A general method for searching customized depressions for large radio telescopes using digital elevation models

The construction of large-aperture radio telescopes, such as the Five-hundred-meter Aperture Spherical radio Telescope (FAST), demands careful site selection to balance performance, cost-efficiency, and electromagnetic shielding. Natural karst depressions are often ideal candidates; however, conventional algorithms, particularly the fill-sink approach, struggle to identify single-sided open depressions, which feature partially open boundaries typically less than [Formula: see text] in width. This study proposes a general method for identifying both enclosed and single-sided open depressions using digital elevation models (DEMs). The method integrates customized template matching, cut-and-fill volume evaluation, and positional weight adjustments to prioritize terrain requiring minimal earthwork while maintaining radio shielding. A case study in Qingyuan, Guangdong, China, identified 32 candidate depressions, including 17 enclosed and 15 single-sided open types. Compared with the fill-sink algorithm (precision = 80.0%, recall = 53.3%), the proposed method achieves significantly higher accuracy (precision = 94.1%, recall = 96.9%). This method offers a robust and adaptable solution for large-scale telescope site selection in complex terrains. Moreover, identifying additional suitable depressions provides strategic opportunities to construct multiple FAST-like telescopes within an appropriate distance, enabling interferometric observations that could greatly enhance detection capabilities.