Abstract
This study presents a robust seven-step framework for curating and analysing extensive radiocarbon (14C) datasets, optimised to construct accurate Summed Probability Distribution (SPD) models. Applied to a comprehensive dataset of 4,657 14C dates from 582 archaeological sites in the Southern Levant spanning the last 50,000 years, this framework systematically consolidates data by eliminating duplicates, refining archaeological periodisation, and integrating key environmental variables like phytogeographic zones and natural regions. Through meticulous 14C classification, outlier identification, and model assessment, the resulting dataset and models are refined, transparent, and adhere to FAIR principles for accessibility. This approach enhances the reliability of SPD models in studying ancient population trends and human-environment dynamics, offering a versatile tool adaptable to similar large-scale datasets across regions.