Abstract:

One of the most important disasters threatening world geography with Global Warming and other human factors is drought. Drought directly/indirectly affects the living ecosystem all over the world, starting from aquatic organisms. The most basic feature of drought, which is one of the most common natural disasters in our country, is that it occurs over time and affects large masses of people. The emergence of drought depending on time creates the possibility of predicting drought. Drought indices are used to define the drought problem. Drought can be classified with these indices, which use meteorological parameters such as precipitation, temperature, and humidity. In this study, the Standard Precipitation Index (SPI) data of the last 13 years in the Aegean Region was examined and forward-looking estimation was made using the Dynamic Mode Decomposition (DMD) method. DMD is a machine learning method that is based on Singular Value Decomposition and can be run on time series that vary depending on time and location. The data set studied was obtained from the World Weather Online climate data portal. The data set has been preprocessed and analyzed with the DMD algorithm. The estimation performance of the DMD method for drought analysis of the Aegean Region is presented in comparison with the real values.

Keywords: