Key Takeaways
- Catherine Nakalembe uses satellite data and AI to aid African smallholder farmers.
- Existing satellite models are often inaccurate for diverse, small-scale African farms.
- Ground-level data collection is crucial for training context-specific AI models.
- True innovation adapts technology to specific problems, delivering actionable insights.
- Connecting data to farmers requires understanding local contexts and accessible channels.
- Securing funding for critical ground-based data collection is becoming difficult.
Deep Dive
- Catherine Nakalembe, a 2025 TED Fellow, is a satellite food security specialist aiding smallholder farmers in Africa.
- She utilizes satellite imagery and machine learning to assist farmers in countries including Uganda, Kenya, Tanzania, Zambia, Mali, and Senegal.
- Over 8,000 satellites orbit Earth, providing daily imagery to map crops and identify areas impacted by weather events.
- This work helps farmers prepare for and mitigate impacts of natural disasters like floods, droughts, and crop failures.
- Nakalembe's team collects ground-level imagery using GoPros on motorcycles or carts, capturing crops like maize, beans, and cassava.
- Over 5 million images were collected in two weeks across Western Kenya to train accurate, context-specific AI models.
- Nakalembe personally installed four soil moisture calibration stations in Kenya, Uganda, and Tanzania.
- These stations ensure data accuracy for new satellite missions, particularly for Africa, a continent highly affected by drought.
- Existing satellite models, often trained on large, single-crop fields from Europe or the US, are inaccurate for diverse African farms.
- Nakalembe identifies a "failure of translation" where data models do not connect to farmers' real-time needs.
- Current data models often lack relevant training data for African farmlands, hindering accuracy.
- Insufficient resolution of satellite imagery also poses a barrier for small, complex fields typical in Africa.
- Data offers actionable insights, such as predicting imminent droughts, which can help farmers avoid wasting resources on planting.
- Aggregated data on continuous droughts can inform decisions about investing in infrastructure like irrigation systems.
- Nakalembe's multi-faceted role bridges academia, government, policy, and farmers to ensure data specificity meets ground-level needs.
- Connecting data insights requires understanding different local contexts and perspectives to make information relevant.
- Scaling contextualized data requires prioritizing ground-level utility over academic validation and understanding the 'why' behind its use.
- Accessible information channels, such as text messages, calls, or community meetings, are essential for farmers.
- A project uses radio programs to translate complex information into relatable terms for farmers, contrasting with academic bulletins.
- Deep listening and intentional observation are crucial for understanding local contexts and building knowledge with communities.
- Securing funding for essential ground-based "root sense" data collection is becoming increasingly difficult.
- Traditional capacity-building methods, such as in-person learning exchanges, are no longer feasible, hindering collaborative problem-solving.
- Understanding challenges from the perspective of those directly affected is critical, as practical application is difficult.
- There is an urgent need to educate and prepare farmers for extreme weather events, like devastating hailstorms, to improve their situations.