Evaluating Modern Tools and Large Language Models for Digitising Sea-Level Data from Marigrams
Abstract
Numerous graphical data digitisation tools exist; however, few have been applied to the recovery of historical sea-level marigrams (tidal charts). Using an early twentieth-century marigram from the Clyde Estuary, Scotland, containing one week of sea level observations, we evaluated the performance of 11 digitisation tools and 2 large language models (LLMs) in digitising marigrams. A multi-user testing framework was used to assess the accuracy and speed of each approach. Both LLMs tested, ChatGPT5.5 and Gemini 3.1Pro, were unsuitable for this application under the tested conditions, because they show errors exceeding 10% of the observed tidal range. Conventional digitisation tools performed substantially better, most digitisation tools successfully recovered sea-level data with errors below 1%. Despite this accuracy, these tools remained dependent on substantial human input. Even the fastest approach, DigitGraph, required on average 29 minutes to recover one week of sea-level data. Scaled to a century-long record, this would require more than one working year to complete, making application to large datasets challenging. Consequently, further development of dedicated, automated digitisation algorithms remains necessary.
Related
- Multi-Isotope Signatures in Modern Pig Bones: Evaluating Dietary and Hydrological Controls on Provenance in South Korea
- Evaluating and leveraging large language models in clinical pharmacology and therapeutics assessment: From exam takers to exam shapers
- Sino-US-DrugQA: A benchmark for evaluating large language models in cross-jurisdictional pharmaceutical regulation
- Genomic structural variations distinguishing archaic and modern humans are associated with neurodevelopment
- Fueling the Modern Endurance Athlete: The High-Carbohydrate Revolution, Exogenous Ketones, and an Evidence-Based Approach for the Sports Medicine Clinician
BibTeX
@misc{kenwright2026evaluating,
title = {{Evaluating Modern Tools and Large Language Models for Digitising Sea-Level Data from Marigrams}},
author = {Christian James Kenwright and Ivan D Haigh and Joanne Williams and Robert J Nicholls and Steve McFarland and Marc Becker and Nurul Tazaroh and Seyedeh Fardis Pourreza Ahmadi and Lily Sharp and Callum Slade and Molly Phillips and Sam E Pearson-Smith and Minnie J Darby and Patrick Sharpe and Duo Chan},
howpublished = {EarthArXiv},
year = {2026},
month = sep,
doi = {10.31223/x5p51q},
url = {https://eartharxiv.org/repository/view/15257/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Evaluating Modern Tools and Large Language Models for Digitising Sea-Level Data from Marigrams
AU - Christian James Kenwright
AU - Ivan D Haigh
AU - Joanne Williams
AU - Robert J Nicholls
AU - Steve McFarland
AU - Marc Becker
AU - Nurul Tazaroh
AU - Seyedeh Fardis Pourreza Ahmadi
AU - Lily Sharp
AU - Callum Slade
AU - Molly Phillips
AU - Sam E Pearson-Smith
AU - Minnie J Darby
AU - Patrick Sharpe
AU - Duo Chan
PY - 2026
DA - 2026/09/28
PB - EarthArXiv
DO - 10.31223/x5p51q
UR - https://eartharxiv.org/repository/view/15257/
AB - Numerous graphical data digitisation tools exist; however, few have been applied to the recovery of historical sea-level marigrams (tidal charts). Using an early twentieth-century marigram from the Clyde Estuary, Scotland, containing one week of sea level observations, we evaluated the performance of 11 digitisation tools and 2 large language models (LLMs) in digitising marigrams. A multi-user testing framework was used to assess the accuracy and speed of each approach. Both LLMs tested, ChatGPT5.5 and Gemini 3.1Pro, were unsuitable for this application under the tested conditions, because they show errors exceeding 10% of the observed tidal range. Conventional digitisation tools performed substantially better, most digitisation tools successfully recovered sea-level data with errors below 1%. Despite this accuracy, these tools remained dependent on substantial human input. Even the fastest approach, DigitGraph, required on average 29 minutes to recover one week of sea-level data. Scaled to a century-long record, this would require more than one working year to complete, making application to large datasets challenging. Consequently, further development of dedicated, automated digitisation algorithms remains necessary.
ER -