CYCLIM: semi-automated cycle counting for robust age model generation

Counting annual-scale cycles can yield extremely high-precision chronological models. However, this process is typically performed by inspection, often making it time-consuming and subjective. While various software packages exist that automate this process, many researchers still count manually because of its technical simplicity and transparency. Here, we present a new Python-based application that combines the benefits of automation and expert judgement using a semi-automated approach. CYCLIM first detects cycle boundaries using a matched filtering approach before then allowing the user to inspect and refine the output. Additionally, CYCLIM estimates age uncertainty via a noise-based Monte Carlo approach and can incorporate additional chronological ties (e.g., radiocarbon or U-series). We demonstrate CYCLIM’s effectiveness using a previously published palaeoclimate reconstruction and show its additional features, without knowledge of the original age model. In this example CYCLIM found 94.4% of the cycles automatically and required a manual tuning of ~9 cycles per 100. The final age model shows strong agreement with the published record with a mean absolute deviation of 0.79 years.