Drought is one of the most complex phenomena to interpret within forecasting systems. Not because of the difficulty in detecting its physical signal, but because of the distance separating that signal from its effects. A deviation in precipitation is not, in itself, an impact: it becomes one when it propagates through agricultural, water, energy and social systems, interacting with conditions of exposure and vulnerability that determine profoundly different consequences.
It is within this continuum between meteorological phenomenon and impact that the concept of impact-based forecasting emerges: an approach aimed at anticipating not only what will happen from a climatic perspective, but what will happen to exposed systems, translating forecasts into expected impacts and possible actions. But what does this approach imply when applied to drought?
Co-development as a scientific methodology: a response to the challenges of drought
Drought can be understood as a systemic risk, whose effects propagate across sectors, territories and actors, influencing food security, water availability, energy systems and socio-economic dynamics. From a scientific perspective, the challenge lies not only in forecasting the hazard, but in modelling its consequences. It is a slow and cumulative process, characterised by delays, memory effects and non-linearities, in which the relationship between climatic signal and impact is strongly context-dependent.
This implies that the development of impact-based early warning systems cannot be reduced to a matter of selecting indicators, but requires a methodological framework capable of integrating heterogeneous dimensions – physical, environmental and socio-economic – into a representation of risk that is coherent with the operational context, local socio-ecological configurations, sectoral priorities and specific decision-making needs.
For this reason, system design requires the direct involvement, from the earliest stages, of continental and regional institutions, climate centres, humanitarian organisations and scientific actors, as an integral part of building an impact-based early warning system.
This approach is not simply an operational choice, but a methodological necessity that takes the form of an explicit co-development process, built through continuous interaction between scientific and operational actors, in which not only the solutions, but also the questions themselves are openly discussed: what does “impact-based” mean in the context of drought? Which sectors should be prioritised? Which information is truly relevant to support decision-making at different levels?
Building a continental impact-based system
Within the African context, this transition takes on relevance. It is within this framework that the integration of impact-based information into the Africa Drought Watch is being developed, as part of the Programme for a Continental Coordination, Early Warning and Action System in Africa. The objective is to transform this monitoring system – currently based on physical indicators – into a tool capable of supporting drought risk management at continental scale by integrating data, models and institutional capacities, while anticipating impacts and supporting decision-making.
This transition has not been approached as a simple technical extension of the existing system, but as a redefinition of its methodological architecture, structured not as a centralised modelling exercise followed by a validation phase, but as an iterative process of co-development and co-production.
With this vision, the Technical Workshop for the Co-Development of Impact-based Drought Analytics took place in Nairobi in December 2025. The workshop initiated a phase of methodological convergence in which assumptions, technical options and operational needs were explicitly discussed, compared and progressively aligned.
A shared and multisectoral representation of risk
One of the outcomes of this process has been the definition of a multisectoral drought risk structure articulated around four main domains: agriculture and livestock, water resources, energy and population, migration and security.
This structure does not respond to a purely theoretical classification, but emerged through the interaction between actors with different responsibilities and operational needs, reflecting the necessity of representing risk in ways that are meaningful for those required to manage it.
At the same time, the process highlighted the interconnected nature of these domains, in which impacts propagate across sectors, generating cascading effects. In particular, the dimension related to population and security was recognised as a cross-cutting component whose level of risk depends largely on the dynamics developing within the other sectors.
Building usable information under conditions of uncertainty
The co-development process also addressed methodological choices related to data and modelling. The availability of exposure and vulnerability information, as well as the quality of impact data, varies significantly across regions and sectors. This makes it difficult to construct models based on quantitative relationships between hazards and observed impacts.
In this context, risk construction has been approached as an iterative process in which different approaches – from indicator combinations to the definition of relative severity classes – are used to produce information that is both scientifically coherent and operationally relevant.
The challenge is therefore to coherently represent these interconnections within complex decision-making systems.
From co-production of knowledge to the capacity to anticipate
The adopted approach reflects a deeper transformation in the way early warning systems are built. Forecasting is no longer conceived as a final product, but as the outcome of a shared process in which scientific knowledge, operational needs and institutional contexts are integrated from the outset. In this sense, co-production is not an accessory element, but an integral part of the system’s structure.
In the case of the Africa Drought Watch, this means building a tool capable not only of describing drought, but also of anticipating its impacts in ways that are coherent with the contexts in which it will be used. It is within this transition – from data production to the shared construction of risk – that forecasting ceases to be mere observation and becomes the capacity to anticipate.