성·연령 구조를 반영한 격자 기반 인구추계 알고리즘 개발과 재현성 평가

초록

Regional extinction risk in South Korea is increasing due to low fertility, population aging, and youth outmigration. However, existing population projections are mostly provided at administrative-unit levels, making it difficult to capture fine-scale spatial differences within regions. This study developed and evaluated a 1 km grid-based population projection algorithm that reflects sex- and five-year age-specific population structures. Using SGIS 1 km gridded census population data, the algorithm generated the population aged 0–4, applied cohort survival for those aged 5 and over, adjusted the open-ended age group of 100 years and over, and then matched the results to resident registration totals by si-gun-gu, sex, and five-year age group. Based on 2018 data, the reconstructed 2023 grid population showed a national total-population MAPE of 33.80% and MAPE-R of 18.30%. Errors were highest for the youth population and lowest for the elderly population. MAPE-R was also used to reduce the influence of extreme values in si-gun-gu-level comparisons. This study provides a grid-based population estimation method that reflects demographic processes more directly than conventional top-down allocation approaches.

키워드