Akademik Veri Yönetim Sistemi
11TH AZERBAIJAN CONGRESS ON LIFE, ENGINEERING,MATHEMATICAL, AND APPLIED SCIENCES CONGRESS, Baku, Azerbaycan, 21 - 22 Haziran 2025, cilt.1, ss.605-614, (Tam Metin Bildiri)
Machine learning methods are used in many fields in the literature. These methods are also widely preferred in the field of hydrology. Machine learning methods provide very effective results in predicting any natural disaster. Although these natural disasters appear as extreme rainfall, extreme heat and strong winds, droughts are actually one of the most studied topics in this field. Droughts are among the most dangerous and insidious natural disasters because they are not noticed when they occur. This increases its importance in terms of tracking, monitoring and prediction. However, there are places in the world where droughts can have positive effects, contrary to what is generally thought. One of these places is Oslo, which is located in the polar region and is affected by snow and ice most of the year. Due to these adverse weather conditions in Oslo, droughts can positively affect this region. For this reason, Oslo was chosen as the study area in this study. The Standardized Precipitation Evapotranspiration Index (SPEI) method, which is frequently preferred in the literature, was used in drought monitoring and evaluation. In machine learning, the Long Short-Term Memory (LSTM) method was preferred. Apart from these, four different model input structures were created to be used in machine learning. The Correlation coefficient (r), the Nash–Sutcliffe efficiency (NSE), and the Root Mean Square Error (RMSE) performance criteria were used to compare model performances. According to the results obtained from the analysis, the LSTMM02 model was ahead of the other methods in terms of performance with the performance values of r = 0.9372, NSE = 0.8732 and RMSE = 0.3271. This study contributes to future planning.