Predicting the Meteorological, Hydrological and Agricultural Droughts in Tehran Using Wavelet Method

According to the heavy damage caused to humans by the occurrence of drought, it is important to predict the drought as accurately as possible.The wavelet-neural network integration method is one of the most accurate methods for this important.There are several categories of drought, including meteorological,hydrological,and agricultural drought.In this study,authors was tried to determine the optimal wavelet for predicting different types of drought.daily precipitation data,daily discharge,and satellite imageryrelated to Tehran from 1969 to 2016 were use as raw data to calculate the indicators.The wave transformations (WT) conversion method and PNN neural network have been used to predict droughts.From each time series of drought, wavelet transformations were performed using haar and bior1.1waves,and the prediction was make by neural network.It was found that the regression coefficient and error concentration for meteorological drought using haar wave are 0.68and 0.033and regression coefficient and error concentration for hydrological drought using haar wave are 0.76and0.066and regression coefficient and error concentration for agricultural drought using haar wave are 0.9269and0.1515.Then, the prediction of any kind of drought was done with bior1.1wave and it was found that the regression coefficient and error concentration for meteorological drought using bior1.1wave are 0.7116and 0.992.and the regression and concentration coefficientThe error for hydrological drought using bior1.1wave is0.14147and 0.0329and the regression and error concentration coefficient for agricultural drought using bior1.1 wave are0.82049and0.0016.The results showed that for meteorological drought, the bior1.1wave appeared better and gave us better results.But in two types of hydrological and agricultural droughts,it was found that the haar wave gives us better results.