Successful completion of the AI for EW4All Pilot Project in Malawi

Around one-third of the global population lacks access to adequate multi-hazard early warning systems, with the greatest gaps observed in Least Developed Countries and Small Island Developing States. The United Nations Early Warnings for All (EW4All) initiative aims to ensure universal access to weather warnings by 2027. WMO is leading Pillar 2 of EW4All, focused on the detection, observation, monitoring, analysis, and forecasting. "MET Norway is supporting the aims of the EW4ALL initiative through capacity building, improvements in weather forecasting, and through advocacy for equitable access to technological advancements," states Roar Skålin, the Director-General of MET Norway. “The WMO CREWS pilot project on AI for EW4All in Malawi is a great example of this.” 

The transformative potential of AI in weather forecasting

Recent technological developments in AI have shown its transformative potential to improve weather forecasting and to empower National Meteorological and Hydrological Services (NMHSs) worldwide to improve their early warning capabilities. ”Compared with Numerical Weather Prediction (NWP) models, AI-based weather prediction (AI-WP) models have a much lower operational cost and complexity.” Håvard Alsaker Futsæter, the lead developer from MET Norway for the AI EW4All pilot project, explains. “Combined with faster run times and easy-to-use tools, this could potentially enable all NMHSs to create early warning systems based on high-resolution weather forecasting data.”

Democratising weather forecasting: the AI EW4All pilot project

However, AI-WP tools are mostly developed within countries that already have strong technical capacity, either by private companies or by NMHSs. To help democratise weather forecasting, MET Norway, ECMWF and DCCMS, undertook the EW4All pilot project in Malawi, to assess how AI-WP can specifically support Malawi’s weather forecasting needs.

Given the ambitious timeline for the pilot project, developments from several ongoing initiatives were used to set the foundation for this work. Notably, the pilot project was built on the Forecast-in-a-Box, implemented by ECMWF and co-funded by Destination Earth, and the AI-WP model, Bris, by MET Norway. It also was linked to a WMO Integrated Processing and Prediction System (WIPPS) pilot project by MET Norway and ECMWF, which aims to integrate the above-mentioned tools into WIPPS. 

Building on the multi-year successful collaboration between MET Norway and DCCMS in Malawi

Most importantly, this work would not have been possible in this timeframe without the existing good collaboration between DCCMS and MET Norway, via the capacity building project, SAREPTA, funded by the Norwegian Agency for Development Cooperation (Norad). The primary objectives of SAREPTA is strengthening weather and climate services in various countries including Bangladesh, Vietnam, Ethiopia, Malawi, Mozambique, and Tanzania. Teferi Demissie, coordinator of SAREPTA at MET Norway, further describes that “the project focuses on capacity building to support sustainable development and enhance resilience to climate-related challenges through targeted training and infrastructure development.” 

“MET Norway and DCCMS have worked together since 2019,” Lene Østvand, a researcher at MET Norway and the project manager of the AI EW4All pilot project, reflects. “The collaboration started with a pilot project investigating how the MET Norway API as a Digital Public Good could be used to improve early warnings in Malawi, through the development of a weather app similar to Yr. Since then, we have had many successful workshops together. These often involve a lot of hard work, but over the years, we have developed a close collaboration where we also genuinely enjoy spending time together and learning from each other.”

AI EW4All pilot project: an interdisciplinary approach

Building operational capacity involves both technical and human aspects, with a specific focus on how AI-WP can transform operational duties, as well as the perceptions of and expectations about these transformations from the DCCMS team. Amos Mtonya, deputy director at DCCMS, is one of the main collaborators and champions of the pilot project. “The AI for EW4All pilot project is a huge opportunity for Malawi’s DCCMS to embrace cutting-edge technology and innovations in meteorological services. It directly supports our commitment to providing Early Warnings for All, and its success will serve as a valuable model for other countries.”

Throughout the various phases of the AI EW4All pilot project, interviews and dedicated focus group sessions were held with operational forecasters at DCCMS.The goal of these discussions were to map expectations about the consequences and opportunities of using AI-WP. Focus groups also functioned as a space for voicing concerns on the operational levels, and establishing vested interest in AI-WP, as both a tool and a process. “Transformative technology, such as AI-WP, thrives at the intersection of technical innovation and human understanding,” states Dina Abdel-Fattah, a researcher at MET Norway, who worked on these elements of the pilot project. “By balancing both dimensions, we can ensure meaningful adoption of these technologies as well as their sustainable, lasting impact.”

Path forward

Overall, this pilot project showcased the dynamic field of AI-WP. The technical novelty of the work, combined with a focus on institutional capacity and international collaboration, shows the potential of AI-WP to transform weather forecasting, helping to meet the growing need for strengthened early warning capabilities. 

The results of the pilot project will be discussed in more detail during the Panel - AI for weather forecasting: closing the digital divide with AI? at the upcoming WMO Ex-Congress (20 - 23 October 2025)  in Geneva. The panel will be livestreamed on Tuesday, 21 October 2025, from 9:00 to 10:30 CEST.

MET Norway looks forward to sharing our results and continuing our collaboration with ECWMF, DCCMS, WMO, and others on potential further phases of the project moving forward.