Prediction of Zonal and Meridional Winds Using the Deep Transformer Model: Case Study at Juanda International Airport

The takeoff and landing phases are the most dangerous phases for aircraft. One of the weather parameters that has a significant influence is wind which determines head wind, tail wind and crosswind. This research aims to improve the accuracy of wind predictions at Juanda Sidoarjo International Airport by using a deep learning model based on the Transformer architecture. The Transformer model was chosen because of its superior ability to capture complex patterns in time series data. Automatic Weather Observing System (AWOS) data which has high temporal resolution and long-term data can be utilized using deep learning modeling. The transformer model was chosen because its architecture can capture the complexity of wind direction and speed at Juanda Sidoarjo International Airport. The model results have good value in predicting zonal and meridional wind direction components. The correlation value between prediction and actual shows a significant value, namely 0.98, while the MSE and MAE values are below 0.29 and 0.36.