Preface: Advancing deep learning for remote sensing time series data analysis

Abstract

This special issue explores the burgeoning field of deep learning for remote sensing time series analysis. The 20 contributed papers showcase diverse applications, including land cover mapping, change detection, atmospheric and biophysical/biochemical parameter retrieval, and disaster monitoring. The articles demonstrate a variety of approaches to address the challenges of irregular time series, such as data compositing, harmonic modeling, and direct ingestion of irregular data using recurrent and attention-based networks (e.g., LSTMs and Transformers). Several studies highlight the potential of integrating physical models with deep learning to improve model trustworthiness and interpretability. Looking ahead, we identify key future directions: the development of globally representative benchmark datasets with time series labels; the creation of readily available, operational time series products and models; the exploration of multi-modal and foundation models tailored to remote sensing time series; and more sophisticated integration of physical knowledge within deep learning frameworks. This collection highlights current progress and fosters innovation in time-aware deep learning for Earth observation.