Document Type : Research Paper



Predicting climate trends, especially forecasting rainfall, provides managers of different fields with
suitable tools so that considering these predictions; they can devise future-state policies. At this
study, after selecting the most effective climate indices applying PCA method, the effects of largescale
climate signals in seasonal rainfall of basin Maharlu - Bakhtegan were investigated both
simultaneously and by delay through statistical methods (Pearson correlation and cross-correlation
coefficient) and by applying stepwise regression model, regression equation for forecasting rainfall
was offered. The results showed that in cross-correlation between the time series of SPI (dependent
variable) at time (t) and climate signals (independent variable) at time (t-k), only SOI index
concurrently has a significant relationship with rainfall, whereas, most of indices turned significant
with standardized precipitation index with different lag times. In season to season study of the
signals with the standard precipitation index using Pearson's correlation coefficient it was found that
climate signals of spring and summer are not significantly correlated with SPI. Representation
coefficients (R2) and standardized regression effect (Beta) in stepwise regression model showed that
simultaneous and with season to season delays signals (for example: SPI index of autumn with four
previous seasons indexes) at method Pearson correlation have higher relationship with seasonal
standardized precipitation index than the cross-correlation in time (t-k), (which signals of all
seasons given is delay together with than SPI of all seasons) show.