A Dataset with Synthetic Landing Trajectories for Zurich Airport

The archive contains synthetic datasets in npy format generated using a TimeGAN-based model designed to capture a range of aircraft landing trajectories at Zurich airport across different operational scenarios and environmental conditions. Each dataset consists of multiple groups representing distinct patterns or behaviors in the trajectory data. The trajectories are segmented in clusters, and to some of them a smoothing filter was applied.

The datasets incorporate a range of variables critical for modeling aircraft landing behaviors. Continuous variables such as longitude, latitude, and altitude exhibit multimodal distributions, capturing different operational phases and conditions within each cluster. The data is stored in an array format with dimensions (number of samples, sequence length, feature dimensions). Here, the number of samples corresponds to the total number of recorded flight trajectories included in the dataset, while the sequence length represents the duration or the number of time steps over which each trajectory is recorded. The feature dimensions denote the various variables (state vector) measured at each time step, consisting of longitude, latitude and altitude. Categorical variables, such as runway identifiers and cluster labels, follow distributions that reflect operational frequencies, with certain clusters or runways being more common under specific conditions. 

The archive contains the following files:

- 5clust0.npy, 5clust1.npy, 5clust2.npy, 5clust3.np & 5clust4.npy (5 clusters of landing trajectories separated)- ma_5clust0.npy, ma_5clust1.npy, ma_5clust2.npy, ma_5clust3.np & ma_5clust4.npy (5 clusters of landing trajectories separated, moving average filter applied)- ma_3clust0.npy, ma_3clust1.npy & ma_3clust2.npy (3 clusters  of landing trajectories separated, moving average filter applied)- run28_syn.npy & run24_syn.npy (groups of landing trajectories per runway)- ma_run28_syn.npy & ma_run24_syn.npy (groups of landing trajectories per runway, moving average filter applied)- go_around_synthetic.npy (go-around landing trajectories on runway 14)