Approaches to transcription of weather observations

Datasets of historical weather observations are vital to our understanding of climate change and variability, and improving those datasets means transcribing millions of observations - converting paper records into a digital form. Here we review a variety of methods for doing this transcription.

It is challenging to intercompare transcription projects, as the difficulty of transcription varies a great deal depending on the difficulty of the source documents: hard projects using older documents can take 10-times as long as easy projects using modern documents. The project funding structure made much less difference: citizen science projects performed about as well as large commercial projects, while being much cheaper. Indeed citizen science has been a success in this field - with several successful projects covering a range of different source documents.

The major remaining difficulty is speed. There are so many weather observations requiring transcription that we need a much faster method than anything that has been used so far.

• What this is, and why

• The transcription problem
• Transcription project case studies
  • Weather observations rescued from RN WW2 logs
  • English East-India Company logbooks, 1785-1835
  • oldWeather1&2 - Royal Navy logbooks from 1913-1923
  • oldWeather3 - U.S. Arctic logbooks
  • New UK-region weather observations digitised from the Daily Weather Reports
  • Weather observations from marine expeditions
  • Automated systems
• Conclusions and recommendations

• Authors and acknowledgements

This document is crown copyright (2018). It is published under the terms of the Open Government Licence.