How a student’s EO journey is powering flood resilience 

When Gloria Pucoe, first began working with satellite data, her focus was on mapping floodwaters after disasters had already struck. Today, her research is evolving moving from looking back to looking ahead, with the aim of predicting floods before they happen. 

As a master’s student in e-Science (Data Science) at Sol Plaatje University in South Africa, Gloria is developing forecasting models that could provide flood-prone communities with early warnings, especially in areas with limited data. Her work started during the Big Data Africa School, where she was part of Team Aqua6, a group comprising students from different disciplines, tasked with mapping flood extent using satellite imagery and machine learning. That experience was a turning point. 

“In South Africa, I have seen people lose everything to floods. That’s what made this more than just a technical problem for me,” she says. 

Learning through collaboration: The Big Data Africa School 

Held in Cape Town in March 2025, the Big Data Africa School brought together 25 participants from across eight countries, pairing them into multidisciplinary teams. Gloria, whose background is in data science, was grouped with peers from GIS and environmental science disciplines. Participants were intentionally grouped to reflect the kind of cross-sector collaboration that’s essential in real-world EO projects. 

Team Aqua6 focused on flood mapping in Beira, Mozambique, a region heavily impacted by cyclones and extreme weather. Using Synthetic Aperture Radar (SAR) data from Sentinel 1 available through the DE Africa platform, the group detected water even through cloud cover. By processing and analysing backscatter data, applying K-means clustering, and comparing results against ESA’s WorldCover dataset, they created flood maps that could support disaster response. They also experimented with pixel-level detection using Fully Convolutional Neural Networks and generated vulnerability maps using elevation and land-use data. 

But the story doesn’t stop at Mozambique. Back home in South Africa, devastating floods in KwaZulu-Natal and the Eastern Cape have left communities displaced and reeling. For Gloria, flood resilience is personal. 

From mapping to forecasting 

That project sparked Gloria’s master’s research, which moves beyond detection into forecasting. Traditional flood prediction models often rely on assumptions that don’t hold up in fast-changing or data-limited environments. Gloria’s approach is transformer-based deep learning models, explainable AI, and synthetic data generated with time-series GANs. 

Her research has four goals: 

1. Test the accuracy and flexibility of Transformer models in data-poor settings. 
2. Use explainable AI tools like SHAP to make model outputs more transparent to stakeholders, showing not just the what, but the why behind predictions. 
3. Create synthetic flood scenarios that preserve spatial and temporal patterns. 
4. Compare how real and synthetic data affect model performance. 

It is a bold shift that reflects both technical ambition and a clear real-world need. 

Turning research into action 

Gloria’s ambition is clear: she doesn’t want her research to stay on paper. She hopes to work with disaster response agencies and local governments, turning models into practical tools that help people prepare. But she’s also realistic about the challenges ahead. 

“I’m still learning how to connect the research side with the implementation side,” she says. “I don’t want to work alone. I want to collaborate with agencies that are already helping communities.” 

Scaling solutions like hers requires partnerships, access to reliable Earth observation data, and shared infrastructure. And while Gloria’s current focus is on algorithms, she’s already thinking about how those systems can plug into real-time early warning efforts. 

From the flooded streets of Beira to the data labs of Sol Plaatje University, Gloria’s journey demonstrates what’s possible when young African innovators are given the tools and the trust to tackle real-world problems using space-based data. Her work is still in progress, but this underscores how DE Africa’s investment in capacity building equips Africa’s next generation of EO leaders.