Boyerahmad province

Drought is one of the most important climate phenomena which occurs in all climate conditions and most of the regions on the earth and brings in a lot of injuries to human and natural ecosystems. Therefore, it is important to investigate drought‛s characteristics including intensity, duration and its frequency and to find out their regional risk and forecasting, too. For this reason, some of rain gauge stations in Kheirabad watershed in which their data had been recorded until the end of water years 2011-2012, were selected. Drought conditions for each station were investigated for four different periods including three monthly and one annual periods, using Standardized Prediction Index. Then, drought prediction was performed using SPI values modeling using ARIMA and SARIMA time series. The results showed that in most of the studied time periods, the droughts in the Khyrababad basin are often of medium drought and wet years are of the same type. Study of the persistence of dry and wet periods in selected stations indicates that drought and wet years are more likely to be 1 and 2 years old and have more frequency than other dry and wet periods but the continuity of the normal situation, even for 5 years or more, has been observed in some stations, including the Boyeri and the Nazmakan. The results of drought prediction using time series show that the ARIMA model can better predict SPI values and increase the time interval from 12 months to 48 months more appropriate prediction. Also, the results of the evaluation of the models show that by increasing the time scale from 12 months to 48 months, the slope of the regression line and the MAE and RMSE values are lower.