SATF: a flight trajectory prediction method incorporating spatial awareness and time–frequency transformation

Flight trajectory prediction (FTP) is crucial for air traffic management, yet current deep learning approaches often struggle with intricate structures and limited accuracy. Although frequency-domain-based methods have achieved state-of-the-art performance for time series tasks, they fail to effecti...

Full description

Saved in:
Bibliographic Details
Main Authors: Jizhao Zhu, Zhuang Zhuang, Bing Han, Liang Zheng, Xinlong Pan
Format: Article
Language:English
Published: Taylor & Francis Group 2025-08-01
Series:International Journal of Digital Earth
Subjects:
Online Access:https://www.tandfonline.com/doi/10.1080/17538947.2025.2512592
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Flight trajectory prediction (FTP) is crucial for air traffic management, yet current deep learning approaches often struggle with intricate structures and limited accuracy. Although frequency-domain-based methods have achieved state-of-the-art performance for time series tasks, they fail to effectively capture the spatial dependencies inherent in flight trajectories with limited features, leading to suboptimal prediction performance. To address these challenges, this paper proposes a novel FTP method called spatial awareness and time frequency (SATF), which incorporates spatial structure awareness and time–frequency transformation. SATF first employs a Spatial Awareness Learner to capture spatial dependencies of flight trajectory and enhance the spatial expressiveness of trajectory representations. Second, SATF uses a Frequency Temporal Learner to convert the flight trajectory from real-valued expressions in the time domain to complex-valued expressions in the frequency domain, enabling the more refined modelling and capture of spatiotemporal dependencies. Extensive experiments are conducted on real-world Automatic Dependent Surveillance-Broadcast (ADS-B) datasets to evaluate the effectiveness of the proposed method. The results demonstrate that SATF outperforms existing mainstream methods across six short-term prediction tasks. Notably, when benchmarked against the state-of-the-art time series forecasting model, SATF reduces the mean absolute error by 40.3%, 52.9%, and 37.2% for longitude, latitude, and altitude predictions, respectively.
ISSN:1753-8947
1753-8955