The ReData Project: Scanning and Digitization of Historical Daily Weather Bulletins through Citizen Scientists

Long-term weather records are often preserved in archives still in paper format. These data come from different sources and represent a crucial starting point for understanding past climate, a reference for validating climate models and an essential input for reanalysis products. This issue affects both data-rich regions and data-sparse areas. Over recent decades, numerous climate data rescue initiatives have been launched worldwide. These projects aim to safeguard historical observations by scanning, transcribing and analyzing them, making the data accessible to the scientific community. One of these initiatives is presented in this study, ReData (Recovery of Data), launched in 2017 by the MeteoNetwork Association in collaboration with the University of Milan. The project aims to scan and digitize daily weather bulletins available from 1879 to 1940, covering Italy as well as some former colonies and surrounding territories, originally issued by the Italian Royal Central Meteorological Office. The scanning phase has been finished, producing a collection of 99,518 pages (about 200 GB) which is now freely available (https://doi.org/10.13130/RD_UNIMI/R1GVKF). Digitization is currently ongoing through the Zooniverse platform (https://www.zooniverse.org/projects/meteonetwork/redata) where volunteers from around the world can contribute by transcribing 12 meteorological parameters for a station of their choice. The transcription of one station day takes about one minute. Since the project launch on Zooniverse in December 2024, more than 10 years of data have been digitized for 37 stations, with an average of about 5000 classification per week and a stable number of 50 to 100 active weekly participants. To facilitate quality control, each measurement was initially transcribed by three different volunteers. Given the high quality of the transcriptions (around 99% of the data show agreement between two out of three independent digitisations), it was later decided to reduce the number of redundant digitisations from three to two, thereby shortening the overall transcription time. Overall, the project provides valuable input for reanalysis models and contributes to a better understanding of historical climate variability over the Italian peninsula, offering substantial scientific and cultural value.