Akademik Veri Yönetim Sistemi
Meteorology and Atmospheric Physics, cilt.138, sa.2, 2026 (SCI-Expanded, Scopus)
Drought is among the most dangerous natural disasters. Drought, defined as a deviation from the average amount of rainfall, negatively affects living things when it occurs. However, there are places where drought positively affects living things. Certain regions of Norway exemplify such conditions. In this study, drought prediction models were created for the regions of Bodo, Oslo, and Tromso in Norway, where droughts can have positive effects. To create these models, the Extreme Gradient Boosting (xGboost), the Long-short Term Memory Network (LSTM), the Support Vector Machine (SVM), the Multi-layer Perceptron (MLP), and the Random Forest (RF) machine learning algorithms were used. The Standardized Precipitation Evapotranspiration Index (SPEI) data spanning from 1901 to 2023 were utilized for drought anaylsis. In addition, cross correlation method was used to determine the input structures in the models, M01, M02, M03, M04, M05 and M06. The evaluation of model performance was conducted using five recognized statistical metrics: the Correlation Coefficient (R), the Root Mean Square Error (RMSE), the Nash-Sutcliffe efficiency (NSE), the Performance Index (PI), and the Kling–Gupta efficiency (KGE). As a result of the analyzes made by creating 6 different input structures, the most successful model was determined for Bodo as LSTMM01 (R = 0.9439, NSE = 0.8816, KGE = 0.8576, PI = 0.2642, and RMSE = 0.3437), for Oslo as SVMM04 (R = 0.9467, NSE = 0.8954, KGE = 0.8555, PI = 0.2696, and RMSE = 0.3231), for Tromso as SVMM04 (R = 0.9456, NSE = 0.8860, KGE = 0.8823, PI = 0.1921, and RMSE = 0.3373). The best performance metrics across all regions were determined in the M04 model with SVM. This study should assist decision-makers in improving the quality of life in the region.