SILT, Inc.

SILT collaborated with CSIDNet (Climate Sensitive Infectious Disease Network) and LDRI’s new Research and Innovation Climate Hub (RICH) for an early morning community-facing session at Global Partnership for Sustainable Data’s Global Data Fest in Nairobi on June 3, 2026.

Building Ecosystems for Environmental Data Use: Trust, Stewardship, and Sustainability featured lighting talks from Dr. Angela Okune (CSIDNet), Dr. Rendani Mbhuva (AfriClimate AI), Leonida Mutuku (RICH) and Mark Irura (Language AI and Digital Governance Specialist). Each expert shared reflections from the field vis-a-vis the gap between environmental data producers and data use. Following this, Rayya El Zein facilitated group discussions revolving around three central themes: trust, equity, and sustainability.

Okune’s remarks, around the central provocation “Data are not fish” argued that the gap between data producers and users persist after nearly two decades of open environmental data. This metaphor underscores how producers release data expecting it to thrive independently. However data, unlike fish, require ongoing care, feeding, nurturing after release. Who will maintain data months and years (decades?) into the future? Okune underscored the experiment with CSIDNet as building the social infrastructures and relationships needed to sustain collaborative open environmental and public health data.

Professor Mbuvhha’s presentation, based on his efforts at AfriClimate AI centered how Africa has 1/8th of the recommended weather station density per WMO standards. Satellite observations are available but the significant spatial and temporal biases in these make data unreliable. This data scarcity is all the more urgent in the context of accelerating climate change, and all the more necessary for the survival of small agricultural operations. Agricultural companies need localized forecasting capabilities but multi-country participation is required for full forecasting potential. His evidence-based dataset recommendations are different for different use cases, using the RainCheck Africa data dashboard to synthesize across datasets.

Irura lifted up experiences working with Kenyan farmers and other agricultural workers, identifying three main problems in the data production and usage cycle. 1) Data fragmentation across systems that don’t communicate; 2) Breakdown in trust where extractive data collection fails to provide communities with benefits; and 3) Ownership disputes, with communities asserting competing rights to given data. These three challenges underscore power dynamics with governments and intergovernmental institutions at the top and farmers (often women) at the bottom. He emphasized the need for transparent communication about data use and benefits to the community and the need to ground formal, state, and inter-state knowledge systems in traditional and indigenous knowledge frameworks.

Finally, Mutuku turned the group’s attention to policy advances at the state and international level. She highlighted UN Environment Assembly Resolution 9 on the environmental sustainability of AI systems and the needed focus on quality, accessibility, and interoperability of environmental data. Her work has championed data sharing frameworks that connect grassroots efforts to governmental and intergovernmental policy. For example, of >60 climate datasets identified, less than 40% are usable in AI systems. The key issues presented include non-interoperability, different collection frequencies, and sustainability challenges.

Following the lighting talks, the (full!) room was invited to break into smaller groups, get to know one another, and surface what challenges and solutions they have seen/are experimenting with around developing trust, building equity, and ensuring sustainability of data in their work.

Trust. Across the groups, participants underscored that transparency about data limitations is essential for building trust (whether with public sector partners, communities, or other parties). We discussed how trust functions at multiple layers — individual, organizational, and across data streams. As an overlap with equity and sustainability, participants pointed to the co-production of data as critical in building sustainable relationships. Importantly, trust is processual and built over time — proof of delivery and meeting standards can build this credibility incrementally. In the agricultural sector in Kenya, participants pointed to the central role of extension workers as a critical bridge between technology and farmers.

Equity. Participants reflected that digital divides continue to leave people behind in data access; the effect of this is particularly acute in environmental data that is often most useful to folks in rural areas where digital access is limited. Participants emphasized the need for inclusive design from the project onset: communities must be partners, not subjects, in research. The benefits of user-centric design also surfaced, again drawing on the context of Kenyan agricultural sector and focusing on farmer decision-making needs.

Sustainability. Finally, the discussion on sustainability centered on the critical relationship between industry and the public sector. Private companies are increasingly holding a large share of environmental data and must charge for access (for their own sustainability). But, participants noted, there is an appetite for collaboration and it is not always clear how data they share with the public sector is used, if it is useful, or what impact it has. One example of this that was shared was WeatherStream – there is a 24 hour data lock, and then it becomes open access). Participants stressed the need for better public-private collaboration frameworks.

In sum, there was a great appetite to discuss trust at many levels — between researchers and communities, between businesses and public-serving institutions, and between scientists and entrepreneurs wherever they work. Veterans may have heard echoes in debates about open data, science, and software over the decades. Importantly, however, there was little need for translation around critical concepts and there was ready enthusiasm about the promise of more collaboration, translation, and exchange.

Further Reading

Mutuku, Leonida and Christine Mahihu, (2014). “Open data in developing countries : understanding the impacts of Kenya open data applications and services.” http://hdl.handle.net/10625/56300

Mutuku, Leonida et al. (2014). “ShambaConnect: Case Study on the Hybrid Design of an Application for Kenyan Extension Officers.” https://dl.acm.org/doi/10.1145/2662155.2662191