Rainfall prediction with intelligent systems: A case study of Tunceli province

Meteorological events hold significant importance in human life as they affect many aspects of our lives. Predicting meteorological events will greatly facilitate our lives as extraordinary weather events influence the course of our lives. The predictability and accurate forecasting of rainfall data are crucial, offering numerous advantages from an engineering perspective. With the help of past rainfall data, this prediction process can be achieved through Artificial Neural Networks without the need for specific mathematical equations. In this study, a model has been created using Artificial Neural Networks by leveraging raw data obtained from meteorology. For this purpose, monthly data from 2012 to 2023, covering a period of 12 years, has been utilized. For the 12-year period from January 2012 to December 2023, monthly average temperature (°C), total open surface evaporation per month, average relative humidity per month (%), average 10 cm soil temperature per month, average wind speed per month (m/s), monthly global solar radiation rates, daily total sunshine duration monthly average, and monthly average rainfall (mm=kg/m2) data were employed. Using input and output parameters for the ANN model, an appropriate simulation modeling of the system was conducted in the MATLAB environment. The numerical predictive results obtained from our ANN model yielded certain error values: RMSE=0.3251, COV=0.3887, MAE=0.2141, and an R2 value of 0.98548. In the modeling study conducted with UBSA, the following error values were obtained: RMSE=0.3087, COV=0.3654, MAE=0.2562, and an R2 value of 0.5165. These error values indicate a good approach to rainfall prediction. When comparing the ANN model with the UBSA model, it is observed that the ANN model performs more successfully.