Climate change impacts on the Nahavand karstic springs using the data mining techniques

Karst resources are sensitive to environmental changes, especially climate change. In this research, monthly data (years 1994–2020) were collected for five karstic springs, namely Famaseb (Sp1), Faresban (Sp2), Ghalebaroodab (Sp3), Giyan (Sp4), and Gonbadkabood (Sp5), located in the Nahavand plain, west of Iran. Data mining models such as KNN, SVM, and M5tree were used to simulate the discharge of springs. The results obtained, based on two statistical indices, correlation coefficient ( r ) and normalized root mean square error (nRMSE), the M5tree model was more accurate than the other models and was selected to simulate the discharge of the springs. The r -value was equal to 0.736, and nRMSE was 0.113. In the later stage, discharge of the springs was projected for four time periods: 2021–2040, 2041–2060, 2061–2080, and 2081–2100 under two scenarios, RCP8.5 and RCP4.5. The results showed that in the monthly survey, the ratio of the variation in the discharge of the springs compared to the historical period in the Sp1 and Sp2 springs had the most variations. It was found that in November, there was an average increase of + 250% in the investigated periods. Also, the highest decrease compared to the historical period of these two springs was observed during May, which were about − 37 and − 70%, respectively. In springs Sp3, Sp4, and Sp5, like the previous two springs, the most fluctuations occurred in November and May, respectively. Graphical abstract