For decades, resource planning has suffered from a top-down bias, beginning with regional climate forecasts, moving to national energy grids, then to river basin authorities, and finally, almost as an afterthought, arriving at the farm gate. But what if we flipped the hierarchy? 

What if the most intelligent signal for the WEF nexus came from a smallholder farmer in the Tana River Basin who notices that her maize is wilting two weeks earlier than last season? 

The EPIC Africa project has made remarkable strides in integrated modelling, with the CLEWs framework, OSeMOSYS tools, and participatory Transition Spaces in Nairobi, Kenya, and Accra, Ghana representing genuine progress. However, a fundamental disconnect remains: most resource models treat agriculture as a “sector” to be managed rather than a network of millions of daily decisions to be understood. To achieve true systems thinking, we must connect farms directly to the WEF nexus, and institutions like KALRO Kenya are showing the path forward.

Learning from EPIC Africa’s Transition Spaces

The Transition Spaces co-created within the EPIC Africa project in the Tana River Basin and the Volta Basin offer a concrete example of the need bring farmers (and not just farm intelligence/data) into the WEF process. What emerged were not just data points but new modelling questions that current WEF frameworks struggle to answer. Participants asked: Why do our national models not represent urbanization’s rapid transformation of agricultural land and labour? Why do they fail to capture the feedback loops of agricultural transformation when energy access, water availability, digital climate services, and extension-based farm management are introduced simultaneously as bundled inputs? These questions exposed a gap: even well-intentioned nexus models often assume linear, separable relationships between sectors, whereas farmers experience them as indivisible. The Transition Spaces demonstrated that when farmers enter the modelling process as co-diagnosticians rather than data sources, they reframe the very problems models are built to solve.

Logic of African Agriculture

The inverse logic of African agriculture makes this connection even more urgent. Unlike highly consolidated agricultural systems dominated by large commercial enterprises in of regions like Europe and North America, Africa’s food system is fundamentally decentralised. It is shaped by approximately 33 million smallholder farms whose day-to-day decisions on production, resource use, technology adoption, and market participation collectively determine food availability, rural livelihoods, and agricultural resilience across the continent. Some of the pain points include: What to plant? When to plant? How to plant? Each of these decisions is, in essence, a WEF nexus trade-off. Choosing a high-yield maize variety (Food) requires more water (Water) and potentially mechanised processing (Energy). Switching to drought-tolerant sorghum reduces water demand but may require different storage infrastructure. Yet virtually none of these farm-level trade-offs enters the national or basin-level models that determine water allocations, energy pricing, or agricultural subsidies. This is the paradox of African systems thinking: the most important data points exist at the smallest scale, and models systematically ignore them.

Limitations of Current WEF Frameworks

Existing WEF frameworks rely heavily on aggregated national datasets, sector-level assumptions, or historical averages. While useful for broad planning, these inputs introduce significant limitations when applied to agricultural realities. 

“Agriculture is often treated as a static demand function rather than a dynamic, adaptive system, flattening differences in soil quality, crop choice, input use, and yield outcomes into averages that obscure production diversity.” 

As a result, nexus models often describe idealised resource systems that fail to reflect the complex, uncertain landscapes farmers navigate. Africa’s food systems are dominated by rainfed dependence and climate sensitivity, diverse cropping systems on small landholdings, rapidly changing access to inputs, limited irrigation and storage infrastructure, and exposure to pests and weather extremes. This variability means agricultural planning cannot be accurately represented without direct farm-level intelligence. To improve the relevance of the WEF nexus, farms must become data-generating nodes in a national or regional intelligence system.

Agricultural Intelligence in the WEF Context

What does agricultural intelligence mean in the WEF context? It refers to the continuous integration of multi-source data to understand, predict, and optimize agricultural systems. This includes farm-level production data such as crop types, planting dates, yields, land size, and input utilization; environmental and climate data including rainfall patterns, soil moisture, and evapotranspiration rates; geospatial intelligence from remote sensing on crop distribution and land use; livelihood and market data on farmgate prices, commodity flows, and farmer income variability; and risk signals from pest surveillance, drought alerts, and yield anomaly detection. When combined, these datasets form a (near) real-time agricultural intelligence layer that can feed directly into WEF decision support models, enabling precision water allocation through accurate estimation of irrigation needs, energy demand optimisation by aligning infrastructure with agricultural cycles, realistic food production forecasting that improves national security assessments, climate risk integration that simulates shock propagation, and sub-national decision support for decentralised resource governance.

The Broken Feedback Loop

Despite significant investment in the digitalisation of agriculture, African countries still face structural challenges. Farmer data is collected but not continuously updated; agricultural surveys are periodic rather than continuously driving seasonal decision making. The more continuously available remote sensing data is underutilised in policy systems, extension services are not fully digitised or integrated, and national planning models rarely aggregate and ingest insightful farm-level datasets. This results in a broken feedback loop between farms and national planning systems, leaving the WEF nexus approach incomplete and less practicable. To address this gap, we need a new multi-scale architecture that connects farms directly to larger systems models at basin level: digital farmer identity systems with unique identifiers linking farmers and plots; sensor and remote sensing integration combining IoT devices, satellite data, and field surveys; digital advisory and feedback loops that collect response data from farmers; interoperable agricultural data platforms allowing exchange between national registries and planning models; and AI-enabled aggregation layers that transform farm-level data into national and regional intelligence.