A tensor-based interpolation method for sparse spatio-temporal field data. Issue 2 (3rd May 2020)

• Record Type: Journal Article
• Authors: Li, D. S.; Yang, L.; Yu, Z. Y.; Hu, Y.; Yuan, L. W.
• Is Part Of: Journal of spatial science. Volume 65:Issue 2(2020)
• Page Start: 307
• Page End: 325
• Publication Date: 2020-05-03
• Subjects: spatio-temporal interpolation -- sparse tensor -- canonical polyadic decomposition; Geographic information systems -- Periodicals; Cartography -- Periodicals; Surveying -- Periodicals; Geodesy -- Periodicals; Photogrammetry -- Periodicals; Cartography; Geodesy; Geographic information systems; Photogrammetry; Surveying; Periodicals; 526.05
• DOI: 10.1080/14498596.2018.1509740
• Languages: English
• ISSNs: 1449-8596

Abstract

ABSTRACT: In this paper, a tensor-based interpolation method for spatio-temporal field data is proposed. The observed data are organized as a sparse tensor, which is then represented as the decomposition of a canonical polyadic (CP) model. The latent factors, abstracting the underlying structure of sparse data, are exploited by introducing a weighted optimization solution. With the reconstruction of these factors, the values of missing data can progressively be recovered. The simulation experiments suggest that our method can interpolate spatio-temporal data with various levels of sparseness. The data can even be interpolated accurately and stably when 90% of the data are missing.