File-based retrieval of large astronomical catalogs using a HEALPix–R-Tree hybrid index

With the rapid growth of astronomical datasets, efficient and accessible methods for catalog management and retrieval are crucial. Database systems offer powerful queries but incur setup and maintenance overhead, while file-based management is straightforward yet lacks efficient indexing. Conventional single-level sky-partitioning indexes may reach their limit as data volume increases. To address these limitations, this study proposes a two-level spatial indexing framework that combines sky partitioning with secondary spatial indexing. HEALPix provides a coarse-grained, first-level index; within each partition, an R-tree index is built for fine-grained spatial filtering. For a cone search centered at ( 29 2 ∘ , 1 0 ∘ ) with a 0 . 1 ∘ radius on the Gaia DR3 catalog, the proposed method reduces the number of angular-distance calculations by 97.1% (from 342,749 to 13,246) and decreases query time from 9098 ms to 1874 ms at partitioning level 7, compared with single-level HEALPix indexing. Experiments across multiple regions further demonstrate stable performance improvements across different source-density conditions. Deployed on the Lijiang 2.4-meter Telescope’s server, this framework furthermore supports batch object identification and catalog fusion, demonstrating its practical scalability as a general-purpose extension to traditional single-level indexing.