Collaborative Climate Data Rescue Workshop with visiting University of Kentucky Students

 

Collaborative Climate Data Rescue Workshop with visiting University of Kentucky Students

Kevin Healion, Caoilfhionn D’Arcy and Simon Noone.

The Climate Data Rescue-Africa (CliDaR-Africa) project, launched in 2022, involves second-year undergraduate students at Maynooth University (MU), Ireland, in a participatory, classroom-based initiative to digitize unique historical meteorological records from data-sparse regions of Africa. This CliDaR project was a collaboration between ICARUS and the MU Geography department and forms part of the students 2^nd year mandatory Methods of Geographical Analysis module. These historical data images are sourced from the African Centre of Meteorological Applications for Development (ACMAD) collection, which contains data dating back to the late 1800’s, see Figure 1 for more details (Noone et al., 2024).

 

Figure 1. Map shows spatial coverage of the ACMAD African data images by countries covered (highlighted in blue). Right panel presents an initial inventory with number of ACMAD stations and temporal coverage by country. Image taken from Noone et al, 2024.

 

Over the past three years, MU students have digitized over 64 years of data in Madagascar, Guinea and Central African Republic. This is nearly 450,000 of unique meteorological observations, which is amazing! The new data enhances coverage in these data-sparse regions, supporting more robust climate extremes assessments and future climate projections crucial for adaptation planning. We have been encouraging other organizations/universities to explore the potential for their students to get involved in the CliDaR-Africa project.

Recently, the CliDaR-Africa project has begun receiving international recognition. Earlier this year, we were excited to hear from Dr. Ann Kingsolver at the University of Kentucky (UOK), who told us that she was bringing fifteen students to MU for a summer school this June. After reading the Noone et al. (2024) paper, she said the UOK students were really keen to get involved in some hands-on climate data rescue. So we were delighted to welcome Dr. Kingsolver and the fifteen students to Maynooth University in June 2025.

 

 

Figure 2 shows the UOK students hard at work digitizing the data.

 

How the project was organised and managed.

To kick things off, the CliDaR-Africa team gave the students some background on ACMAD data, why it is important, previous CliDaR-projects and information on Madagascar. Students learned that Madagascar is a large island in the Indian Ocean, just off the east coast of Africa, and is renowned for its incredible biodiversity. Positioned south of the equator and straddling the Tropic of Capricorn, Madagascar has a distinct seasonal climate, with its rainy season running from November to March, and a drier period from April to October. However, Madagascar has several localized microclimates. The students also explored how Madagascar’s geography plays a big role in its climate variability. Factors like proximity to the coast and the island’s north-south mountain range contribute to regional differences in rainfall and weather patterns. The team also pointed out a major gap in historical data for the region, with very little data available from before the 1970s. The team also showed how Madagascar is increasingly vulnerable to extreme weather events, especially cyclones, which are becoming more frequent and severe. These storms have far-reaching impacts on the country’s infrastructure, economy, livelihoods, and ecosystems.

For the four-day workshop, the team specifically tailored the CliDaR-Africa project to enable the UOK students to digitize data from Marolambo in Madagascar. Marolambo, is situated in the Atsinanana region in eastern Madagascar, close proximity to the coast with an amazing National Park spanning over 95,000 hectares. This region experiences heavier rainfall than other regions, including the southern arid areas and western regions of Madagascar. The students were taught about life in Marolambo and how important the local weather patterns are for people's livelihoods. The majority of people in Marolambo are dependent on small-scale farming for subsistence and trade. Rice, cassava, and maize are the primary crops grown in the region. Any delay to the rainy season, which starts in November, can have a major impact on the production of these crops. Marolambo is a biodiversity hotspot with very unique ecosystems including many endemic flora and fauna species. Again, any changes to the weather patterns could have a knock-on effect for this area of natural importance.

Each UOK student was allocated six months of data images to digitize, collectively covering approximately four years (1951–1954), including double keying for quality assurance. Half of the students worked on one set of images, while the other half keyed in a matching set. The observations were variously taken at 3 times a day and once a day intervals for each climate variable (see example in Figure 3). The first sheet contains precipitation and evaporation observations, the second sheet contains temperature, cloud and humidity observations and the third sheet contains pressure and wind observations. These UOK students were tasked with entering the data exactly as it appeared in the images into the corresponding cells of an Excel template, these specifically match the layout of cells in each of the three data image form types (Figure 4). Each column ended with a total of the entered values, acting as a first quality check. Students used built-in formulas to verify these totals, and if something didn’t add up, they went back to double-check the data in that column. It was a simple but effective way to catch any mistakes early on. As the completed data sheets were double-keyed, this allowed the team to run Python scripts to compare both datasets and automatically flag any mismatched values, a great way to catch and correct errors before finalizing the data.

