A subdivision-based electromagnetic map model for unified multi-dimensional electromagnetic characterization and fusion

The scientific characterization and information fusion of complex electromagnetic spectral spaces remain fundamental challenges because existing models often rely on inconsistent electromagnetic representations and loosely organized data structures across dimensions. To address these two problems, in this paper, a subdivision-based electromagnetic map (SEM) model for the unified encoding and fusion of electromagnetic information across six essential dimensions: space, time, frequency, power, polarization angle, and propagation angle, with GeoSOT providing the hierarchical space–time address. The SEM model further constructs a grid tensor data model and an attribute encoding algebra, facilitating rapid retrieval and high-speed computation of massive heterogeneous electromagnetic data. Moreover, employing a grid attention optimization mechanism, the model supports accurate and interpretable spatio-temporal prediction of electromagnetic fields. Compared with conventional baseline models, experimental results show that the SEM model improves encoding/decoding, visualization, and retrieval/compression efficiency by factors of 4, 7.7, and 15, respectively. Thus, SEM enables complex electromagnetic characterization and organization, extends Digital Earth research from conventional geographic, remote-sensing, and environmental layers to the electromagnetic dimension of Earth space.