Article Title

Uydu verilerinden karar ağaçları kullanarak orman yangını tahmini


Destruction of forests, which are important for all vitality, with fires, seriously threatens the safety of life and property. Wildfires occur by natural ways and conscious human behavior. The prediction and early detection of wildfires will allow for rapid intervention and prevention. In the literature, meteorological data and remote sensing data were used to predict forest fires. In addition, by using meteorological data, the behavior of existing wildfires could also be determined. In this study, wildfires were predicted from data received from satellites. Wildfires were predicted by using Normalized Differential Vegetation Index (NVDI), Land Surface Temperature (LST) and Thermal Anomaly (TA) data calculated from downloaded satellite data. Decision trees were used to make predictions using mentioned data. 70% of the data in the data set were used for the training of decision trees. The trained model was tested with the remaining 30% data. Training and testing process were repeated 10 times with different data and the average performance of the proposed model was determined. Occurred fires were correctly predicted on the average of 98.62% sensitivity in the experiments. For the predictions in all trials, the actual situation was determined with an average of 93.11% accuracy