Meteorological drought prediction in Konya with machine learning algorithms

Drought is a significant natural disaster that occurs under various meteorological and environmental conditions and has substantial impacts on living organisms. Unlike other natural disasters, its effects are felt in the long term, making drought one of the most damaging natural disasters in the long term. This thesis, a meteorological drought analysis was conducted using the Reconnaissance Drought Index (RDI) in Konya, located in the Central Anatolia Region, which is known as the driest region in our country. RDI values were determined using monthly total precipitation data and potential evapotranspiration values calculated from average temperature data obtained from nine meteorological stations between 1985-2021 in the study area. The severity, duration, and intensity of drought were determined for 3-, 6-, 9-, and 12-months periods using Run analysis for each station. The longest drought period lasted 45 months and occurred at the Konya airport station in the 12-month period. Subsequently, maps were created using Geographic Information Systems (GIS) for RDI values and meteorological parameters, including precipitation, temperature, relative humidity, wind speed, and air pressure. Using meteorological parameters, RDI values, and pixel values obtained from maps of these meteorological parameters, various input combinations were created to develop drought models using machine learning algorithms. The most suitable model results were obtained from models developed using pixel values. Finally, the trends of RDI values in all drought models were determined using the Mann-Kendall trend test.