 

Figure 3 example of the 3 form types for one month of data during January 1951 at Marollambo , Madagascar. The first sheet on the left contains precipitation and evaporation observations, the middle sheet contains temperature, cloud and humidity observations and the sheet on the right contains pressure and wind observations.

Figure 4 presents an example of the excel template forms that students use to enter the values from the data image sheet.

 

Output and initial results

In the final hour of the last day, we wrapped things up by sharing the data the students had digitized so far. Even though it only covered a few years, we were able to highlight some initial analysis results, giving everyone a glimpse of the valuable insights this work can uncover. It was a great way to celebrate what had been accomplished in just a few days! To showcase the real potential of this data, we shared analysis results from a previously digitized station at Nosy Varika, covering the years 1954 to 1964. The data revealed several extreme events, including heatwaves, severe drought, and occurrences of intense rainfall. Most strikingly, it uncovered details of several long-forgotten devastating tropical cyclones that hit Madagascar during March 1959. To bring these impacts to life, we showed the students old_video_footage, historical newspaper clippings, and old photographs that documented the widespread impacts across Madagascar during that time (see Figures 5 and 6). It was a powerful reminder of just how vulnerable Madagascar is to extreme weather events and how valuable rescued climate data can be in understanding and preparing for them.

Figure 5 Images showing the impacts after the March 1959 cyclones that hit Madagascar. Top left image is the harbour quarter in the capitol Tananarive, the top right shows Tananarive’s Lake Mandroseza, bottom left image is the rural town of Sonierana, in the Analanjirofo Region and bottom right shows the Anosizato zone in Tananarive. Images taken from (Aldegheri, 1959)

Figure 6 Online newspaper articles from the Canberra Times Thursday 2^nd April 1959 page 6, Belfast Newsletter published on March 27^th, 1959, page 5, The Belfast Newsletter published on April 4th, 1959, page 8 and the Irish Examiner published on April 4th, 1959, page 9. Showing reports on the extreme weather events during March 1959 in Madagascar (Irish Newspaper Archives). 

This devastation was caused by a series of tropical cyclones, which frequently impact Madagascar between January and April. The lack of sub-daily and daily weather data in Madagascar before the 1970s means that it is difficult for climate scientists to analyse these extreme weather events. However, thanks to the digitization efforts carried out by the students, in the future it will be possible for scientists to analyse and reconstruct these extreme weather events and study how climate change is affecting the characteristics of the cyclones.
Thanks to their efforts of the UOK students over 33,000 meteorological observations were digitized covering the period January 1951 to July 1954 (Figure 7). Many thanks again to the students for the great work and for such positive constructive feedback. In the near future, the digitized data will be made openly and freely available to data users via the Copernicus Climate Data Store, and the National Oceanic and Atmospheric Administration’s (NOAA) National Centers for Environmental Information (NCEI) data centre in the US (https://www.ncei.noaa.gov/).

 

Figure 7 Plot shows six main variables at station Marolambo, Madagascar 1951-1954 digitized by the UOK students. Maximum and minimum temperature variables are original daily values. Wind speed and station level pressure derive from the arithmetic average of the sub-daily values for plotting visualization purposes.

The CliDaR–Africa project is just scratching the surface when it comes to digitizing the millions of climate data images in the ACMAD collection. There's still a long way to go, and we’d love to have more universities and institutions join the effort! If your team is interested in getting involved, feel free to reach out. Please contact Simon Noone @: simon.noone@mu.ie. We are happy to share images, templates, and all the resources you’ll need to get started.

  

References

Aldegheri, M . Cyclones of March 1959 in Madagascar. In: Hydrological Yearbook of Overseas France: Year 1957. Paris: ORSTOM, pp. 33-55,1959. Accessed 16th June 2025 online at: https://horizon.documentation.ird.fr/exl-doc/pleins_textes/divers12-05/16864.pdf

Noone, S., D’Arcy, C., Donegan, S., Durkan, W., Essel, B., Healion, K., Hersbach, H.,  Madden, S., Marshall, J., McConnell, L., Mensah, I.,  Scroxton, N., Thiesen, S.,  Thorne, P. Investigating the potential for students to contribute to climate data rescue: Introducing the Climate Data Rescue Africa project (CliDaR-Africa). Geoscience Data Journal, 00, 1–17. Available from: https://doi.org/10.1002/gdj3.248, 2024