Estimation of Meteorological Parameters Using Data Driven Methods

In this study, two different methods were analyzed on time series analysis for the forecasting of temperature and precipitation data. Random Forest, Long Short Term Memory Network, a specialized deep neural network model used for algorithm and time series analysis, from the average precipitation data collected for 4 different provinces in the Eastern Anatolia Region of Turkey was used. Although the data used for the study differ for each province, it covers an average of 54 years of data. The successes of the algorithms used on the data sets were compared in the modeling made after the data preprocessing stages. The results obtained as a result of the analyzes showed that the studies on temperature estimation gave more positive results due to the large number of data. In the average precipitation data used in precipitation analysis, it was observed that the success of the models used was lower due to the low number of data. In temperature analysis, the LSTM model produced the most successful results.