Retrieval of Missing Remotely Sensed Tropospheric NO2 Data Using Tensor Completion

Abstract

Missing values in remotely sensed satellite data present a significant challenge for accurate analysis and interpretation of environmental data. Factors such as dead pixel values or cloud coverage can lead to significant gaps in datasets, degrading its overall quality and value. Tensor completion utilizes high-dimensional arrays known as tensors and their low-rank decompositions to recover missing values. This paper demonstrates the application and value of applying Tensor Completion to enhance remotely sensed Nitrogen-Dioxide satellite data and proposes an algorithm for its use. An efficient and accurate recovery method is proposed by leveraging relationships within the tensor and by employing suitable tensor decomposition methods. The proposed algorithm will enhance the analysis, interpretation, and utilization of satellite data. The algorithm is validated on real-world satellite datasets and its superiority demonstrated against existing alternate data recovery methods such as IDW and Kriging.