Threshold-Based Hybrid Data Mining Method for Long-Term Maximum Precipitation Forecasting

In this paper, the application of two data mining techniques (decision tree and association rules) was offered to discover affiliation between several thresholds of monthly precipitation (MP) values of Tabriz and Kermanshah synoptic stations (located in Iran) and de‐trend sea surface temperature (SST) of the Black, Mediterranean and Red Seas. Two major steps of the modeling in this study were the classification of de‐trend SST data and selecting the most effective groups and extracting hidden predictive information involved in the data. The decision tree techniques which can identify the good traits from a data set for the classification purpose were used for classification and selecting the most effective groups and association rules were employed to extract the hidden predictive information from the large observed data. To examine the accuracy of the rules, confidence and lift measures were calculated and compared for different thresholds of precipitation at different lag times. The computed measures confirm reliable performance of the proposed hybrid data mining method to forecast extreme precipitation events considering higher threshold values and the results show a relative correlation between the Mediterranean, Black and Red Sea de‐trend SSTs and maximum MP of Tabriz and Kermanshah synoptic stations so that the confidence between the threshold of 35% of MP values and the de‐trend SST of seas is higher than 70 for Tabriz and 60% for Kermanshah. It was also shown that the geographical location of stations and the distribution of precipitation data affect the measures of the rules and forecasting outcomes.