VUB software for rescuing historical weather data receives TIME recognition among world’s top-400 Best Inventions of 2025

VUB software for rescuing historical weather data receives TIME recognition among world's top-400 Best Inventions of 2025

Brussels, 10 October 2025 - A groundbreaking tool developed at the Vrije Universiteit Brussel (VUB) has been recognised by TIME as one of the Best Inventions Special Mentions of 2025. MeteoSaver, an AI-based software created by VUB researcher Derrick Muheki, received a special mention in TIME's annual list of the world's most impactful innovations. Earlier this year, Nature also featured Muheki's work in a News article. "It's incredible to see this kind of recognition. It shows how important technology is in helping us protect our planet's climate history", says Muheki.

Our understanding of climate change depends on the quality of the data that underpins it. Yet for vast regions of the world, especially in the Global South, reliable weather observations remain scarce or incomplete. "Because of a lack of information about past temperatures, the forested heart of Africa remained a blind spot in the 2021 IPCC report, which assessed the pace and severity of global warming around the world", explains Prof. Wim Thiery, climate scientist at the VUB Water and Climate Department. "Recovering and digitising this information is crucial if we want truly global and accurate climate models."

From handwritten logs to digital data

This is where MeteoSaver comes in. Developed by Muheki, a doctoral researcher in the bclimate group led by Prof. Thiery, the software uses machine learning to automatically transcribe handwritten weather records into digital datasets in a matter of minutes. By simply photographing an old weather log, the program can process the image and generate structured spreadsheets that can be analysed, shared, and safely stored. This is a vital step toward preserving fragile data that might otherwise be lost to time, floods, or fire.

Fieldwork deep in the Congo Basin

To test the system, Muheki travelled to the Yangambi branch of the National Institute for Agronomic Research (INERA) in the Democratic Republic of the Congo during the summer of 2023. Over two months, he scanned nearly 9,000 pages of historical weather logs, ​ some dating back to the 1960s, under challenging conditions. Reaching the remote site required several flights, a river journey along the Congo, and hours by motorbike over unpaved roads. Electricity was unavailable, so Muheki brought enough rechargeable batteries to power his camera equipment throughout the campaign.

Back at the VUB, Muheki processed the photos with the MeteoSaver software with over 90% accuracy, a level comparable to human transcription and unprecedented for handwritten meteorological tables.

Scientific impact and global collaboration

According to Prof. Thiery, MeteoSaver represents a new frontier in climate science: "Climate models are only as good as the data we feed into them. With MeteoSaver, we can finally close long-standing data gaps and improve our understanding of regional climate change, particularly in areas that have historically been overlooked".

The MeteoSaver software is open source and can be used free of charge by anyone in the world. There are already partnerships growing with many institutes, including the Royal Meteorological Institute of Belgium (RMIB) and the University of Reading. The Royal Netherlands Meteorological Institute (KNMI) in the Netherlands is testing to apply MeteoSaver to historical archives preserved there.

Through months of refining and training the AI algorithms that make up his MeteoSaver tool, Muheki has managed to make it work with many handwriting styles, paper size, and maintenance conditions of the archives, making his tool extremely robust. "It took months to get the software to work, particularly because these types of AI tools had not been used to transcribe tabular handwritten data before. Think of all the various handwriting styles and table structures one can find in archives. Training the software to recognise hundreds of handwriting styles was the hardest part. But now the software is working really well and we hope to apply MeteoSaver to many more historical archives around the world. There are so many places we could bring MeteoSaver to".

Muheki is now analysing over 900,000 newly digitised weather observations from the Congo Basin to better understand how local weather extremes such as droughts and heatwaves have evolved under human-induced climate change. "What is interesting is that we will be able to quantify whether and how extreme events like heatwaves or severe droughts happening today intensified because of man-made climate change. Data rescue at the level of quality provided by MeteoSaver enables answering a lot of new interesting questions about our changing climate, particularly in the Global South", highlights Thiery.

In recognition of the development of this incredible tool for historical weather data rescue, TIME has selected Muheki's MeteoSaver as a special mention in their TIME Best Inventions of 2025, marking it part of the 100 special mentions of the year. "I'm deeply grateful for this recognition, as this means that MeteoSaver will gain more visibility, helping us in our quest, and many others', to rescue essential weather records around the globe. Even in the Global North, we are also sitting on incredibly rich archives that still need to be unlocked", concludes Muheki.

More information:

• MeteoSaver in the TIME's best inventions Special Mention of 2025: https://time.com/collections/best-inventions-special-mentions/7320840/meteosaver
• Nature News article about work of Derrick Muheki: "'Revolutionary' AI tools rescue old weather data to improve climate models" https://www.nature.com/articles/d41586-025-02798-y
• Preprint of scientific article describing MeteoSaver: https://egusphere.copernicus.org/preprints/2025/egusphere-2024-3779/ (note: paper is still under review and its content may therefore still be updated)

Read the VUB Today interview with Derrick Muheki and Wim Thiery here:

We would like to acknowledge that these results wouldn't have been possible without the close collaboration with colleagues in Ghent University, who have worked in Yangambi for decades and brought expertise in text mining, and our DRC local partner INERA.

Contact:

• Derrick Muheki, [email protected], +32470346423
• Advisor & head of the bclimate group Prof. Wim Thiery, [email protected], +3226293029
• Valorisation Manager Dr. Marie Cavitte, [email protected], +32470192415

Read more